Page 1
Improving Pre-operative Cefazolin Use in Patients with Reported Beta-Lactam Allergy Undergoing Elective
Surgery
by
Dr. Philip Wai-Hei Lam
A thesis submitted in conformity with the requirements for the degree of Master of Science in Quality Improvement
Institute of Health Policy, Management and Evaluation University of Toronto
© Copyright by Dr. Philip Wai-Hei Lam 2020
Page 2
ii
Improving Pre-operative Cefazolin Use in Patients with Reported Beta-Lactam Allergy Undergoing Elective
Surgery
Philip Lam
Master of Science in Quality Improvement
Institute of Health Policy, Management and Evaluation
University of Toronto
2020
Abstract
Cefazolin is the preferred agent for antimicrobial prophylaxis in most
surgical procedures but is often avoided in patients who report an
allergy to a beta-lactam antibiotic. A retrospective cohort study was
conducted to assess the relationship between self-reported beta-
lactam allergy and surgical site infection (SSI). Using a multivariable
logistic regression model, a reported beta-lactam allergy was
associated with a significant increase in SSI risk (adjusted odds ratio
1.61, 95% CI: 1.04 - 2.51, p = 0.03) completely mediated by receipt of
an alternate antibiotic to cefazolin. An anesthesiologist-led intervention
consisting of a standardized allergy and antibiotic selection algorithm
was implemented in January 2019 which resulted in a sustained
increase in pre-operative cefazolin use in patients with reported beta-
lactam allergy (42.7% to 85.6%; p=0.002). Due to its simplicity and
minimal resource requirements, this intervention is scalable to other
institutions with potential to reduce SSI risk across a broad number of
surgeries.
Page 3
iii
Acknowledgments
First and foremost, I would like to thank my supervisor, Dr. Jerome Leis for his patience,
expertise and mentorship over the past two years.
I would like to acknowledge the study team involved in this project, including Dr. Jordan
Tarshis, Dr. Avery Nathens, Dr. Daniel Riegert, Dr. Melinda Li, Dr. Payam Tarighi, Dr. Keith
Gunaratne, Marion Elligsen BScPhm, Ariel Kwan BScPhm, Shaun Barry BScPhm, Phoebe Quek
BScPhm, Samantha Quach BScPhm, Andrea Clarke RN and Ellie Lee. The success of this
project would not have been possible without your input, feedback and encouragement.
I would also like to thank Dr. Theis Lange (University of Copenhagen) for his guidance in
conducting the mediation analysis for the study.
Finally, thank you to rest of my thesis committee: Dr. Patricia Trbovich, Dr. Kaveh Shojania, Dr.
Avery Nathens, Dr. Barbara Haas and Dr. Dominik Mertz.
Page 4
iv
Table of Contents
Acknowledgments ..................................................................................................................... iii
Table of Contents ...................................................................................................................... iv
List of Tables ............................................................................................................................ ix
List of Figures ............................................................................................................................ x
List of Appendices.................................................................................................................... xii
Part I: Self-Reported Beta-Lactam Allergy and the Risk of Surgical Site Infection ................. xiii
Chapter 1 Introduction ................................................................................................................ 1
Introduction ........................................................................................................................... 1
Chapter 2 Background and Literature Review ............................................................................. 2
Background ........................................................................................................................... 2
2.1 The Beta-Lactam Class of Antibiotics ............................................................................. 2
2.1.1 Beta-Lactam Allergy: True Allergy versus Reported Allergy ............................... 3
2.1.2 Established Risk Factors for Beta-Lactam Allergy ............................................... 5
2.2 Surgical Site Infection ..................................................................................................... 6
2.2.1 Established Risk Factors for Surgical Site Infection ............................................. 7
2.2.2 The Role of Beta-Lactam Antibiotics in the Prevention of Surgical Site
Infections ............................................................................................................ 8
2.3 Impact of Reported Beta-Lactam Allergy ...................................................................... 10
2.3.1 Treatment of Infection ....................................................................................... 10
2.3.2 Prevention of Surgical Site Infection ................................................................. 11
Chapter 3 Objectives ................................................................................................................ 15
Objectives ............................................................................................................................ 15
Chapter 4 Methodology ............................................................................................................ 16
Methodology ....................................................................................................................... 16
4.1 Overview of the Study Design ...................................................................................... 16
Page 5
v
4.2 Ethical Considerations .................................................................................................. 16
4.3 Population of Interest .................................................................................................... 16
4.4 Sources of Data ............................................................................................................. 17
4.5 Variables and Confounders ........................................................................................... 18
4.5.1 Exposure Variable: Reported Beta-Lactam Allergy ........................................... 18
4.5.2 Mediator Variable: Receipt of An Alternate Antibiotic to Cefazolin .................. 18
4.5.3 Outcome Variable: 30-day Surgical Site Infection ............................................. 18
4.5.4 Confounders ...................................................................................................... 19
4.6 Statistical Analysis ........................................................................................................ 19
4.6.1 Data Management ............................................................................................. 19
4.6.2 Descriptive Statistics ......................................................................................... 20
4.6.3 Multivariable Regression ................................................................................... 20
4.6.4 Mediation Analysis ........................................................................................... 21
4.6.5 Sample Size and Power Calculation ................................................................... 21
Chapter 5 Results ..................................................................................................................... 24
Results ................................................................................................................................. 24
5.1 Characteristics of the Cohort ......................................................................................... 24
5.2 Characteristics of the Cohort by Reported Beta-Lactam Allergy.................................... 24
5.2.1 Patient and Intra-operative Characteristics ......................................................... 24
5.2.2 Pre-Operative Antibiotic Use ............................................................................. 24
5.2.3 Reported Beta-Lactam Allergy Characteristics .................................................. 25
5.3 Primary Objective: The Impact of a Reported Beta-lactam Allergy on Surgical Site
Infection ....................................................................................................................... 25
5.4 Secondary Objective: Mediation Analysis ..................................................................... 26
Chapter 6 Discussion ................................................................................................................ 36
Discussion ........................................................................................................................... 36
Page 6
vi
6.1 Summary of Results ...................................................................................................... 36
6.2 Comparison with Previous Studies: Association between Reported Beta-Lactam
Allergy and Surgical Site Infection ............................................................................... 36
6.3 Study Strengths and Limitations.................................................................................... 37
6.4 Future Direction ............................................................................................................ 39
6.5 Conclusions .................................................................................................................. 39
Part II: Evaluation of a Standardized Pre-operative Allergy and Antibiotic Assessment
Algorithm on Cefazolin Use in Patients with Reported Beta-Lactam Allergy Undergoing
Elective Surgery ................................................................................................................... xl
Chapter 7 Introduction .............................................................................................................. 41
Introduction ......................................................................................................................... 41
7.1 Importance and Relevance of the Problem .................................................................... 41
7.2 Causes of the Problem................................................................................................... 42
Chapter 8 Literature Review ..................................................................................................... 44
Literature Review ................................................................................................................ 44
8.1 Beta-Lactam Allergy Skin Testing ................................................................................ 44
8.2 Direct Drug Provocation Testing ................................................................................... 47
8.3 Structured Allergy History ............................................................................................ 48
8.4 Implementation Challenges ........................................................................................... 49
Chapter 9 Objectives ................................................................................................................ 55
Objectives ............................................................................................................................ 55
9.1 Study Aim .................................................................................................................... 55
9.2 Study Objectives ........................................................................................................... 55
Chapter 10 Methodology .......................................................................................................... 56
Methodology ....................................................................................................................... 56
10.1 Overview of the Study Design ...................................................................................... 56
10.2 Ethical Considerations .................................................................................................. 56
Page 7
vii
10.2.1 Setting ............................................................................................................... 56
10.2.2 Study Team ....................................................................................................... 59
10.2.3 Study Participants .............................................................................................. 59
10.3 Quality Improvement Diagnostics ................................................................................. 60
10.3.1 Process Map ...................................................................................................... 60
10.3.2 Semi-Structured Interviews ............................................................................... 61
10.3.3 Anesthesiologist Survey .................................................................................... 62
10.3.4 Ishikawa Diagram ............................................................................................. 63
10.4 Change Concept and Interventions ................................................................................ 63
10.4.1 Allergy and Pre-operative Antibiotic Assessment Algorithm ............................. 64
10.4.2 Stamp Communication Tool .............................................................................. 65
10.4.3 Education .......................................................................................................... 65
10.5 Family of Measures ...................................................................................................... 66
10.5.1 Outcome Measures ............................................................................................ 66
10.5.2 Process Measures .............................................................................................. 67
10.5.3 Balancing Measures .......................................................................................... 68
10.6 Iterative Tests of Change .............................................................................................. 68
10.6.1 Usability Testing ............................................................................................... 68
10.6.2 Assessing Fidelity with Expansion to Nursing ................................................... 69
10.6.3 Assessing for Internalization of Practice ............................................................ 70
10.6.4 Expansion to Nurse Practitioners ....................................................................... 71
10.7 Additional Variables of Interest .................................................................................... 71
10.8 Statistical Analysis ........................................................................................................ 72
10.8.1 Data Management ............................................................................................. 72
10.8.2 Descriptive Statistics ......................................................................................... 72
10.8.3 Statistical Process Control Chart ........................................................................ 72
Page 8
viii
10.8.4 Stratification of Primary Outcome ..................................................................... 73
10.8.5 Segmented Regression Analysis ........................................................................ 73
10.8.6 Sample Size Calculation .................................................................................... 74
Chapter 11 Results.................................................................................................................... 83
Results ................................................................................................................................. 83
11.1 Characteristics of the Cohort ......................................................................................... 83
11.2 Primary Outcome Measure: Cefazolin Use .................................................................... 84
11.3 Secondary Outcome Measure: Surgical Site Infection ................................................... 85
11.4 Process Measure: Stamp Utilization .............................................................................. 86
11.5 Balancing Measure: Peri-operative Allergic Reaction ................................................... 86
Chapter 12 Discussion .............................................................................................................. 99
Discussion ........................................................................................................................... 99
12.1 Summary of Results ...................................................................................................... 99
12.2 Comparison with Previous Interventions ......................................................................100
12.3 Study Limitations.........................................................................................................102
12.4 Cost Considerations .....................................................................................................104
12.5 Sustainability and Spread .............................................................................................105
12.6 Future Directions .........................................................................................................106
12.7 Conclusions .................................................................................................................107
Glossary of Terms ...................................................................................................................109
References ...............................................................................................................................111
Appendices ..............................................................................................................................117
Page 9
ix
List of Tables
Table 2-1: Beta-Lactam family of antibiotics ............................................................................ 13
Table 2-2: Gell & Coombs classification of drug hypersensitivity reactions .............................. 14
Table 5-1: Baseline characteristics of cohort ............................................................................. 27
Table 5-2: Pre-operative antibiotic administration characteristics.............................................. 29
Table 5-3: Antibiotic dose and timing of cefazolin.................................................................... 30
Table 5-4: Antibiotic dose and timing of clindamycin ............................................................... 31
Table 5-5: Reported beta-lactam allergy antibiotic .................................................................... 32
Table 5-6: Allergic reactions to beta-lactam antibiotics ............................................................. 33
Table 5-7: Estimated effects of reported beta-lactam allergy and covariates on 30-day surgical
site infection in the multivariable logistic regression model ...................................................... 34
Table 8-1: Summary of interventions aimed at improving pre-operative beta-lactam antimicrobial
use ............................................................................................................................................ 53
Table 11-1: Baseline characteristics of patients with reported beta-lactam allergy during baseline
and intervention periods ........................................................................................................... 87
Table 11-2: Distribution of reported beta-lactam antibiotic allergy ........................................... 88
Table 11-3: Pre-operative antibiotic administration during baseline and intervention period ..... 89
Table 11-4 Surgical site infection outcomes during study period............................................... 95
Page 10
x
List of Figures
Figure 4-1: Directed acyclic graph illustrating the potential relationship between reported beta-
lactam allergy and surgical site infection .................................................................................. 23
Figure 5-1: Natural effects model estimates for receipt of an alternate antibiotic to cefazolin as
the mediator ............................................................................................................................. 35
Figure 10-1: Process map outlining allergy assessments in the pre-operative period ................. 76
Figure 10-2: Ishikawa diagram ................................................................................................. 77
Figure 10-3: Driver diagram ..................................................................................................... 78
Figure 10-4: Allergy assessment and pre-operative antibiotic selection algorithm ..................... 79
Figure 10-5: Stamp communication tool ................................................................................... 80
Figure 10-6: Allergy assessment and pre-operative antibiotic selection algorithm redesigned
following usability testing ........................................................................................................ 81
Figure 10-7 Example of a laminated algorithm attached to anesthesia drug cart in the operating
room ......................................................................................................................................... 82
Figure 11-1: Statistical process control chart (P-chart) for outcome measure: Proportion of
patients with reported beta-lactam allergy receiving cefazolin as pre-operative prophylaxis ...... 90
Figure 11-2: Segmented regression analysis on the primary outcome measure (monthly cefazolin
use) .......................................................................................................................................... 91
Figure 11-3 Pareto chart: Reasons for avoiding cefazolin when alternate antibiotic administered
during intervention period (n=22) ............................................................................................. 92
Figure 11-4: Pre-operative cefazolin use during intervention period stratified by algorithm use in
pre-operative clinic ................................................................................................................... 93
Figure 11-5: Cefazolin use in the baseline and intervention periods, stratified by description of
allergic reaction ........................................................................................................................ 94
Page 11
xi
Figure 11-6: Statistical process control chart (P-chart) for process measure: Proportion of
patients with reported beta-lactam allergy with stamp imprinted on anesthetic record (All
patients assessed in pre-operative clinic) ................................................................................... 96
Figure 11-7: Statistical process control chart (P-chart) for process measure: Proportion of
patients with reported beta-lactam allergy with stamp imprinted on anesthetic record (Patients
assessed by anesthesiologist of pharmacist only) ...................................................................... 97
Figure 11-8: Statistical process control chart (T-chart) for balancing measure: Time between
peri-operative allergic reactions during study period ................................................................. 98
Figure 12-1: Statistical process control chart (P-chart): Proportion of patients with reported beta-
lactam allergy undergoing elective Caesarean section receiving cefazolin as pre-operative
prophylaxis ............................................................................................................................. 108
Page 12
xii
List of Appendices
Appendix 1: Sampling methodology and surgical procedures selected for NSQIP .................. 117
Appendix 2: Definitions used for covariates included in multivariable logistic regression model
............................................................................................................................................... 119
Appendix 3: Literature review search strategy ........................................................................ 121
Appendix 4: Anesthesiologist survey and responses ............................................................... 122
Page 13
Part I: Self-Reported Beta-Lactam Allergy and the Risk of
Surgical Site Infection
Page 14
1
Chapter 1 Introduction
Introduction
Beta-lactams are a family of antibiotics which share a common beta-lactam ring
molecular structure, and include penicillins, cephalosporins and carbapenems. Beta-lactams are
the preferred class of antibiotic for the treatment and prevention of a variety of infections. In the
peri-operative setting, cefazolin (a first-generation cephalosporin antibiotic) is the preferred
agent for antimicrobial prophylaxis in most surgical procedures(1). Due to misperceptions
regarding allergy cross-reactivity however, prescribers often avoid cefazolin in patients who
report an allergy to any beta-lactam antibiotic(2).
The impact of avoiding the preferred beta-lactam antibiotic for the treatment of bacterial
infection has been well documented and includes an increased risk of adverse events(3),
antibiotic-resistant organisms(4), and mortality(5). To date, only one study has identified an
association between penicillin allergy and an increased risk of surgical site infection; the effect
was demonstrated to be mediated through receipt of a non-beta-lactam antibiotic(6). However,
this cohort was limited to only five surgical procedure types, and included only patients with
reported penicillin allergy(6). It is not known whether the association exists at other institutions
where alternate antibiotic use significantly differs.
We hypothesized that patients with a reported beta-lactam allergy have an increased risk
of surgical site infections across a broad range of surgeries, and that this risk is mediated by
avoidance of cefazolin. The following retrospective cohort study was conducted to assess this
association and the potential mediators of this effect.
Page 15
2
Chapter 2 Background and Literature Review
Background
The importance of beta-lactam antibiotics in the treatment of infection, and the potential
harm in avoiding their use in the context of a reported allergy have been well documented(3–
5,7). However, the impact of a reported beta-lactam allergy on the prevention of surgical site
infection is less clear. This literature review will begin with the epidemiology and risk factors for
beta-lactam allergy and surgical site infection, followed by a summary of the evidence pertaining
to the impact of a reported beta-lactam allergy on the treatment and prevention of infection.
2.1 The Beta-Lactam Class of Antibiotics
Beta-lactams are a family of antibiotics which are characterized by a beta-lactam ring
molecular structure. Beta-lactam antibiotics exert their bactericidal activity by binding to
transpeptidase enzymes which are essential in the cross-linking of peptidoglycans in the bacterial
cell wall(8). Inhibition of this enzyme results in loss of bacterial cell wall integrity and
eventually cell lysis(8). Penicillin was the first beta-lactam antibiotic discovered and was
originally isolated from the Penicillium mould(8). Through the chemical modification of the
penicillin side chain and adjacent ring structures, other beta-lactam derivatives were developed
with improved antimicrobial spectrum of coverage. Examples of beta-lactam subclasses include:
aminopenicillins, ureidopenicillins, cephalosporins, monobactams and carbapenems (Table 2-1).
Due to their established efficacy and safety profile, beta-lactams are the preferred class of
antibiotics for the treatment of a variety of infections, including those caused by streptococci,
Enterobacteriaceae, Listeria monocytogenes, Actinomyces species, Staphylococcus aureus,
Neisseria species, and Treponema pallidum(9).
Page 16
3
2.1.1 Beta-Lactam Allergy: True Allergy versus Reported Allergy
A beta-lactam allergy refers to an immune-mediated drug hypersensitivity reaction which
occurs upon exposure to an antibiotic containing a beta-lactam ring. Drug hypersensitivity
reactions are traditionally categorized into four types using the Gell & Coombs classification
system (Table 2-2), and reflect the different immune-mediated mechanisms. Type I reactions
occur as a result of Immunoglobulin E (Ig-E) mediated mast cell degranulation and can manifest
as urticaria, angioedema, bronchospasm or anaphylaxis minutes to hours after drug exposure.
Type II reactions are mediated by antibody-antigen binding on a host cell resulting in blood cell
depletion. Type III reactions result in antibody-antigen immune complex deposition in organ
systems which can result in serum sickness, vasculitis and glomerulonephritis. Type IV reactions
occur as result of T-cell activation and manifest as delayed reactions ranging from a benign rash
to life-threatening organ damage. The baseline risk for any hypersensitivity reaction to beta-
lactam antibiotics is approximately 2.0% with cutaneous manifestations (representing Type I and
IV reactions) being the most commonly reported(10).
Despite the low incidence of allergy to beta-lactam antibiotics, the prevalence of patient-
reported “allergy” is high. Penicillin has been the most well-studied beta-lactam antibiotic in the
context of reported allergy. Studies have consistently demonstrated that approximately 10% of
the general population report a penicillin allergy. However, only 1% of the general population
have a true allergy when testing is performed(9).
The discrepancy between the incidence of reported versus true beta-lactam allergy have
been attributed to several factors. Firstly, misclassification of drug intolerance as an allergy by
both physicians and patients contribute to over-labelling of allergy. In a retrospective cohort
study of 232,616 adult patients seen by 199 primary care providers over a 5-year period, 15.6%
Page 17
4
of patients had a reported beta-lactam allergy documented by a healthcare professional.
Moreover, there was a 3-fold variability in reported allergy prevalence among providers (7.8% to
24.8%), suggesting that some patients were being misclassified as having a true allergy when
their reaction may have reflected drug intolerance(11). Secondly, concurrent viral infections can
produce rashes that mimic drug hypersensitivity reactions. This has been best described in
children with infectious mononucleosis secondary to Epstein-Barr virus, where up to 90% of
patients with this viral infection who receive ampicillin develop a morbilliform rash(12). A rash
caused by such a viral infection could therefore be mistaken for an allergy to a beta-lactam if one
had been prescribed. Thirdly, even with a history consistent with a true allergic reaction, patients
can lose their drug hypersensitivity over time. This was demonstrated by Blanca et al. who
performed annual skin testing on 31 penicillin skin-test positive patients over a 5 year
period(13). At 5 years of follow-up, nearly 50% of penicillin allergic patients became skin test
negative, suggesting that beta-lactam allergies do not necessarily persist lifelong(13).
Clarifying a reported beta-lactam allergy and determining whether it represents a true
drug allergy can be achieved by various methods such as a detailed allergy history, skin testing
and direct drug provocation. Unfortunately, healthcare provider knowledge surrounding how to
clarify reported beta-lactam allergies is poor. Wanat et al. performed a systematic review of
prescriber views on antibiotic use in the context of penicillin allergy(14). Of the four studies
evaluating prescriber knowledge, 39-57% of prescribers believed a penicillin allergy was
permanent(14). In a clinical audit of 30 hospital physicians practicing in the United Kingdom,
13% of prescribers were unaware of the characteristics of an allergic reaction, with 23%
considering an antibiotic side effect as an allergic reaction(14).
Page 18
5
These findings suggest that antibiotic selection is heavily influenced by the presence of a
reported beta-lactam allergy regardless of whether it represents a true drug hypersensitivity
reaction. For this reason, a reported beta-lactam allergy was chosen as the exposure variable of
interest in this study. Identifying an association between reported allergy and surgical site
infection would represent a significant opportunity to improve patient outcomes with
interventions aimed at improving prescriber awareness and comfort around clarification of
reported beta-lactam allergies.
2.1.2 Established Risk Factors for Beta-Lactam Allergy
Increasing age and female gender are associated with an increased likelihood of a
reported beta-lactam allergy(15). In a retrospective review of electronic health records from
411,543 patients in San Diego County (California, USA), the prevalence of reported allergy to
beta-lactam antibiotics in females was significantly higher compared to males (11.0% vs 6.5%)
regardless of the decade of life(15). Furthermore, the authors found that increasing age was
significantly correlated with beta-lactam allergy prevalence, with 20% of women over the age of
90 years reporting an allergy to penicillin(15). Combined with the finding that women had higher
beta-lactam antibiotic use compared to men, the study authors hypothesize that gender and age as
risk factors for reported beta-lactam allergy may be related to antimicrobial use, with higher
prevalence of reported allergy in cohorts with higher antibiotic use(15).
Similarly, frequent repeated antibiotic exposures as a result of chronic medical illness
appear to increase the risk of beta-lactam allergy. This has been best described in the cystic
fibrosis population, who require frequent administration of beta-lactam antibiotics for treatment
of exacerbations of disease(16,17). Burrows et al. performed a retrospective review of 150
patients with cystic fibrosis and found that 36% had allergic reactions to one or more beta-lactam
Page 19
6
antibiotics(16). A similar study of 90 patients with cystic fibrosis and found that 26 patients
(29%) developed one or more reactions to beta-lactam antibiotics and appeared greater in
penicillin-type antibiotics compared to cephalosporins(17). More recently, Imlay et al. studied
2153 solid organ and hematopoietic cell transplant patients and found that 16% of patients
reported a beta-lactam allergy(18). Within this cohort, 94% of patients received an antibiotic
within the first 100 days post-transplant, which supports the notion that frequent antibiotic
exposure is an underlying risk factor for developing an allergy(18).
Patients who report an allergy to one compound are also more likely to develop a reaction
to a second structurally unrelated compound. For example, Strom et al. performed a retrospective
cohort study of 969 patients with an allergic reaction to a sulfonamide (non-beta-lactam)
antibiotic. Compared to patients with no allergy to sulfonamide antibiotics, patients with a
sulfonamide allergy were more likely to develop a subsequent allergic reaction to penicillin
(adjusted odds ratio 3.9)(19). The mechanism behind this finding is unclear but is hypothesized
to be related to a genetic predisposition to drug hypersensitivity reactions(19).
2.2 Surgical Site Infection
A surgical site infection (SSI) is defined as an infection related to a surgical procedure
that occurs at or near the incision site. SSIs complicate 5-20% of surgical procedures annually,
and are associated with significant morbidity, including an increased length of hospitalization,
intensive care unit utilization, and a five times increased risk of hospital re-admission(20,21).
Compared to other hospital-acquired infections, SSIs are the costliest, and in the United States is
estimated to have an annual expenditure of $3.3 billion USD(22).
SSIs can be categorized into superficial incisional, deep incisional, and organ space
infection which reflect the extent of tissue involvement. The Centers for Disease Control (CDC)
Page 20
7
National Healthcare Safety Network (NHSN) publishes annual definitions for SSIs for these
three categories to assist healthcare institutions in performing surveillance(23).
SSIs are frequently caused by pathogens which colonize the skin at the site of the surgical
incision. Staphylococcus aureus and coagulase-negative staphylococci are Gram-positive
organisms which colonize the skin and are frequent pathogens isolated in clean surgical
procedures(24). In surgical procedures involving the gastrointestinal or genitourinary tract,
Gram-negative bacilli (such as Escherichia coli), Enterococcus species and anaerobes (such as
Bacteroides fragilis) can also contribute to infection(24).
2.2.1 Established Risk Factors for Surgical Site Infection
Risk factors which have been associated with an increased the risk of SSI include
increasing age(25,26), cigarette smoking(27), diabetes(28), radiation(29) and
immunosuppressive therapy(30). It is generally accepted that for all these risk factors, the
increased SSI risk is mediated through the impairment of wound healing. Increasing age results
in a decrease in cutaneous nerve and blood vessel supply along with atrophy of the dermal
tissues which contribute to impaired wound healing(25). In a retrospective cohort study of
144,485 patients undergoing surgical procedures, the risk of SSI increased by 1.1% per year
between ages 17 and 65(26). The mixture of compounds found in tobacco smoke are thought to
increase the risk of SSI through a relative increase in tissue hypoxemia, and a blunted
inflammatory and bactericidal response(27). Pre-operative smoking cessation has been
demonstrated to significantly reduce the incidence of surgical site infections(27).
Immunosuppressive therapy and radiation also contribute to impaired wound healing and have
been identified as important risk factors for SSI(29,30).
Page 21
8
Scoring systems used to assess the general fitness of a patient prior to surgery have been
useful in predicting intra-operative and post-operative complications including infection. The
most widely used risk index is the American Society of Anesthesiologists (ASA) Physical Status
Classification System which grades a patient’s level of health on a five-category scale. In a
retrospective review of 84,691 surgical procedures by the CDC National Nosocomial Infections
Surveillance (NNIS), the surgical wound infection rate in patients with an ASA score of 1
(healthy, non-smoker, minimal alcohol consumption) was 1.5 per 100 cases. In patients with
ASA score of 3 (severe systemic disease), the rate of SSI increased to 3.7(31). The utility of this
system in predicting SSIs likely reflects the accumulation of co-morbidities which impair the
wound healing process(31).
There is a significant variability in SSI risk depending on the type of surgical procedure.
Using data from the NNIS as an example, bowel surgeries, due to their proximity to
gastrointestinal bacterial flora, are associated with higher rates of SSI (rate: 3.98 – 11.25 per 100
cases) compared to knee prothesis surgery (rate: 0.88 – 2.26 per 100 cases)(32). Similarly, the
degree of wound contamination at the time of surgery has been associated with SSI. Wound
contamination has been classified by the CDC NHSN into four categories of ascending infection
risk: clean, clean-contaminated, contaminated and dirty / infected. Data published by the NNIS
from 1987 to 1990 reported surgical wound infection rates for clean, clean-contaminated,
contaminated and dirty-infected wounds of 2.1, 3.3, 6.4 and 7.1 per 100 cases respectively (31).
2.2.2 The Role of Beta-Lactam Antibiotics in the Prevention of Surgical Site Infections
The use of antimicrobial prophylaxis for the prevention of SSI is focused on decreasing
the burden of bacterial organisms at the surgical site during the procedure. An effective
prophylactic antibiotic agent must possess an appropriate antimicrobial spectrum of activity to
Page 22
9
kill common bacterial pathogens implicated in SSIs and must be administered at an appropriate
time prior to incision. Inappropriate timing of pre-operative antibiotic prophylaxis has been
associated with an increased risk of SSI(33).
Cefazolin is an intravenously administered first-generation cephalosporin antibiotic
which possesses an appropriate antimicrobial spectrum of coverage against pathogens implicated
in the development of SSI. For patients weighing less than 120 kilograms, a 2-gram dose is
administered intravenously 0 to 60 minutes before surgical incision, with a subsequent dose
given every 4 hours for as long as the duration of surgery(1). The efficacy of cefazolin in the
prevention of SSI is well established among several surgical specialties(1). Clinical practice
guidelines developed jointly by the American Society of Health-System Pharmacists, Infectious
Diseases Society of America, the Surgical Infection Society, and the Society for Healthcare
Epidemiology of America in 2013 recommend cefazolin as surgical prophylaxis in all surgical
procedure types (except for liver transplantation) when antibiotic prophylaxis is indicated(1).
The addition of metronidazole has been recommended in certain small bowel and colorectal
surgeries to provide additional antimicrobial activity against anaerobic organisms(1). Due to a
different bacterial flora and pathogenesis of infection in procedures involving the lower urinary
tract, fluoroquinolones and trimethoprim/sulfamethoxazole are also appropriate first-line agents
for specific urologic procedures(1).
In the setting of an allergy to beta-lactam antibiotic, these guidelines recommend
clindamycin and vancomycin as alternate agents for surgical prophylaxis(1). However, use of
clindamycin and vancomycin have been associated with higher rates of SSIs in orthopedic,
gynecologic, otolaryngology and neurosurgical procedures compared to cefazolin(34). Potential
explanations for the reduced efficacy of these alternate antibiotics include inferior bactericidal
Page 23
10
activity, and reduced antimicrobial activity against Gram-negative organisms(34). Furthermore,
vancomycin and fluoroquinolones should be administered starting 120 minutes before incision
due to the need for longer infusion times(1). This poses a challenge in the timing of pre-operative
antibiotic administration and may result in delays to surgical incision(35).
Although these guidelines acknowledge that cross-allergic reactions between penicillins
and cephalosporins are uncommon, the authors (based on expert opinion) suggest avoidance of
cephalosporins in patients with an allergy consistent with a Type 1 (IgE-mediated)
hypersensitivity reaction(1). Since this guideline was published, several studies have emerged
which demonstrate the lack of cross-reactivity between penicillins and cefazolin(2).
2.3 Impact of Reported Beta-Lactam Allergy
2.3.1 Treatment of Infection
The harms of avoiding beta-lactam antibiotic therapy in the treatment of infections have
been well documented, which include increased adverse events(3), drug-resistant organisms(4)
and mortality(5). For example, in a study of 318 patients with reported penicillin allergy and
Staphylococcus aureus bacteremia, patients who received vancomycin (a non-beta-lactam
antibiotic) had a 3-fold higher mortality rate compared to those who received beta-lactam
therapy(7). Interestingly, penicillin allergy alone was not a predictor of mortality, which
suggested that the causal pathway for increased mortality was exclusively due to receipt of an
alternate non-beta-lactam antibiotic(7). More locally, MacFadden et al. performed a multicenter
prospective cohort study of 507 patients with reported beta-lactam allergy across 3 Toronto
academic hospitals(3). Patients who did not receive preferred beta-lactam therapy were at higher
risk of adverse events (adjusted odds ratio 3.1) compared to those without reported allergy(3). To
evaluate the impact of a reported penicillin allergy on the prevalence of drug-resistant bacterial
Page 24
11
infections, Macy et al. conducted a retrospective matched cohort study of patients admitted to
Kaiser Foundation hospitals in southern California over a 3-year period and found that patients
with a reported penicillin allergy were more likely to have longer hospitalizations (0.59 more
days), and a higher risk of C. difficile (23%), VRE (20%) and MRSA (14%) infections(4).
The use of alternate antibiotics in patients with reported beta-lactam allergy has also been
associated with increased antimicrobial utilization and associated cost(36). In a retrospective
review of antibiotic costs at a tertiary care center in Quebec, alternate antibiotic use in patients
with penicillin allergy resulted in an additional annual cost of $15,672 CAD(36).
2.3.2 Prevention of Surgical Site Infection
Only one study to date has examined the impact of a reported penicillin allergy on
surgical site infection(6). Blumenthal et al. performed a 5-year retrospective review of 9004
procedures over a 5-year period (2010 – 2014) at Massachusetts General Hospital(6). The
authors focused specifically on patients with reported penicillin allergy (rather than any reported
beta-lactam allergy), and five specific surgery types (hysterectomy, hip arthroplasty, knee
arthroplasty, colon surgery and coronary artery bypass grafting). Of the 8385 patients included in
the cohort, 11% reported a penicillin allergy. Patients with a reported penicillin allergy were less
likely to receive cefazolin as pre-operative prophylaxis (12 vs 92%, p < 0.001) and more likely to
receive clindamycin (49% vs 3%, p < 0.001), vancomycin (35% vs 3%, p < 0.001) and
gentamicin (24% vs 3%, p< 0.001) compared to those without allergy(6). In a multivariable
logistic regression model accounting for surgery type, age, sex, race, ASA classification,
procedure duration and wound classification, a reported penicillin allergy was associated with an
increased risk of surgical site infection with an adjusted odds ratio of 1.51 (95% CI 1.02 – 2.22, p
= 0.04)(6). Further mediation analysis was performed which confirmed that this effect was
Page 25
12
completely mediated by the receipt of a non-beta-lactam antibiotic (indirect effect adjusted OR
2.05, 95% CI 1.40 – 3.00)(6). The authors estimated that between 112 and 124 patients with
reported penicillin allergy would need to be assessed in order to prevent one SSI(6).
The single-center nature of the study combined with the inclusion of patients with
penicillin allergy only undergoing five specific procedure types limits the external validity of the
study. Furthermore, this study reported a relatively high use of vancomycin as an alternate
antibiotic in those with penicillin allergy (34.7%), of which 97.5% of doses were not
administered in the correct timeframe(6). It is therefore unclear whether the impact of alternate
antibiotic use is solely due to inferior antibiotic efficacy or related to inappropriate antibiotic
administration time as well.
Considering these limitations, we conducted a retrospective cohort study to assess the
impact of a reported beta-lactam allergy on the risk of surgical site infection over a broad range
of surgeries, and to determine whether this effect was mediated by a receipt of an alternate
antibiotic to cefazolin.
Page 26
13
Table 2-1: Beta-Lactam family of antibiotics
Class Examples
Penicillin Penicillin V, Penicillin G
Aminopenicillin Ampicillin, Amoxicillin
Ureidopenicillin Piperacillin
Carboxypenicillin Ticarcillin
Cephalosporin 1st generation: cephalexin, cefadroxil, cefazolin
2nd generation: cefuroxime, cefaclor, cefprozil, cefoxitin, cefotetan
3rd generation: ceftriaxone, cefotaxime, ceftazidime
Monobactam Aztreonam
Carbapenem Ertapenem, meropenem, imipenem
Page 27
14
Table 2-2: Gell & Coombs classification of drug hypersensitivity reactions
Classification Immune Mechanism Typical Onset Examples
I Ig-E – antigen complex
activates mast cells
Minutes to hours Urticaria,
angioedema,
anaphylaxis
II Ig-G binds to antigen-
target cell complex and
activates complement
pathway
Hours to days Hemolytic anemia,
neutropenia,
thrombocytopenia
III IgG – antigen immune
complex deposition in
organ systems
Weeks Serum sickness,
vasculitis,
glomerulonephritis
IV T-lymphocyte and
macrophage activation
Days to weeks Benign
maculopapular rash,
Steven-Johnson
syndrome, toxic
epidermal necrolysis
Page 28
15
Chapter 3 Objectives
Objectives
The primary objective of this study was to assess the impact of a reported beta-lactam
allergy on the risk of surgical site infection.
The secondary objective of this study was to determine whether the impact of a reported
beta-lactam allergy on the risk of surgical site infection was mediated by receipt of an alternate
antibiotic to cefazolin
Page 29
16
Chapter 4 Methodology
Methodology
4.1 Overview of the Study Design
A retrospective cohort study of patients undergoing a surgical procedure between January
1, 2017 and June 30, 2018 was conducted using the American College of Surgeons (ACS)
National Surgical Quality Improvement Program (NSQIP) database at Sunnybrook Health
Sciences Centre (SHSC). SHSC is a 627-bed academic health sciences centre located in Toronto,
Ontario, Canada. NSQIP is a voluntary benchmarking program aimed at improving the quality
and safety of surgical care. An estimated 15,000 surgical procedures are performed at SHSC
annually, with approximately 2,500 of them sampled by NSQIP.
4.2 Ethical Considerations
The research protocol for this study was submitted to the Research Ethics Board at
Sunnybrook Health Sciences Centre for delegated review and received approval on January 15,
2019 (Project Identification Number: 439-2018).
4.3 Population of Interest
Patients who underwent a surgical procedure at SHSC during the study period and
sampled by the institutional NSQIP were included in the study. During the study period, data
pertaining to the following surgical specialties were collected for NSQIP: general surgery,
vascular surgery, gynecology-oncology, neurosurgery, urology, orthopedic surgery and plastic
surgery. A complete list of the specific procedures and sampling technique is included in
Appendix 1. Cases which were excluded from NSQIP sampling included: patients less than 18
years of age, brain-death organ donors, cases involving hyperthermic intraperitoneal
Page 30
17
chemotherapy, trauma cases, transplant cases, any surgical procedure related to an occurrence or
complication of a prior procedure, and multiple surgical cases from the same patient within 30
days(37). Due to a markedly different pathogenesis of post-operative infections in transurethral
procedures, and the fact that urinary tract infections are coded as a separate outcome measure
from surgical site infections, these procedures were excluded from the study.
4.4 Sources of Data
The institutional NSQIP database served as the main source of data for the study. As part
of NSQIP, 135 variables relating to pre-operative, intra-operative, and post-operative care are
collected for each sampled surgical procedure via retrospective chart review.
To obtain data pertaining to beta-lactam allergy, patients identified from the NSQIP
database were cross-referenced with the Stewardship Program Integrating Resource Information
Technology (SPIRIT) database, which aggregates microbiology, laboratory, and pharmacy
information for patients admitted to hospital(38). A search query for the following strings were
used to identify patients with beta-lactam allergy: “-cillin-”, “pcn”, “pen g”, “pen v”, “cef-”,
“ceph-”, “-penem”.
Finally, a manual review of the anesthetic record for each surgical procedure was
performed to extract the type, dose and time of pre-operative antibiotic administration. This
review also served to identify additional cases of reported beta-lactam allergy which were not
documented electronically or detected by the search query.
Page 31
18
4.5 Variables and Confounders
4.5.1 Exposure Variable: Reported Beta-Lactam Allergy
The exposure of interest was the presence of a self-reported beta-lactam allergy (binary
variable), defined by any allergy to an antibiotic containing a beta-lactam ring molecular
structure. Since the current literature describing the association between penicillin and surgical
site infection is thought to be mediated by receipt of an alternate antibiotic (rather than a direct
immune mechanism)(6), any type of patient-perceived allergic reaction was included regardless
of whether it constituted a true drug hypersensitivity reaction.
4.5.2 Mediator Variable: Receipt of An Alternate Antibiotic to Cefazolin
The mediator variable of interest was an alternate antibiotic to cefazolin. An alternate
antibiotic to cefazolin included other beta-lactam and non-beta-lactam antibiotics. Cefazolin was
specifically chosen because it is the recommended agent for almost all surgical procedures
requiring prophylaxis(1). Demonstration of a mediated effect through receipt of an alternate to
cefazolin would provide greater impetus to develop interventions to improve cefazolin use pre-
operatively, which theoretically would be easier to accomplish based on emerging literature
suggesting that cefazolin is a non-cross reactive beta-lactam antibiotic(2).
4.5.3 Outcome Variable: 30-day Surgical Site Infection
The primary outcome was the presence of surgical site infection at 30 days post
procedure. Surgical site infection was defined by ACS NSQIP(37), which are based on the
National Healthcare Safety Network (NHSN) definitions for SSI(23). This variable was collected
as part of NSQIP surveillance based on retrospective chart review. When post-operative follow-
up notes were unavailable, the patient was contacted to inquire about occurrence of SSI.
Page 32
19
4.5.4 Confounders
Based on the known risk factors for beta-lactam allergy and surgical site infections, a
directed acyclic graph was developed to identify variables on the causal pathway between the
exposure (beta-lactam allergy) and outcome (surgical site infection) variables (Figure 4-1). Using
this diagram, the following variables were selected a priori for use in the multivariable logistic
regression model in order to adjust for potential confounders while minimizing the risk of
collinearity: age, gender, smoking status, diabetes, steroid/immunosuppression and American
Society of Anesthesiologists (ASA) classification, and wound classification. Definitions for
diabetes, steroid/immunosuppression and wound classification are based on criteria set out by
NSQIP Operations Manual(37). The definitions used for each covariate are summarized in
Appendix 2.
4.6 Statistical Analysis
4.6.1 Data Management
Data obtained from the NSQIP database were inspected for completeness and reformatted
for use in the analysis. All patient data was de-identified to maintain confidentiality (replacing
medical record numbers with a study identification number). The master list linking patient
medical record numbers with study identification numbers was stored separately on a secure
hospital server and only accessible to the researchers. Only encrypted devices or internal servers
were used to store electronic information. All statistical analyses were performed using R
Statistical Software 3.4.3 (Foundation for Statistical Computing, Vienna, Austria) with the
following packages: tableone, car, geeglm and medflex.
Page 33
20
4.6.2 Descriptive Statistics
A descriptive analysis of patient demographics (age, gender), co-morbidities (diabetes
mellitus, current smoker, history of severe chronic obstructive pulmonary disease, ascites,
congestive heart failure, hypertension, need for dialysis, disseminated cancer, and
corticosteroid/immunosuppressant use), ASA physical status classification and operative factors
(elective surgery, surgical specialty, wound classification, procedure duration, length of stay)
were carried out for the entire cohort followed by stratification based on reported beta-lactam
allergy. To assess whether the covariates were balanced between the stratified groups, Chi-
square and Wilcoxon signed-rank tests were performed for categorical and continuous variables
respectively. Chi-square testing was also performed to compare the type of pre-operative
antibiotic received in patients with and without reported beta-lactam allergy
4.6.3 Multivariable Regression
The association between a reported beta-lactam allergy and surgical site infection while
accounting for confounding variables was evaluated using a multivariable logistic regression:
ln (𝑝
1 − 𝑝) = 𝛼 + 𝛽1𝐴𝑙𝑙𝑒𝑟𝑔𝑦 + 𝛽2𝐴𝑔𝑒 + 𝛽3𝐺𝑒𝑛𝑑𝑒𝑟
+ 𝛽4𝑆𝑚𝑜𝑘𝑖𝑛𝑔 + 𝛽5𝐷𝑖𝑎𝑏𝑒𝑡𝑒𝑠 + 𝛽6𝐼𝑚𝑚𝑢𝑛𝑜𝑠𝑢𝑝𝑝𝑟𝑒𝑠𝑠𝑖𝑜𝑛 + 𝛽7𝐴𝑆𝐴
+ 𝛽8 𝑊𝑜𝑢𝑛𝑑𝐶𝑙𝑎𝑠𝑠𝑖𝑓𝑖𝑐𝑎𝑡𝑖𝑜𝑛
Where (𝑝
1−𝑝) = 𝑜𝑑𝑑𝑠 𝑜𝑓 𝑑𝑒𝑣𝑒𝑙𝑜𝑝𝑖𝑛𝑔 𝑎 𝑠𝑢𝑟𝑔𝑖𝑐𝑎𝑙 𝑠𝑖𝑡𝑒 𝑖𝑛𝑓𝑒𝑐𝑡𝑖𝑜𝑛
𝛼 = 𝑏𝑎𝑠𝑒𝑙𝑖𝑛𝑒 𝑜𝑑𝑑𝑠 𝑤ℎ𝑒𝑛 𝛽1, 𝛽2 … 𝛽8 𝑖𝑠 𝑒𝑞𝑢𝑎𝑙 𝑡𝑜 𝑧𝑒𝑟𝑜
𝛽1, 𝛽2 … 𝛽8 = 𝑡ℎ𝑒 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑎𝑠𝑠𝑜𝑐𝑖𝑎𝑡𝑒𝑑 𝑤𝑖𝑡ℎ 𝑒𝑎𝑐ℎ 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑚𝑜𝑑𝑒𝑙
Page 34
21
The adjusted odds ratio for developing a surgical site infection in patients with beta-
lactam allergy (compared to those without allergy) was expressed by the natural exponential of
coefficient β1. The Hosmer-Lemeshow test was used to confirm that the model adequately fit the
data. The variance inflation factor was calculated for each covariate in the model to ensure there
was no evidence of multi-collinearity. To account for model overspecification, the “ASA”
covariate was re-categorized as a dichotomous variable (<3 , ≥ 3). Since a single patient could
have been sampled more than once during the study period, a generalized estimating equation
approach was used to carry out the regression.
4.6.4 Mediation Analysis
In order to understand the causal mechanisms underlying a reported beta-lactam allergy
and SSI, mediation analysis was carried out to determine whether the effect was mediated by
receipt of an alternate antibiotic to cefazolin. The analysis was performed using a natural effects
model, which is an extension of marginal structural models for mean nested counterfactuals(39).
This method allows for direct parameterization of each mediated pathway, while producing
estimates that correspond to the natural scale of the outcome model (an adjusted odds ratio)(39).
A multivariable logistic regression model using the same covariates selected a priori was used as
the outcome model for the mediation analysis. To minimize the risk of mediator-outcome
confounding and intertwining causal pathways, surgical procedures where no pre-operative
antibiotic was administered were excluded from mediation analysis.
4.6.5 Sample Size and Power Calculation
Using a relative precision of 50%, confidence level of 95%, an estimated SSI prevalence
of 3% in non-allergic patients (based on local data), an expected odds ratio of 1.5(6), and a ratio
of allergic to non-allergic individuals of 0.08, it was estimated that 2634 patients would be
Page 35
22
required to demonstrate the association. A study period of 18 months would yield approximately
3600 surgical procedures which would satisfy the power requirements for this study.
Page 36
23
Figure 4-1: Directed acyclic graph illustrating the potential relationship between reported
beta-lactam allergy and surgical site infection
Page 37
24
Chapter 5 Results
Results
5.1 Characteristics of the Cohort
Over the 18-month study period, 3708 surgical procedures were sampled by NSQIP.
After excluding transurethral procedures, there were 3589 procedures performed among 3499
patients. A total of 181 surgical site infections (5.0%) occurred in the cohort. Of the 3589
procedures, 369 (10.3%) occurred among patients who reported a beta-lactam allergy.
5.2 Characteristics of the Cohort by Reported Beta-Lactam Allergy
5.2.1 Patient and Intra-operative Characteristics
Patient demographics, co-morbidities and surgical factors are summarized in Table 1.
Patients with reported beta-lactam allergy were older (68.4 vs 65.3 years, p=0.005), and more
likely to be female (67.8% vs 54.5%, p<0.001) compared to those without allergy. Both groups
appeared balanced in terms of co-morbidities, ASA physical status classification, surgical
specialty, wound classification and duration of surgery.
5.2.2 Pre-Operative Antibiotic Use
Pre-operative antibiotic use stratified by reported beta-lactam allergy is summarized in
Table 5-2. The most commonly prescribed antibiotics were cefazolin (89.6%), metronidazole
(25.5%), and clindamycin (5.5%). There were 66 cases (1.8%) that were balanced across groups
where no pre-operative antibiotic was given.
Patients with reported beta-lactam allergy were significantly less likely to receive
cefazolin (38.8% vs 95.5%, p< 0.001) and more likely to receive clindamycin (52.0% vs 0.2%,
Page 38
25
p<0.001) or vancomycin (2.2% vs 0.1%, p<0.001) compared to those without allergy. Of the 226
patients with reported beta-lactam allergy who did not receive cefazolin, the most common
alternate antibiotics administered were clindamycin (85.0%), metronidazole (33.2%) and
vancomycin (3.5%).
When cefazolin was used, selection of a 2-gram dose (97.2% vs 98.0%), and
administration within 60 minutes before surgical incision (95.1% vs 94.2%) was similar between
patients with and without a reported beta-lactam allergy (Table 5-3). When clindamycin was
used, selection of a 600-milligram dose (92.7% vs 100%), and administration within 60 minutes
before surgical incision (90.1% vs 80.0%) was similar between patients with and without a
reported beta-lactam allergy (Table 5-4).
5.2.3 Reported Beta-Lactam Allergy Characteristics
The three most commonly reported antibiotic allergies were penicillin (85.4%),
amoxicillin (4.0%) and cephalexin (3.7%). Seven patients (1.9%) reported an allergy to two
different beta-lactam antibiotics (Table 5-5). The five most commonly reported reactions were
rash (36.8%), unknown to patient (24.9%), urticaria (13.2%), swelling or angioedema (10.3%)
and anaphylaxis (9.8%) (Table 5-6).
5.3 Primary Objective: The Impact of a Reported Beta-lactam Allergy on Surgical Site Infection
In the univariable analysis, 30-day surgical site infection occurred in 4.8% of patients
without reported beta-lactam allergy and 7.3% of patients with reported allergy. In the
multivariable logistic regression model, smoking status, an ASA physical status classification ≥
3, and a wound classified as contaminated or dirty were significantly associated with an
increased risk of SSI (Table 5-7). After accounting for these variables and other covariates, a
Page 39
26
reported beta-lactam allergy was associated with a statistically significant increase in 30-day
surgical site infection risk (adjusted odds ratio [aOR]= 1.61, 95% CI: 1.04 - 2.51, p = 0.03).
5.4 Secondary Objective: Mediation Analysis
Following exclusion of 66 procedures where no pre-operative antibiotics were
administered, the effect of a reported beta-lactam allergy on surgical site infection was
completely mediated by the receipt of an alternate antibiotic to cefazolin (indirect effect aOR =
1.68, 95% CI: 1.17 – 2.34, p=0.005) (Figure 5-1). When receipt of an alternate antibiotic to
cefazolin was accounted for, the direct effect of a reported beta-lactam allergy on the risk of
surgical site infection was not significant (direct effect aOR = 0.99, 95% 0.55 – 1.78, p=0.96).
Page 40
27
Table 5-1: Baseline characteristics of cohort
All
(n=3589)
No reported beta-
lactam allergy
(n = 3220)
Reported beta-
lactam allergy
(n = 369)
Patient Factors
Age (median [IQR]) 65.5 [54.4, 74.5] 65.3 [54.3, 74.1] 68.4 [57.2, 76.6]*
Gender (% male) 1582 (44.1) 1463 (45.4) 119 (32.2)*
Co-morbidities Δ (%)
Diabetes mellitus 559 (15.6) 499 (15.5) 60 (16.3)
Current smoker 507 (14.1) 453 (14.1) 54 (14.6)
History of Severe COPD 115 (3.2) 101 (3.1) 14 (3.8)
Ascites 42 (1.2) 40 (1.2) 2 (0.5)
Congestive Heart Failure 11 (0.3) 8 (0.2) 3 (0.8)
Hypertension 1673 (46.6) 1489 (46.2) 184 (49.9)
Dialysis 20 (0.6) 15 (0.5) 5 (1.4)
Disseminated Cancer 342 (9.5) 302 (9.4) 40 (10.8)
Steroid / Immunosuppressant
Therapy 115 (3.2) 98 (3.0) 17 (4.6)
ASA Classification (%)
1 71 (2.0) 66 (2.0) 5 (1.4)
2 519 (14.5) 470 (14.6) 49 (13.3)
3 1728 (48.1) 1549 (48.1) 179 (48.5)
4 1243 (34.6) 1111 (34.5) 132 (35.8)
5 28 (0.8) 24 (0.7) 4 (1.1)
Page 41
28
Operative Factors
Elective Surgery (%) 2656 (74.0) 2385 (74.1) 271 (73.4)
Surgical Specialty (%)
General Surgery 1184 (33.0) 1063 (33.0) 121 (32.8)
Gynecology 511 (14.2) 467 (14.5) 44 (11.9)
Neurosurgery 638 (17.8) 571 (17.7) 67 (18.2)
Orthopedic Surgery 321 (8.9) 280 (8.7) 41 (11.1)
Plastic Surgery 119 (3.3) 107 (3.3) 12 (3.3)
Urology 338 (9.4) 304 (9.4) 34 (9.2)
Vascular Surgery 478 (133) 428 (13.3) 50 (13.6)
Wound Classification (%)
Clean 1683 (46.9) 1504 (46.7) 179 (48.5)
Clean - Contaminated 1769 (49.3) 1596 (49.6) 173 (46.9)
Contaminated 59 (1.6) 53 (1.6) 6 (1.6)
Dirty/Infected 78 (2.2) 67 (2.1) 11 (3.0)
Duration (minutes, median [IQR]) 120.0 [84.0, 191.0] 120.0 [84.0, 190.0] 125.0 [87.0, 209.0]
Length of Stay (days, median [IQR]) 3.0 [1.0, 6.0] 3 [1.0, 6.0] 4.0 [1.0, 7.0]*
Δ Definitions for co-morbidities are based on American College of Surgeons National Surgical Quality Improvement
Program Operations Manual (37)
* Statistical significance (p < 0.05) on univariable analysis
ASA = American Society of Anesthesiologists, COPD = chronic obstructive pulmonary disease, IQR = interquartile
range,
Page 42
29
Table 5-2: Pre-operative antibiotic administration characteristics
Pre-operative
Antibiotic† (%)
All
(n=3589)
No reported beta-
lactam allergy
(n = 3220)
Reported beta-
lactam allergy
(n = 369)
Cefazolin 3217 (89.6) 3074 (95.5) 143 (38.8)*
Ampicillin 10 (0.3) 10 (0.3) 0 (0.0)
Ceftriaxone 35 (1.0) 35 (1.1) 0 (0.0)
Ertapenem 1 (0.0) 0 (0.0) 1 (0.3)
Meropenem 2 (0.1) 1 (0.0) 1 (0.3)
Piperacillin-
Tazobactam 50 (1.4) 47 (1.5) 3 (0.8)
Ciprofloxacin 20 (0.6) 7 (0.2) 13 (3.5)*
Clindamycin 197 (5.5) 5 (0.2) 192 (52.0)*
Metronidazole 915 (25.5) 840 (26.1) 75 (20.3)*
Gentamicin 13 (0.4) 0 (0.0) 13 (3.5)*
Vancomycin 12 (0.3) 4 (0.1) 8 (2.2)*
None 66 (1.8) 62 (1.9) 4 (1.1)
* Statistical significance (p < 0.05) on univariable analysis
† Cumulative antibiotic use exceeds cohort size as some patients received multiple antibiotics
Page 43
30
Table 5-3: Antibiotic dose and timing of cefazolin
All
(n=3217)
No reported beta-
lactam allergy
(n= 3074)
Reported beta-
lactam allergy
(n=143)
Dose
1 gram 50 (1.6%) 48 (1.6%) 2 (1.4%)
2 gram 3150 (97.9%) 3011 (98.0%) 139 (97.2%)
3 gram 3 (0.1%) 3 (0.1%) 0 (0%)
Unknown 14 (0.4%) 12 (0.4%) 2 (1.4%)
Administration Time
Within 60 minutes of
surgical incision
3033 (94.3%) 2897 (94.2%) 136 (95.1%)
Outside 60-minute
window
157 (4.9%) 152 (4.9%) 5 (3.5%)
Unknown 27 (0.8%) 25 (0.8%) 2 (1.4%)
Page 44
31
Table 5-4: Antibiotic dose and timing of clindamycin
All
(n=197)
No reported beta-
lactam allergy
(n= 5)
Reported beta-
lactam allergy
(n=192)
Dose
1 gram 3 (1.5%) 0 3 (1.6%)
600 milligram 183 (92.9%) 5 (100.0%) 178 (92.7%)
900 milligram 7 (3.6%) 0 7 (3.6%)
Unknown 4 (2.0%) 0 4 (2.1%)
Administration Time
Within 60 minutes of
surgical incision
177 (89.8%) 4 (80.0%) 173 (90.1%)
Outside 60-minute
window
15 (7.6%) 1 (20.0%) 14 (7.3%)
Unknown 5 (2.5%) 0 5 (2.6%)
Page 45
32
Table 5-5: Reported beta-lactam allergy antibiotic
Beta-Lactam Antibiotic Frequency (%)
Penicillin 323 (85.4)
Amoxicillin 15 (4.0)
Cephalexin 14 (3.7)
Cefazolin 6 (1.6)
Piperacillin-tazobactam 5 (1.3)
Cephalosporin (unspecified) 5 (1.3)
Cefprozil 4 (1.1)
Ceftriaxone 3 (0.8)
Cefaclor 2 (0.5)
Ampicillin 1 (0.3)
Cumulative total exceeds reported beta-lactam allergy cohort size as some patients reported more
than one type of reaction
Page 46
33
Table 5-6: Allergic reactions to beta-lactam antibiotics
Reaction No. (%)
Rash 139 (36.8)
Unknown to patient 94 (24.9)
Hives 50 (13.2)
Swelling or angioedema 39 (10.3)
Anaphylaxis 37 (9.8)
Dyspnea 6 (1.6)
Nausea, vomiting, or diarrhea 5 (1.3)
Pruritis 2 (0.5)
Seizure 2 (0.5)
Syncope 2 (0.5)
Dizziness 2 (0.5)
Cough 2 (0.5)
Acute interstitial nephritis 1 (0.3)
Headache 1 (0.3)
Blistering rash 1 (0.3)
Cumulative total exceeds reported beta-lactam allergy cohort size as some patients reported more
than one type of reaction
Page 47
34
Table 5-7: Estimated effects of reported beta-lactam allergy and covariates on 30-day
surgical site infection in the multivariable logistic regression model
Variable Adjusted Odds Ratio 95% Confidence Interval
Age (per 10-year increase) 1.09 0.96 – 1.24
Gender (male) 1.32 0.97 – 1.81
Current smoker 1.55 1.03 – 2.32
Steroid / Immunosuppressant therapy 1.17 0.48 – 2.84
Diabetes 1.19 0.80 – 1.76
ASA Classification ≥ 3 2.95 1.47 – 5.88
Wound Classification (compared to clean)
Clean - Contaminated
Contaminated
Dirty / Infected
5.78
1.39
12.00
3.80 – 8.81
0.19 – 9.97
5.78 – 24.90
Reported Beta-Lactam Allergy 1.61 1.04 – 2.51
ASA = American Society of Anesthesiologists
Page 48
35
Figure 5-1: Natural effects model estimates for receipt of an alternate antibiotic to cefazolin
as the mediator
Although not shown in this diagram, the following covariates were included in the natural effects
model: age, gender, current smoker, diabetes mellitus, steroid/immunosuppressant use, American
Society of Anesthesiologists physical status classification, and wound classification.
aOR = adjusted odds ratio, CI = confidence interval
Page 49
36
Chapter 6 Discussion
Discussion
6.1 Summary of Results
Using a NSQIP-sampled, surgically diverse cohort spanning an 18-month period, a
reported beta-lactam allergy was associated with an increased risk of surgical site infection. This
effect was completely mediated by receipt of an alternate antibiotic to cefazolin. Our findings
add to the mounting evidence demonstrating the harms of avoiding preferred beta-lactam therapy
due to reported allergy.
6.2 Comparison with Previous Studies: Association between Reported Beta-Lactam Allergy and Surgical Site Infection
One prior study assessed the relationship between reported penicillin allergy and the risk
of SSI. Blumenthal et al. conducted a 5-year retrospective review of 9004 surgical procedures
and reported an adjusted odds ratio of 1.51 for developing a surgical site infection in those with a
reported penicillin allergy(6). Their study found that this association was completely mediated
by receipt of a non-beta-lactam antibiotic but was limited to five surgical procedures (coronary
artery bypass, colon surgery, hip arthroplasty, hysterectomy and knee arthroplasty)(6).
Our study demonstrated a similar effect size (adjusted odds ratio 1.61) that was mediated
through the receipt of an alternate antibiotic to cefazolin. In terms of covariates, patient age, and
gender distribution were similar between the two studies. However, our patient cohort tended to
have higher ASA physical classification status scores likely representing a less surgically fit
population. This finding is likely a result of differing inclusion criteria used between these two
studies. Blumenthal et al. included only 5 surgical procedures in their cohort, with only 1.6% of
Page 50
37
procedures performed in an emergency setting. In contrast, our study included over 20 surgical
procedures from 7 surgical specialties, with 26% of surgeries performed emergently. The
inclusion of a wider range of surgeries with a higher proportion of emergent surgeries likely
resulted in more varied patient population, higher medical complexity and less time for medical
optimization before surgery which greatly increases the external validity of our findings.
Peri-operative antibiotic use was also significantly different compared to Blumenthal et
al. In patients with a reported beta-lactam allergy, our institution reported a significantly higher
use of cefazolin (38.8%) compared to Blumenthal (12%)(6). Despite this finding, the distribution
of reported reactions was similar between both studies, which suggests that differences in
cefazolin use were likely related to the comfort level of surgeons and anesthesiologists at their
respective institutions. Potential explanations for greater comfort in using cefazolin at our
institution include the availability of allergy assessment services by the allergy and infectious
diseases services, the presence of hospital-wide antimicrobial stewardship initiatives to increase
awareness of the impact of reported allergies, and the presence of some anesthesiologists who
were aware of the misconceptions surrounding reported beta-lactam allergies and use of
cefazolin(34).
6.3 Study Strengths and Limitations
Our study strengthens the external validity of the association between reported allergy
and SSI. In contrast to Blumenthal et al, our study included 7 surgical specialties encompassing
over 20 different surgical procedures. This finding suggests that improving the use of cefazolin
prophylaxis among these patients could have a widespread impact on SSIs across surgical
specialties. In addition, our study included patients who reported any beta-lactam allergy rather
than penicillin allergy alone. We hypothesize that the impact of a reported penicillin allergy is
Page 51
38
not drug-specific, but rather applies to all beta-lactam antibiotics, since both result in the
avoidance of cefazolin.
Our study also provides additional insight into pre-operative antibiotic use and the
mediators of surgical site infection. In Blumenthal’s study, 34.7% of patients who reported
penicillin allergy received vancomycin as an alternate antibiotic(6). Pre-operative administration
of vancomycin is fraught with challenges including the need for more stringent weight-based
dosing, and unique to vancomycin, completion of the antibiotic infusion ideally 60 to 120
minutes before surgical incision(1). The challenges with antibiotic administration are reflected in
Blumenthal’s study, where 97.5% of patients did not receive vancomycin in the recommended
timeframe(6). Given a third of their patients received this antibiotic, it was unclear whether
inferior antibiotic efficacy or inappropriate timing of alternate antibiotics was the main
contributor to the increased SSI risk. In contrast, vancomycin utilization as pre-operative
prophylaxis at our institution is significantly lower, with only 2.2% of patients with reported
beta-lactam allergy receiving this antibiotic. By demonstrating a similar mediation effect,
combined with negligible use of vancomycin, our study strongly suggests that inferior efficacy of
alternate antibiotic agents is the main explanation for the increased SSI risk observed among
patients with reported beta-lactam allergy.
Our study has several limitations. First, the prevalence of reported beta-lactam allergy
(10.3%) is probably underestimated in our study, as accessibility to inpatient and outpatient
allergy testing services at our institution likely resulted in de-labeling of a significant proportion
of patients(40). Similarly, cefazolin use in patients with reported beta-lactam allergy were three
times higher (38.8%) compared to the results reported by Blumenthal et al. (12.2%)(6). It is
therefore possible that the impact of a reported allergy on surgical site infection is more
Page 52
39
pronounced at other institutions where pre-operative cefazolin utilization is lower. We cannot
exclude the possibility of residual confounding variables which could not be included in the
model due to limitations in sample size and outcome frequency. Finally, the findings of this
study cannot be applied to transurethral procedures because they were excluded from this study.
6.4 Future Direction
Patient and physician misconceptions surrounding penicillin allergies and cross-reactivity
with cephalosporins are prevalent(34), and interventions are urgently needed for clinicians to
accurately assess a reported allergy and safely choose the most appropriate pre-operative
antibiotic. Interventions such as beta-lactam skin testing and direct drug provocation have been
used in the peri-operative setting but have been limited in their scalability and sustainability due
to resource requirements. It is now recognized that cefazolin is a non-cross-reactive
cephalosporin with penicillin due to a structurally dissimilar side chain(2), which suggests that
the decision to administer cefazolin pre-operatively could theoretically be achieved with a
clinical history alone. Further studies are needed to determine whether such a strategy is
sustainable, and whether an increase in cefazolin use in patients with reported beta-lactam
allergy results in a decline in surgical site infections.
6.5 Conclusions
Reported beta-lactam allergy is associated with increased odds of surgical site infection
across a broad range of surgical specialties which appears to be mediated by use of alternate
agents to intravenous cefazolin. Given recent data suggesting that cefazolin is a non-cross-
reactive cephalosporin, there is great opportunity to improve SSI rates through interventions that
promote the safe use of cefazolin prophylaxis among patients reporting a beta-lactam allergy.
Page 53
xl
Part II: Evaluation of a Standardized Pre-operative
Allergy and Antibiotic Assessment Algorithm on Cefazolin Use in Patients with Reported Beta-Lactam Allergy
Undergoing Elective Surgery
Page 54
41
Chapter 7 Introduction
Introduction
Cefazolin is a cephalosporin belonging to the beta-lactam class of antibiotics and is the
recommended agent for prophylaxis in most surgical procedures(1). However, cefazolin is often
avoided in patients with a reported beta-lactam allergy. Patients with a reported beta-lactam
allergy who do not receive cefazolin as pre-operative prophylaxis are at an increased risk of
surgical site infections(6). Although several interventions have been described to improve pre-
operative use of cefazolin in this population, a scalable, and sustainable intervention developed
using quality-improvement methodology is lacking. With the growing evidence describing the
potential harms of a reported beta-lactam allergy, improving the use of cefazolin has the
potential to reduce surgical site infections. The second part of this thesis will review the scope of
the problem, potential causes and consequences, and interventions employed to date. A
description and evaluation of a quality improvement intervention consisting of a standardized
pre-operative allergy and antibiotic assessment algorithm in patients with reported beta-lactam
allergy follows.
7.1 Importance and Relevance of the Problem
The use of cefazolin as surgical prophylaxis is frequently avoided in patients with a
reported beta-lactam allergy. In a 5-year retrospective review of 9004 surgeries at Massachusetts
General Hospital (Boston, Massachusetts, USA), only 12.2% of patients with a reported
penicillin allergy received cefazolin as surgical prophylaxis(6). More locally, Michael Garron
Hospital (Toronto, Ontario) reported that approximately only 20% of patients with a reported
beta-lactam allergy received cefazolin as pre-operative prophylaxis prior to implementation of an
intervention(41). Similarly, Haslam et al. at Kingston General Hospital (Kingston, Ontario)
Page 55
42
reported that only 27% of patients with penicillin allergy undergoing primary hip and total knee
arthroplasties over a 4-year period received cefazolin as surgical prophylaxis(42). At
Sunnybrook Health Sciences Centre (SHSC), approximately 39% of patients with a reported
beta-lactam allergy undergoing elective surgery received cefazolin as surgical prophylaxis over
an 18-month period using NSQIP sampling data (Table 5-2). No interventions have been
employed at SHSC to improve pre-operative cefazolin use in this population.
Unfortunately, current clinical practice guidelines for surgical prophylaxis are outdated
with respect to which patients with beta-lactam allergy can safely receive cefazolin(1). Some
antimicrobial stewardship groups across the country have begun addressing this issue through the
release of practice recommendation documents(43). Choosing Wisely campaigns in Canada and
the United States have highlighted the potential harms of avoiding beta-lactam antibiotics in the
treatment of infection, but have not specifically addressed this in the peri-operative
setting(44,45). Consensus recommendations published by a working group published in the
British Journal of Anesthesia in 2019 provided guidance on the management of surgical patients
with penicillin allergy, but ultimately still recommended traditional diagnostic techniques such as
skin testing, and direct drug provocation despite acknowledging the lack of cross-reactivity
between penicillin and cefazolin(46).
7.2 Causes of the Problem
The tendency to avoid cefazolin (and other cephalosporins) in patients with a reported
penicillin allergy (the most commonly reported beta-lactam allergy) originate from two
retrospective studies of clinical trial data from the 1970s(47,48). In these studies, the risk of
cephalosporin allergy in patients with an underlying penicillin allergy were estimated to be as
high as 8.1%(47). The risk reported in these studies have since been found to be grossly
Page 56
43
overestimated(49). Cephalosporins manufactured prior to 1980 were contaminated with trace
amounts of penicillin, which likely contributed to the high incidence of allergic reaction in
patients with known penicillin allergy(49). Furthermore, the definition of an allergic event was
poorly defined, and the studies failed to account for the increased general tendency for drug
reactions in patients with penicillin allergy(49).
Over the past decade, there has been accumulating evidence dispelling the overestimated
risk of cross-reactivity. Current evidence suggests that most allergic reactions to beta-lactams are
driven by the structure of the R1 side chain, rather than the beta-lactam ring itself(2). Therefore,
cephalosporins with a structurally dissimilar R1 side chain (such as cefazolin) carry no increased
cross-reactivity risk(2).
Despite these revelations, there is marked variability in physician knowledge and comfort
when prescribing cephalosporins in patients with penicillin allergy(50,51). A literature review by
Wanat et al. identified 21 papers describing patient and prescriber behaviors surrounding
penicillin allergies(14). Common prescriber themes included avoidance of beta-lactams in
penicillin-allergic patients to “play it safe”, with the view that alternate antibiotics with
comparable safety and efficacy exist(14). Even in the presence of a negative allergy skin test
result, both physicians and patients continued to demonstrate reluctance to prescribe and receive
penicillin antibiotics, citing fear of a subsequent reaction, uncertainty of the severity of reaction,
and lack of confidence in the safe administration(14). These findings suggest that physician and
patient misconceptions surrounding penicillin allergies need to be directly addressed before
changes in prescribing behavior can occur.
Page 57
44
Chapter 8 Literature Review
Literature Review
Interventions aimed at improving pre-operative prescribing of cefazolin in patients with
reported beta-lactam allergy have included a standardized allergy history, allergy skin testing,
and direct drug provocation. A literature review identified six studies utilizing beta-lactam
allergy skin testing(35,52–56), one study utilizing direct drug provocation(57), and two studies
using a structured allergy history(41,58) (Table 8-1). In the seven studies where baseline or
control data was available, all but one study demonstrated either a reduction in alternate
antibiotic use or increase in cefazolin use (41,52–54,56,58). A review of each study based on
intervention type is summarized below. The search strategy used for this literature review is
included in Appendix 3.
8.1 Beta-Lactam Allergy Skin Testing
Beta-lactam allergy skin testing has been the most commonly reported intervention
employed in the pre-operative setting. Testing consists of skin prick and intra-dermal
administration of a negative control (normal saline), positive control (histamine) and a series of
commercially or hospital-produced penicillin derivatives. The formation of a wheal with
erythema when an antibiotic is administered in the presence of a valid negative control test
indicates a positive reaction. With the high negative predictive value of penicillin skin testing (97
– 99%), patients with a negative result are effectively “de-labelled” of their beta-lactam allergy
and can tolerate all beta-lactam antibiotics(59).
The first study to illustrate the utility of penicillin skin testing in the pre-operative
population was published by Li et al. in 2000(52). Patients with penicillin or cephalosporin
Page 58
45
allergy planned for orthopedic surgery at the Mayo Clinic (Rochester, Minnesota, USA) were
referred by their surgeon for same-day allergy consultation and penicillin skin testing. Over a
one-year period, 60 patients were referred for allergy assessment. Of the 54 patients cleared to
receive cefazolin, 49 (89%) ultimately received cefazolin as pre-operative prophylaxis(52).
Compared to the 6-month period prior to the intervention, use of vancomycin decreased from
30% to 11% (p < 0.05)(52).
The same institution published results from a larger surgical cohort two years later(53).
Over a 14-month study period, 999 patients underwent skin testing for penicillin allergy, of
which 946 (94.7%) tested negative(53). Of the 905 patients who were permitted to receive
cefazolin, 660 (72.9%) ultimately received cefazolin; reasons for non-adherence to
recommendations were not explored(53). Use of vancomycin decreased to 16% compared to
baseline controls (30%, p < 0.01), but baseline use of cefazolin was not known(53). Four patients
with a negative penicillin skin test developed a probable adverse event to cefazolin, but all
recovered without any complications(53). The authors concluded that penicillin skin testing in
the pre-operative setting was an effective means to reduce vancomycin use. This intervention
excluded cardiovascular, thoracic and pediatric surgeries as these surgical clinics were not
situated close to the allergy clinic(53).
In 2008, Frigas et al. from the Mayo Clinic further expanded on the surgical scope of
their penicillin skin testing(54). In contrast to the previous two studies which enrolled patients by
physician referral, patients with penicillin allergy undergoing all types of surgery were identified
using a patient-completed questionnaire in the pre-operative clinic(54). Those with non-life-
threatening reactions to penicillin were offered penicillin allergy skin testing in the presence of
an allergy consultant. Over a 6-month period, 412 consecutive patients were included in the
Page 59
46
study(54). Following testing, 68% of patients with reported penicillin allergy received cefazolin
in comparison to only 33% at a control institution where allergy skin testing was not offered
(p<0.001)(54). Baseline prescriber and patient characteristics between the intervention and
control institutions were not reported, thereby limiting the internal validity of the results. No
peri-operative drug reactions were noted during the intervention period.
Cook et al. reported a referral-based penicillin allergy skin testing service for patients
undergoing cardiac surgery at the Mayo Clinic in 2009(55). Patients without a history suggestive
of anaphylaxis were referred by their surgeon for penicillin skin testing. Over a five-year period,
276 patients underwent pre-operative skin testing(55). Vancomycin use in the penicillin-allergic
population did not change appreciably before (7.1%) or after the intervention (7.9% at year
4)(55). The authors do not state how the results of the skin test were communicated to the
surgeon. Given the known incidence of penicillin allergy, the skin testing service received an
unexpectedly low number of referrals (48% of patients)(55). The authors attributed the low
uptake to poor provider awareness, remoteness of the skin testing facility from clinic, scheduling
challenges, and patient refusal(55).
Over a 2-year period, McDanel et al. (University of Iowa Hospitals, Iowa, USA)
described 161 patients undergoing joint arthroplasty referred to a drug allergy clinic for
penicillin skin testing, and direct drug provocation(56). Of the 145 patients deemed safe to
receive cefazolin, 90% received cefazolin as pre-operative prophylaxis, compared to 77% of
patients who were not seen in the drug allergy clinic (p=0.005)(56). No perioperative adverse
reactions were noted during the study period. However, there were 143 patients unable to be
evaluated in the drug allergy clinic despite being eligible(56). The authors attributed this to
Page 60
47
errors in placing the electronic consult, administration errors, and nursing misinterpretation of
the screening allergy questionnaire(56).
In Canada, Moussa et al. at McGill University Health Centre (Montreal, Quebec)
implemented an allergist-led evaluation of beta-lactam allergies in surgical patients consisting of
a history-based risk assessment, beta-lactam skin testing and direct drug provocation(35). Of the
194 patients referred for evaluation, 94.3% of patients were successfully de-labelled of their
beta-lactam allergy(35). Of those who were de-labelled, 77.4% of patients received cefazolin as
antibiotic prophylaxis; there was no baseline data or control group for comparison(35). The
authors also reported a 22 minute delay to surgical incision in those receiving vancomycin
compared to cefazolin, and suggest that reducing pre-operative vancomycin use in their entire
surgical population has the potential to save $350,867 CAD in annual operating time costs(35).
8.2 Direct Drug Provocation Testing
There is growing evidence that direct drug provocation testing (without preceding skin
testing) can be safely used in patients with a low-risk penicillin allergy history(60,61). Although
allergy skin testing has a high negative predictive value, there are concerns that a positive result
may not accurately predict an actual reaction, hence unnecessarily excluding a proportion of
patients who could have otherwise tolerated the antibiotic(61). As such, direct drug provocation
may represent a simpler, more cost-effective alternate compared to beta-lactam skin testing in
certain individuals.
To date, only one direct drug provocation intervention has been studied in the pre-
operative population. Savic et al. shared their experiences with a nurse-led allergy questionnaire
coupled with direct drug provocation in patients undergoing elective surgery(57). Recruitment
into study was performed on an ad hoc basis based on nursing availability. Patients who were
Page 61
48
deemed low-risk for an IgE-mediated allergy were offered an oral antibiotic challenge on site,
followed by an incremental 3-day dose of amoxicillin to be completed at home(57). Of the 74
patients who met the low-risk criteria, 56 patients agreed to participate in the study, and 55
completed the oral challenge successfully(57). At follow-up, 17 of 19 patients who successfully
completed the oral challenge received beta-lactam surgical prophylaxis(57). The authors do not
comment on the outcome of the other 37 patients who successfully tolerated the oral challenge.
The authors also sought to identify patient barriers with their intervention; they found that
anxiety levels on the day of testing were generally low(57). Interestingly, 70% of patients
interviewed stated that they would not be comfortable with removal of an allergy label based on
history alone(57). Patients valued the presence of a medical professional as an additional form of
security(57).
8.3 Structured Allergy History
Two studies to date have utilized an allergy assessment as the sole modality for
improving pre-operative cefazolin use. Vaisman et al. at Michael Garron Hospital in Toronto
published an intervention which consisted of a structured allergy history assessment by
pharmacists in their pre-operative clinic(41). The allergy history was subsequently reviewed by
an Infectious Disease physician followed by an antibiotic recommendation(41). Patients were
deemed safe to receive cefazolin unless they described a history of an IgE-mediated reaction in
keeping with anaphylaxis, or a severe non-IgE mediated reaction such as Steven Johnson
Syndrome(41). Over a 3-year period, 485 patients with beta-lactam allergy underwent an allergy
history assessment(41). With this intervention, the proportion of patients with beta-lactam
allergy who received a non beta-lactam antibiotic decreased from 81.9% to 55.9% (p <
0.001)(41). The monthly proportion of patients with reported beta-lactam allergy who received
Page 62
49
cefazolin increased to approximately 85% by the end of the study period(41). There were no
adverse reactions to cefazolin noted.
More recently, Margallo et al. at Cedars-Sinai Medical Centre (Los Angeles, California,
USA) described a 5-question structured allergy assessment implemented in the pre-operative
clinic and performed by nursing staff(58). This intervention was supplemented by pharmacists
providing in-service education sessions to nursing staff to raise awareness of the importance of
an accurate allergy assessment. The intervention did not include a pre-operative antibiotic
recommendation, and surgeons were allowed to prescribe antimicrobial prophylaxis at their own
discretion. Over a 5-month period, 364 patients reported a beta-lactam allergy with 313 receiving
pre-operative prophylaxis. A beta-lactam antibiotic was prescribed in 245 of 313 patients
(78.3%) which was modestly increased from the 24-month pre-intervention period (72.5%,
p=0.04)(58). There were no adverse effects noted during the intervention period. Compared to all
other studies reviewed, this study reported the highest use of beta-lactam antibiotics in patients
with reported beta-lactam allergy during the baseline period.
8.4 Implementation Challenges
Resource utilization was a common theme identified in most studies. Beta-lactam skin
testing for example, requires rapid access to an allergist for assessment, a trained nurse for
intradermal and skin prick testing, and procurement of normal saline, histamine and penicillin
derivatives for skin testing. As a result, all beta-lactam skin testing studies were performed at a
single-center, and many limited their intervention to a select surgical population (52,53,55,56).
Given these requirements, the scalability and adaptability of this intervention to smaller hospitals
with limited resources poses a significant challenge. Without studies to demonstrate the cost-
Page 63
50
effectiveness of skin testing in the pre-operative setting, hospitals may be reluctant to devote
staffing and infrastructure to support this type of intervention.
Even with implementation of a structured allergy assessment described by Vaisman et al.,
this required one-on-one pharmacist training with an Infectious Diseases physician over a 4-
month period. Since the study has been published, budget cuts have resulted in a reduction in
pharmacist staffing in the pre-admission clinic, thereby affecting the sustainability of their
intervention (personal communication, Dr. Janine McCready).
Further to the challenge of resource utilization, five studies relied on physician referral to
enroll patients into their intervention(35,52,53,55,56). In three studies, the authors noted that the
capture rate of patients with beta-lactam allergies was suboptimal, citing low physician
awareness of the availability of the intervention, and challenges with clinic
accessibility(35,55,56). The use of a referral system may result in selection bias of patients
enrolled in the intervention. For example, clinicians may choose to avoid referring certain
patients with a reported allergy due to their underlying suspicion of a true allergy. Similarly,
some physicians may not appreciate the consequences of prescribing alternate antibiotics and
therefore may not see the value of a formal allergy assessment. The pitfalls of a referral system
highlight the importance of developing an intervention that can be scaled to encompass the entire
pre-operative population.
Furthermore, prescriber uptake of antibiotic recommendations (in those deemed safe to
receive cefazolin) was not universal. There was significant variability in how the results of an
allergy assessment were communicated to the prescribing physician. Forms of communication
included notifying the referring surgeon of the test result(52–54,56,57), giving the patient a card
with their updated allergy status(53,54,57), and entering results into the electronic health
Page 64
51
record(35,54,57). Considering the complex pathways that patients follow in the pre-operative
period, potential points of miscommunication include: conveying the recommendations to a
physician who does not ultimately prescribe the pre-operative antibiotic, failing to access the
antibiotic recommendations on the day of surgery, and resistance to change prescribing
behaviors due to ongoing misconceptions surrounding penicillin allergies.
The converse situation, where a physician inadvertently prescribes cefazolin despite a
recommendation against its use could potentially cause significant harm. Park et al. reported in
their study that 164 patients received cefazolin or a beta-lactam antibiotic despite an allergist
advising against its use(53). One of these patients subsequently developed hypotension to
cefazolin but recovered without complication(53). This event prompted the institution to enter
the allergy test results into the electronic health record and provide each patient a card with the
allergist’s recommendations(53).
The method of communication may be as important as the actual intervention itself. For
instance, Vaisman et al. conveyed their recommendations by directly modifying the electronic
pre-operative antibiotic order using a pharmacist medical directive(41). With an allergy history
assessment alone, only 9.5% of patients in which cefazolin was recommended received an
alternate antibiotic(41). In contrast, Moussa et al. conveyed their antibiotic recommendations by
uploading documentation through an electronic health record module(35). Despite utilizing a
more extensive and definitive allergy assessment, 23% of these patients still received an alternate
antibiotic(35). An editorial published by these authors acknowledged this shortcoming and
attributed the low uptake of antibiotic recommendations to delays in the electronic health record
updating process(62). The study authors have since modified their intervention to include real-
time updating of the health record(62). The authors also noted that 15 patients who were
Page 65
52
successfully de-labelled but still received an alternate antibiotic, continued to claim they were
allergic to beta-lactam antibiotics on the day of surgery(62). These contrasting results suggest
that effective communication of antibiotic recommendations not only from provider to provider,
but from provider to patient, are important contributors to the success of an intervention.
Finally, none of the studies identified in the literature utilized quality improvement
methodology or principles to develop or optimize their intervention. Baseline data was generally
lacking in most studies and when reported, were presented in a pre-post fashion rather than a
time series analysis. One study used a run chart to display the outcome and process measures
during the baseline and intervention period, but iterative testing was not explicitly used to refine
their intervention(41). Margallo et al. reported implementing their structured allergy assessment
using the “Plan, Do, Study, Act” principle but did not provide any details on their prediction or
iterative changes made to refine their intervention (58). Their intervention, which did not target
the prescribers of pre-operative antibiotics, was associated with a rather modest increase in beta-
lactam utilization (72.5% to 78.3%) and was evaluated using an uncontrolled before-after
evaluative design which calls into question any causal inference.
Given the implementation challenges identified above, the ideal intervention that will be
both effective and sustainable is one that is not resource intensive, prescriber-led and can be
integrated into the current workflow of perioperative assessment and management. Iterative
quality improvement methodology is essential to ensuring that these ‘key ingredients’ are
embedded in the intervention.
Page 66
53
Table 8-1: Summary of interventions aimed at improving pre-operative beta-lactam antimicrobial use
Author (Year) Beta-Lactam Allergy
Surgical Population
Patient Identification Intervention Outcome Safety
Li et al.
(2000)(52)
Orthopedic
(n = 60)
Physician referral Penicillin skin testing Reduced vancomycin use (11%
vs 30%, p<0.05)
No peri-operative
reactions to cefazolin
Park et al.
(2002)(53)
Excluded cardiovascular,
thoracic, and pediatric
surgeries
(n = 1204)
Physician referral Penicillin skin testing Reduced vancomycin use (16%
vs 30%, p<0.05)
4 peri-operative reactions
to cefazolin in skin test
negative patients – no
long-term sequelae
Frigas et al.
(2008)(54)
All
(n=412)
Patient-completed
questionnaire in pre-
operative clinic
Penicillin skin testing 68% of those with negative test
result received cefazolin
(33% at control institution,
p<0.001)
No peri-operative
reactions to cefazolin
Cook et al.
(2014)(55)
Cardiac Surgery
(n=276)
Physician referral Penicillin skin testing No change in vancomycin use
after 4 years (7.1% vs 7.9%)
Not reported
Margallo et al.
(2019)(58)
All patients
(n=364)
Patient-completed
questionnaire in pre-
operative clinic
Structured allergy history
by nurse
Increased cefazolin use from
72.5% to 78.3% (p=0.04)
No peri-operative
reactions
Page 67
54
McDanel et al.
(2017)(56)
Orthopedic
(n=161)
Physician referral Penicillin skin testing
and direct drug
provocation
Increased cefazolin use in those
assessed in drug allergy clinic
(90% vs 77%, p=0.005)
No peri-operative
reactions to cefazolin
Moussa et al.
(2018)(35)
All
(n=194)
Physician referral Penicillin skin testing
and direct drug
provocation
94.3% of patients successfully
de-labelled of allergy
77.4% of patients de-labelled
received cefazolin
No baseline or comparator group
Not reported
Savic et al.
(2018)(57)
All patients at low-risk of
IgE-mediated allergy
(n=219)
Patient-completed
questionnaire in pre-
operative clinic –
dependent on nursing
availability for enrolment
Direct drug provocation
98.2% of low-risk patients
successfully de-labelled of
allergy
17 of 19 patients received beta-
lactam prophylaxis
No baseline or comparator group
Not reported
Vaisman et al.
(2017)(41)
All
(n=485)
Patient-completed
questionnaire in pre-
operative clinic
Structured allergy history
by Pharmacist, reviewed
with Infectious Diseases
physician
Reduction in alternate antibiotic
use after the intervention (55.9%
vs 81.9%, p<0.001)
No peri-operative
reactions to cefazolin
Page 68
55
Chapter 9 Objectives
Objectives
9.1 Study Aim
The following aim statement was developed using the SMART framework (specific,
measurable, actionable, realistic and timely) at the beginning of the quality improvement project
to set a concrete goal for the study team:
“By May 2019, the proportion of patients with reported beta-lactam allergy undergoing elective
surgery who receive cefazolin as pre-operative prophylaxis will increase by 40% (from 45 to
85%).”
9.2 Study Objectives
Primary Objective
• Compare the proportion of patients with reported beta-lactam allergy undergoing elective
surgery who received cefazolin as pre-operative prophylaxis before and after the
intervention (primary outcome measure)
Secondary Objectives:
• Compare the proportion of patients with reported beta-lactam allergy undergoing elective
surgery who developed a surgical site infection before and after the intervention
(secondary outcome measure)
• Compare the proportion of patients with reported beta-lactam allergy undergoing elective
surgery who developed a peri-operative allergic reaction before and after the intervention
(balancing measure)
Page 69
56
Chapter 10 Methodology
Methodology
10.1 Overview of the Study Design
An interrupted time series analysis was conducted to evaluate the impact of the quality
improvement intervention on pre-operative antibiotic use (primary outcome measure) during the
Baseline (May 1, 2018 – November 30, 2018), Implementation (December 1, 2018 – January 31,
2019) and Intervention (February 1, 2019 – August 31, 2019) periods of the study. An
uncontrolled before-after study design was used to evaluate the impact of the intervention on
surgical site infection (secondary outcome measure) and peri-operative allergic reactions
(balancing measure).
10.2 Ethical Considerations
The Ethics Review – Self Assessment Tool (ER-SAT) is a tool developed by the
Research Ethics Board (REB) at Sunnybrook Health Sciences Centre to help facilitate decisions
surrounding whether formal REB approval is required for quality improvement projects. Using
ER-SAT, a formal REB review was deemed unnecessary for this quality improvement project.
However, because additional variables were collected as part of the formal evaluation of the
intervention, a research protocol was submitted to the REB for delegated review. Approval was
received on July 22, 2019 (Project Identification Number: 235-2019).
10.2.1 Setting
Sunnybrook Health Sciences Centre (SHSC) is a 627-bed academic tertiary care centre
which performs 15,000 surgical procedures per year. Approximately 70% of surgical procedures
performed at SHSC are elective in nature. Over 70 surgeons and 40 anesthesiologists provide
Page 70
57
peri-operative care to patients at SHSC. Patients who are scheduled for an elective surgery are
referred to a pre-operative clinic where they are seen 2 to 3 weeks before surgery to ensure their
medical co-morbidities are optimized prior to surgery. Patients referred to the clinic are triaged
by a nurse who determines the most appropriate appointment type.
Patients with an uncomplicated medical history who are not undergoing major surgery
(i.e. open body cavity surgery) and physically and psychologically prepared for the procedure
undergo a telephone assessment with a nurse. Patients who do not meet criteria for telephone
assessment or those who require additional pre-operative testing are scheduled for an on-site
appointment with a nurse. Depending on the patient’s medical history, prior anesthetic
complications, or concerns raised by the surgeon, patients may be additionally assessed by an
anesthesiologist, internist and/or pharmacist. On any given day, the pre-operative clinic is
operated by 2 to 3 administrative staff, a nursing team leader, 4 nurses, 1 to 2 anesthesiologists,
and 1 pharmacist. An allergy history is taken as part of the health assessment by nursing,
pharmacy and anesthesia and documented electronically. Pre-printed orders for pre-operative
antibiotics are available for some elective surgical procedures but guidance for which antibiotic
is most appropriate in the presence of an allergy is limited.
Suspicions of inappropriate pre-operative antibiotic selection in patients with reported
beta-lactam allergy at SHSC were raised by data from Michael Garron Hospital illustrating the
extent of this problem at their institution(41). Parallel to this, concerns were raised by the
Infection Prevention & Control Department following a review of cardiac surgical site infection
cases during 2018. Findings from their review revealed several cases of surgical site infection in
patients with reported beta-lactam allergy who received vancomycin instead of cefazolin. Further
Page 71
58
review of these cases revealed that these patients could have received cefazolin safely based on
their allergy history.
These findings prompted an institution-wide review of pre-operative cefazolin use in
patients with reported beta-lactam allergy, and whether receipt of an alternate antibiotic to
cefazolin was associated with increased surgical site infection risk (Part I). A retrospective
review of 3589 surgical procedures sampled by NSQIP over an 18-month period found that only
38.8% of patients with reported beta-lactam allergy received cefazolin. Using a multivariable
regression model to account for possible confounders, a reported beta-lactam allergy was
associated with an increased risk of surgical site infection (adjusted odds ratio 1.61, 95% CI:
1.04 – 2.51, p =0.03)(Table 5-7). Mediation analysis confirmed that this effect was completely
mediated by receipt of an alternate antibiotic to cefazolin. The findings of this study served as a
burning platform to change allergy and pre-operative antibiotic assessments.
Prior to this project, there were no quality improvement interventions aimed at improving
pre-operative cefazolin use in patients with reported beta-lactam allergy. Improving pre-
operative cefazolin use however, aligned well with the strategic goals outlined by the institution.
SHSC published eight strategic quality goals for the five-year period spanning 2016 to 2019,
with its first goal to eliminate preventable harm (“Lead in Providing the Safest Care Based on
Best Evidence and Practice”). This goal explicitly targeted reducing cardiac surgical site
infection rates to below NSQIP benchmarks. Through improving cefazolin use in patients with a
reported beta-lactam allergy, this project has the potential to eliminate preventable harm (SSI)
across multiple surgical specialties and eliminate the downstream healthcare costs related to
infections.
Page 72
59
10.2.2 Study Team
A study team was formed at the beginning of the project to facilitate stakeholder
engagement, quality improvement diagnostics, procurement of data, and ongoing feedback after
implementation. The chief of surgery at SHSC served as the executive sponsor for the project,
which was essential in the removing barriers to change. As the pre-operative clinic served as the
initial target of our intervention, the nursing team leader, and medical director of the clinic were
part of the team. Four pharmacists, and one anesthesiologist (all of whom worked in the pre-
operative clinic) were also included on the team. The anesthesiologist was chosen based on an
established interest in this quality improvement problem and additionally served as a champion
for this project. An anesthesiology fellow who was completing a certification course in quality
improvement also joined the team to provide additional insight. To ensure timely access to
procedural logs and allergy information, a surgical quality improvement analyst, and
antimicrobial stewardship pharmacist were essential members of the team. Finally, given the
relationship between antibiotic prophylaxis and surgical site infection, the Medical Director of
Infection Prevention and Control (IPAC) and three IPAC coordinators were involved to ensure
ongoing surveillance of surgical site infections.
10.2.3 Study Participants
Adult patients (18 years or older) with a reported beta-lactam allergy undergoing an
elective surgical procedure at the Bayview site of SHSC during the study period were included in
the intervention.
Page 73
60
10.3 Quality Improvement Diagnostics
10.3.1 Process Map
In order to understand the process of an allergy assessment in the pre-operative period,
nursing, pharmacists and anesthesiologists working in the pre-operative clinic, same-day unit and
operating room were shadowed during their workday. Completing a process map was essential in
understanding all the steps leading up to pre-operative antibiotic selection and identifying all
potential stakeholders which could be impacted by the intervention. The process map for allergy
assessment and pre-operative antibiotic selection is summarized in Figure 10-1.
Process mapping revealed that 3 to 6 allergy assessments were conducted by different
healthcare providers in the pre-operative period and documented in separate sections of the
patient’s chart, highlighting significant waste in the system. During the shadowing process, it
was noted that pre-operative antibiotic orders were often absent in the chart on the day of
surgery. To verify this observation, an audit of 88 medical charts of patients undergoing elective
surgery revealed that 45% of patients did not have a pre-written antibiotic order. Furthermore, in
the cases where a pre-operative antibiotic order was written on the medical chart, about one half
of patients received a different antibiotic than what was written in the orders. These findings
strongly suggested that an additional assessment was taking place in the operating room suite
either by the surgeon or anesthesiologist which ultimately determined antibiotic selection. This
finding prompted additional qualitative data collection with anesthesiologists in order to clarify
the factors driving pre-operative antibiotic selection in the operating room.
Page 74
61
10.3.2 Semi-Structured Interviews
Appreciating that written pre-operative antibiotic orders were often discordant with the
antibiotic ultimately administered, semi-structured interviews were conducted with 5
anesthesiologists at SHSC. The following questions were asked to each anesthesiologist:
• Who ultimately decides which pre-operative antibiotic is given in the operating room?
• Do you look at the pre-operative antibiotic orders written by the surgeon on the order
sheets (when it is present)?
• What factors influence your decision on whether you can safely administer cefazolin in
patients with reported beta-lactam allergy?
All five anesthesiologists who were interviewed stated that the choice of pre-operative
antibiotic was determined during the “surgical time out” period before surgery and involved a
discussion between the surgeon and anesthesiologist. However, surgeons often deferred to
anesthesiologists’ expertise for antibiotic selection in many cases. More importantly, all five
anesthesiologists reported that pre-operative antibiotic orders written by the surgeon on the
medical chart were not reviewed in the operating room. These interviews strongly suggested that
interventions focused on changing anesthesiologist (rather than surgeon) prescribing behavior
would have the highest likelihood of success.
Semi-structured interviews also revealed that paper documents were the main mode of
communication between the anesthesiologist in the pre-operative clinic and the anesthesiologist
providing care to the patient on the day of surgery. As a result, an electronic-based intervention
would likely introduce additional barriers compared to an intervention that could be integrated
into the current paper documentation system.
Page 75
62
Based on these interviews, pertinent factors influencing whether cefazolin could be safely
given in patients with reported beta-lactam allergy included: type of allergic reaction, severity of
reaction, how long ago the reaction occurred and the offending agent. All anesthesiologists
reported that the main reason for avoiding cefazolin in patients with a reported beta-lactam
allergy was concern for precipitating an allergic reaction. Due to the risk for sampling bias
(anesthesiologists participating in the semi-structured interview may not reflect general
prescribing practices), conclusions could not be made regarding the relative importance of these
factors influencing the avoidance of cefazolin.
10.3.3 Anesthesiologist Survey
In order to assess antibiotic prescribing practice patterns on a larger scale, a survey was
administered to anesthesiologists at SHSC in November 2018 (Appendix 4) The goal of this
survey was to identify which aspects of a patient’s allergy history were most influential in their
decision to avoid cefazolin, and to gauge their interest in potential change concepts. There were
17 (38%) responses to the survey.
When asked about the most important components of an allergy history, the description
of the allergic reaction (12 out of 17, 86%) and when the reaction occurred (9/17, 56%) were the
most common responses. When presented with a scenario of a patient with a reported penicillin
allergy, respondents were most comfortable prescribing cefazolin when the reaction was
gastrointestinal in nature (16 out of 17, 94%), a childhood reaction (13 out of 17, 76%) and when
the reaction consisted of rash (13 out of 17, 76%). Respondents were least likely to prescribe
cefazolin when the patient reported angioedema (0 out of 17), laryngeal edema (0 out of 17) or
anaphylaxis (0 out of 17). Finally, when asked about type of interventions which would be most
useful in improving cefazolin use in patients with reported beta-lactam allergy, the three most
Page 76
63
commonly chosen interventions were an algorithm or decision-support tool (12 out of 15, 80%),
an educational intervention (8 out of 15, 53%) and formal allergy skin testing (6 out of 15, 40%).
Despite evidence demonstrating a lack of allergic cross-reactivity between penicillin and
cefazolin, almost all respondents did not feel comfortable prescribing cefazolin in patients with
moderate to severe reactions to penicillin (e.g. angioedema and anaphylaxis). This finding was in
keeping with published literature describing the prevalence of misconceptions surrounding
penicillin allergy and cross-reactivity. This finding highlighted the need to address these
knowledge gaps in order to maximize uptake of the proposed intervention.
10.3.4 Ishikawa Diagram
Based on the information gleaned from the various diagnostic tools, an Ishikawa diagram
was developed to identify the most pertinent contributors to avoidance of cefazolin in patients
with reported beta-lactam allergy (Figure 10-2). From a provider point of view, misconceptions
surrounding penicillin and cefazolin cross-reactivity, and an under-appreciation of the harms of
avoiding cefazolin were most important. System factors also contributing to avoidance of
cefazolin include the lack of a standardized process to assess a reported beta-lactam allergy and
guide pre-operative antibiotic selection.
10.4 Change Concept and Interventions
We theorized that anesthesiologists were inappropriately avoiding cefazolin in patients
with beta-lactam allergy due to fears of precipitating an allergic reaction. These fears were based
on outdated knowledge which grossly over-estimated the risk of cross-reactivity between
penicillin and cefazolin. Secondly, anesthesiologists were more likely to choose an alternate
antibiotic because they were not aware of the increased risk of surgical site infection, an
observation which has only been recognized in the past year.
Page 77
64
A driver diagram was developed to identify the primary and secondary drivers of changes
and the potential change ideas associated with each driver (Figure 10-3). We hypothesized that
standardizing the allergy assessment and pre-operative antibiotic selection process would
increase the use of cefazolin in patients with reported beta-lactam allergy. While avoidance of
cefazolin originated from misconceptions surrounding penicillin allergies, educational initiatives
aimed at addressing these misconceptions alone would have low impact and limited
sustainability particularly at a teaching hospital where there is large rotating cohort of
anesthesiology staff and trainees. In contrast, standardization although more challenging
intervention to implement, would have a greater impact and higher likelihood of a sustained
effect. Furthermore, we hypothesized that standardization in addition to reducing variability in
prescribing practice, would significantly simplify the allergy assessment process and save the
prescriber time compared to the current process. As a result, standardization of the allergy
assessment and pre-operative antibiotic selection was chosen as the main change concept for this
intervention.
10.4.1 Allergy and Pre-operative Antibiotic Assessment Algorithm
In order to achieve standardization, an algorithm was developed to determine whether a
patient with reported beta-lactam allergy could safely receive cefazolin (Figure 10-4). This
novel tool was developed through consensus by allergists, anesthesiologists and infectious
diseases specialists at our institution and in discussion with an allergist in the United States.
Anesthesiologists and pharmacists were trained to use the algorithm in the pre-operative clinic
whenever they encountered a patient who reported a beta-lactam allergy. Anesthesiologists were
also encouraged to use the algorithm on the day of surgery if it was not completed in the pre-
operative clinic. Based on prior chart reviews, we estimated that over 95% of patients with a
reported beta-lactam allergy could safely receive cefazolin. The algorithm was presented to an
Page 78
65
anesthesiologist staff meeting for their input and approval in December 2018. Use of the
algorithm was officially launched on January 14, 2019. A copy of the algorithm was posted in
each of the pre-operative clinic assessment rooms.
10.4.2 Stamp Communication Tool
Considering prior studies demonstrating the importance of communication, a stamp was
developed to convey the recommendation that cefazolin could be safely given on the day of
surgery (Figure 10-5). The stamp was imprinted in the allergy section of the anesthetic record
during the pre-operative assessment by pharmacists and anesthesiologists based on the result of
the algorithm. The anesthetic record (and stamp) would be subsequently reviewed by the
anesthesiologist on the day of surgery prior to antibiotic administration. The development of this
change idea was driven by semi-structured interviews which indicated that anesthesiologists
primarily utilize a paper record to communicate clinical issues identified during the pre-operative
assessment. A stamp was physically placed next to each algorithm in the pre-operative clinic
assessment rooms and was launched on January 14, 2019.
10.4.3 Education
Since misconceptions surrounding penicillin allergies and cross-reactivity were an
important factor identified during our diagnostics, we felt that an educational component aimed
at addressing these misconceptions would help maximize uptake of the algorithm. We
hypothesized that providing an evidence-based rationale for how the algorithm was developed
would increase anesthesiologists’ confidence in using the algorithm.
This quality improvement project was presented at Anesthesia Quality Improvement
Rounds on January 18, 2019 and was attended by approximately 20 anesthesiologists and
trainees. The session was also uploaded to YouTube and was viewed by 25 users(63). A
Page 79
66
presentation to the operating room nurses was carried out on January 11, 2019 which was
attended by over 30 nurses. Presentation to the clinical pharmacists at Clinical Pharmacology
Grand Rounds took place February 28, 2019 and was attended by over 30 pharmacists.
Finally, because our intervention was focused on changing anesthesiologist prescribing
behavior, a shift in ownership of pre-operative antibiotic selection in patients with reported beta-
lactam allergy was essential. To facilitate this process, a surgery-wide communication was sent
by our Executive Sponsor informing surgeons of this new intervention and asking them to defer
antibiotic selection to the anesthesiologist since they are using an evidence-based tool.
10.5 Family of Measures
10.5.1 Outcome Measures
The primary outcome measure was the proportion of patients with reported beta-lactam
allergy undergoing elective surgery who received cefazolin as pre-operative prophylaxis. The
denominator was the total number of patients with reported beta-lactam allergy who received a
pre-operative antibiotic for elective surgery. The numerator was the number of patients with
reported beta-lactam allergy who received cefazolin as pre-operative prophylaxis for an elective
surgical procedure.
In order to calculate this measure, a monthly log of patients undergoing elective surgery
was provided by the surgical performance analyst at SHSC. The list was cross-referenced with
the Stewardship Program Integrating Resource Information Technology (SPIRIT) database(38)
using the following allergy search query: “-cillin”, “pcn”, “pen g”, “pen v”, “cef-”, “ceph-”, or “-
penem” in order to identify those with a reported beta-lactam allergy. Furthermore, all
electronically documented allergies for patients undergoing elective surgery were manually
reviewed to ensure no beta-lactam allergies were missed using the query. For each patient with a
Page 80
67
reported beta-lactam allergy, the pre-operative antibiotic administered was collected by
reviewing the anesthetic record.
The secondary outcome measure was the crude rate of 30-day surgical site infection
based on the 2018 CDC/NHSN definition(23). Since surgical site infection outcome data was
unavailable for all patients with reported beta-lactam allergy during the study period, SSI data
collected by NSQIP and IPAC (irrespective of reported beta-lactam allergy) were used as an
indirect measure. SSI outcomes sampled by NSQIP were analyzed for the following procedures:
general surgery (hepatectomy, pancreatectomy, colectomy, low anterior resection), gynecology
(hysterectomy and myomectomy), neurosurgery (brain tumour procedures), and urology
(cystectomy, nephrectomy and radical prostatectomy). SSI outcome data collected by IPAC were
analyzed for consecutive cardiac bypass and valve replacement surgeries. These surgical
procedures were specifically chosen because outcome data was available throughout the entire
study period. Using both data sources, the rates of SSI during July to December 2018 were
compared to the rates during January to June 2019. Crude SSI rates for NSQIP and IPAC were
reported separately due to different data collection methods (sampling versus consecutive cases
respectively). Given the low frequency of surgical site infections, relatively short period of data
collection and the inability to capture SSI data for every surgical procedure, a before-after study
design was chosen to assess this outcome rather than a time series analysis.
10.5.2 Process Measures
To assess the use of the algorithm in the pre-operative clinic, evidence of stamp use was
assessed by chart review daily and aggregated into a monthly rate. A list of patients assessed in
the pre-operative clinic was generated daily, and each chart was reviewed for a reported beta-
lactam allergy (denominator) and the presence of a stamp (numerator).
Page 81
68
10.5.3 Balancing Measures
The number of patients who developed a peri-operative allergic reaction during the study
period was enumerated. A peri-operative allergic reaction was defined as the development of
symptoms intra-operatively compatible with an IgE-mediated hypersensitivity reaction (e.g.
urticaria, angioedema, bronchospasm, hypotension, anaphylaxis) and requiring treatment with
corticosteroids, epinephrine, or anti-histamines. The anesthetic record for each surgical
procedure during the study period was reviewed to assess for evidence of a peri-operative
allergic reaction. In the event of any peri-operative allergic reaction, the case was reviewed to
determine all medications received that could be implicated in causing the reaction.
10.6 Iterative Tests of Change
10.6.1 Usability Testing
During the first week of the Implementation period, pharmacists and anesthesiologists
working in the pre-operative clinic were engaged to provide input on the usability of the
algorithm. Users were asked to demonstrate the use of the algorithm (based on patients seen
earlier in the day) and provide suggestions on how the appearance of the algorithm could be
improved. We hypothesized that improvements in the design of the algorithm would improve the
uptake.
Common themes which emerged from these interactions included: users squinting at the
algorithm due to small font, confusion as to where the algorithm began, a suggestion of adding
colours (green for “safe to give cefazolin” versus red for “avoid cefazolin”) and a “take-home
point” to reinforce the rationale behind the algorithm. Based on this feedback, an updated version
of the algorithm was developed and posted in the pre-operative clinic assessment rooms (Figure
10-6). Users found the re-designed algorithm easier to follow, and appreciated the additional
Page 82
69
rationale included on the algorithm which helped to clarify previously held misconceptions
surrounding beta-lactam cross-reactivity.
10.6.2 Assessing Fidelity with Expansion to Nursing
Following usability testing, the process measure (stamp utilization) was tracked over the
following month to determine how frequently the algorithm was being utilized in patients with
reported beta-lactam allergy. Prior to implementation, it was known that healthy patients with no
co-morbidities undergoing minor surgical procedures were not seen by pharmacists or
anesthesiologists in the pre-operative clinic and would bypass this algorithm but the actual
proportion was unclear.
Following a two-week audit (January 14 to 29, 2019), only 73% (33 out of 45) of patients
with reported beta-lactam allergy were being assessed by a pharmacist or anesthesiologist.
Furthermore, of the patients with reported beta-lactam allergy that were seen by these two
professions, only 61% (20 out of 33) had the stamp imprinted. We hypothesized that having
nurses use the algorithm and stamp would help to improve the proportion of patients who have
the algorithm completed.
To test whether this was feasible, this idea was presented at the pre-operative clinic staff
meeting in February 2019. During this meeting, several nurses voiced concerns that use of the
algorithm would affect their workflow and potentially increase patient wait times. The nursing
staff also felt uncomfortable performing the algorithm because the act of recommending
cefazolin as pre-operative prophylaxis was felt to be beyond their scope of practice. Given this
feedback, the idea of having nursing staff perform the algorithm was not feasible and therefore
abandoned.
Page 83
70
10.6.3 Assessing for Internalization of Practice
Following the pre-operative clinic staff meeting, we chose to stratify patients with
reported beta-lactam allergy based on whether they were assessed by a pharmacist or
anesthesiologist versus nursing alone. We hypothesized that if the success of our intervention
truly depended on the use of the algorithm and stamp in the pre-operative clinic, then patients
who were assessed by nursing alone (no algorithm) would not experience an increase in
cefazolin use compared to their counterparts seen by a pharmacist or anesthesiologist.
The proportion of patients with reported beta-lactam allergy who received cefazolin as
pre-operative prophylaxis stratified by pharmacist or anesthesiologist versus nursing alone was
calculated. Contrary to our hypothesis, patients who were seen by nursing alone demonstrated a
parallel increase in the use of cefazolin following the roll-out of the intervention (see Results
Section 11.2). This finding strongly suggested that anesthesiologists were using the algorithm on
the day of surgery and that the stamp was not critical to their change in practice. This
stratification process also revealed that 25% of patients assessed by nursing alone did not receive
pre-operative antibiotic prophylaxis because it was not indicated for their surgery (e.g. minor
procedures such as cataract surgery, and hysteroscopy). Thus, the consequences of patients
bypassing the algorithm in the pre-operative clinic were less than predicted.
In order to confirm these findings, six anesthesiologists were contacted to ask about their
practices in choosing pre-operative prophylaxis in patients who were not assessed in the pre-
operative clinic. All anesthesiologists reported using the algorithm on the day of surgery, usually
through memory recall. Four anesthesiologists stated that because the algorithm only consisted
of three questions, it was easy to memorize once completed a few times. Two anesthesiologists
suggested that having the algorithm available in the operating room would help improve access.
Page 84
71
Based on this feedback, a laminated copy of the algorithm was attached to each anesthesia drug
cart in every operating room at SHSC (Figure 10-7).
10.6.4 Expansion to Nurse Practitioners
In March 2019, our study team was contacted by a cardiology nurse practitioner who was
interested in implementing the algorithm into the assessment of patients planned for cardiac
device implantation. This had come to their attention because anesthesiologists working in the
operating rooms were using the algorithm for their patients and administering cefazolin despite
pre-operative orders for an alternate antibiotic.
The population served by cardiology nurse practitioners was a unique in that most
patients bypass the algorithm in the pre-operative clinic because they are assessed by nursing
staff only. Furthermore, these patients underwent device implantation by an interventional
cardiologist rather than a surgeon and were therefore unaware of this quality improvement
project. Since cardiology nurse practitioners were responsible for ordering pre-operative
antibiotics prior their device implantation, we hypothesized that integrating the algorithm into
their practice (upstream from the pre-operative clinic) would further improve antibiotic
utilization.
An educational session with 4 nurse practitioners took place in April 2019. An e-mail
communication was also sent to the interventional cardiologists to inform them of this quality
improvement initiative.
10.7 Additional Variables of Interest
In addition to the outcome, process and balancing measures described above, the
following variables were collected for each patient: age, gender, American Society of
Page 85
72
Anesthesiologists Physical Status Classification, wound classification, surgical specialty,
description of allergic reaction, and type of pre-operative antibiotic administered. The
description of allergic reaction was obtained by review of the electronic health record. The type
of pre-operative antibiotic administered was collected through manual chart review of the
anesthetic record. All other variables automatically accompanied the monthly log of surgical
data.
10.8 Statistical Analysis
10.8.1 Data Management
The monthly logs of surgical procedure data were inspected for completeness and
additional chart review was completed to supplement any missing values. Care was taken to
maintain confidentiality by de-identifying all patient data and linking to a master list that was
stored separately. Encrypted devices and internal servers were exclusively used to store the study
data.
10.8.2 Descriptive Statistics
A descriptive analysis of patient demographics (age, gender), surgical specialty, ASA
physical status classification, wound classification and beta-lactam antibiotic allergy, and allergic
reaction was conducted in patients undergoing elective surgery during the Baseline and
Intervention periods. When comparing variables between the two time periods, Chi-square and
Wilcoxon signed-rank tests were performed for categorical and continuous variables
respectively.
10.8.3 Statistical Process Control Chart
A statistical process control chart was employed to measure change over time and to
assess for special cause variation. A p-chart was used for the primary outcome measure (monthly
Page 86
73
proportion of patients with reported beta-lactam allergy who received cefazolin) and process
measure (monthly proportion of patients with reported beta-lactam allergy who had stamp
imprinted on anesthetic record).
A T-chart was used to measure the time between peri-operative allergic events during the
Baseline and Intervention period given the rare incidence of these events. Statistical process
control charts were created using the QIMacros 2018 software (KnowWare International Inc,
Denver, Colarado, USA) extension for Microsoft Excel Office 365.
10.8.4 Stratification of Primary Outcome
Stratification based on whether patients were assessed by an anesthesiologist or
pharmacist (users of the algorithm) versus nursing alone (non-users) was carried out for the
primary outcome measure on a monthly basis. Stratification based on description of allergic
reaction was also performed for the primary outcome measure during the Baseline and
Intervention periods.
10.8.5 Segmented Regression Analysis
Following completion of the Intervention period, the primary outcome measure was
further evaluated using a segmented regression analysis. A level model was proposed and
modelled using a Poisson regression. The model was formulated as follows:
ln 𝜇𝑥 = ln 𝑡𝑥 + 𝛽0 + 𝛽1 × 𝑇 + 𝛽2 × 𝐼𝑥
where: µx = the number of patients with reported beta-lactam allergy who received cefazolin as
pre-operative prophylaxis for an elective surgical procedure each month
tx = the total monthly number of patients with reported beta-lactam allergy who received pre-
operative prophylaxis for an elective surgical procedure (an offset variable)
Page 87
74
β0 = the baseline use of cefazolin as pre-operative prophylaxis in the Baseline period
β1 = the monthly change (slope) in cefazolin use in the Baseline and Intervention period
(assumed to be equal)
β2 = the coefficient of change associated with Intervention
Ix = presence of the intervention (coded as 0 or 1)
The effect of the intervention was expressed using the natural exponential of β2 as an
incidence rate ratio. Data collected during the two-month Implementation period (December 1,
2018 to January 31, 2019) was censored from the analysis to account for the suboptimal uptake
during this period, and the lag period between the pre-operative assessment (where the
intervention was introduced) and the day of surgery (where the pre-operative antibiotic was
administered). In order to account for potential autocorrelation and heteroscedasticity, weighted
empirical adaptive variance estimators were used.
The secondary outcome measure (surgical site infection) and balancing measure (peri-
operative allergic event) were analyzed using a Chi-square test. The regression analysis and
statistical tests were performed using R Statistical Software 3.4.3 (Foundation for Statistical
Computing, Vienna, Austria) with the following packages: car, ggplot2 and sandwich.
10.8.6 Sample Size Calculation
In order to detect a change in proportion of patients with beta-lactam allergy who receive
cefazolin as pre-operative prophylaxis from 45% to 85%, with an alpha of 0.05 and power 0.80,
19 patients in the Baseline and Intervention period were needed. Approximately 30-50 patients
with beta-lactam undergo an elective surgery each month, so the specified time period was
powered to detect this change in antimicrobial use.
Page 88
75
A seven-month baseline and intervention period were chosen in order to confidently
assess for special cause variation and sustainability of the effect over time. Given the low
frequency of beta-lactam allergy (10% of the general population), and surgical site infections (1-
5% in the general population), this study was underpowered to detect a significant change in
surgical site infection. Assuming a baseline SSI rate of 8%, a 10% incidence of reported beta-
lactam allergy of which only 30% of patients receive cefazolin, a total sample size of 166,000
patients would be required to detect a 0.5% reduction in SSI rate. To be adequately powered, a
follow up study over 10 years would be required to assess this secondary outcome.
Page 89
76
Figure 10-1: Process map outlining allergy assessments in the pre-operative period
Page 90
77
Figure 10-2: Ishikawa diagram
Page 91
78
Figure 10-3: Driver diagram
Page 92
79
Figure 10-4: Allergy assessment and pre-operative antibiotic selection algorithm
Page 93
80
Figure 10-5: Stamp communication tool
Page 94
81
Figure 10-6: Allergy assessment and pre-operative antibiotic selection algorithm
redesigned following usability testing
Page 95
82
Figure 10-7 Example of a laminated algorithm attached to anesthesia drug cart in the
operating room
Page 96
83
Chapter 11 Results
Results
11.1 Characteristics of the Cohort
From May 2018 to August 2019, 745 patients with a reported beta-lactam allergy
underwent 789 elective surgical procedures where a pre-operative antibiotic was administered.
Of the 789 procedures, 279 (35.4%), 71 (9.0%) and 439 (55.6%) took place during the Baseline,
Implementation and Intervention periods. A summary of patient demographics, operative factors,
and reported allergy type and description are summarized in Table 11-1. Compared to the
Baseline period, patients in the Intervention period tended to have lower ASA physical status
classification scores. Surgical procedures performed in the patients with a reported beta-lactam
allergy in the Intervention period tended to be longer, with a lower proportion of burn surgery
procedures.
The distribution of the reported beta-lactam antibiotic allergy was similar between the
Baseline and Intervention periods, with the three most common reported antibiotics being
penicillin (77.9%), amoxicillin (8.4%) and cephalexin (3.5%) (Table 11-2). The allergic
reactions described were also similar during the two periods, with the three most commonly
described reactions being rash (38.7%), unknown reaction (19.2%) and hives (16.6%) (Table 11-
2). Patients in the Intervention period were less likely to have a documented unknown allergy
(16.2% vs 24.0%; p-value=0.01). Four patients during the entire study period reported an allergy
history suggestive of a severe delayed hypersensitivity reaction. Of the 13 patients who reported
a history of cefazolin allergy during the study period, 12 reported signs and symptoms suggestive
of a Type I (IgE-mediated) hypersensitivity reaction.
Page 97
84
11.2 Primary Outcome Measure: Cefazolin Use
During the Baseline period, 42.7% (119 out of 279) of surgical procedures in patients
with reported beta-lactam allergy received cefazolin as pre-operative prophylaxis. As illustrated
in the statistical process control chart (Figure 11-1), there was no evidence of special cause
variation during the Baseline period. At the beginning of the Implementation period (December
2018) when the algorithm was presented to anesthesiologists, an increase in cefazolin use was
noted, despite the intervention not being officially implemented until January 2019. A further
increase in the proportion of patients with reported beta-lactam allergy receiving cefazolin
occurred during the Intervention period and remained sustained. The observation of data points
exceeding the upper control limit, and more than 8 data points above the average baseline value
confirmed the presence of special cause variation. During the Intervention period, (376 out of
439) 85.6% of patients with a reported beta-lactam allergy received cefazolin as pre-operative
antibiotic prophylaxis. The segmented regression analysis confirmed a statistically significant
increase in the proportion of patients receiving cefazolin as pre-operative prophylaxis, with an
incidence rate ratio of 2.03 (95% CI 1.73 – 2.40, p = 0.002) (Figure 11-2).
The distribution of pre-operative antibiotic use in the Baseline and Intervention period is
summarized in Table 11-3. In addition to an increase use of cefazolin, there was a corresponding
decrease in clindamycin, gentamicin, metronidazole and vancomycin use during the Intervention
period. Of the 63 procedures in the Intervention period where an alternate antibiotic was
administered, 22 procedures were reviewed with the anesthesiologist assigned to the case to
understand their rationale for avoiding cefazolin (Figure 11-3). The most common reasons for
avoidance were concern for allergic cross-reactivity with cefazolin (8 out of 22, 36.4%), an
appropriate reason based on the algorithm (5 out of 22, 22.7%) and a different beta-lactam agent
Page 98
85
was administered at the request of the surgeon (4 out of 22, 18.2%). Of the 41 remaining surgical
procedures which were not reviewed, 17 (41%) of these cases involved a reported anaphylactic
reaction to a beta-lactam that was not cefazolin.
When the monthly proportion of patients with reported beta-lactam allergy who received
pre-operative cefazolin was stratified by whether patients were assessed by an anesthesiologist or
pharmacist (users of the algorithm and stamp) versus nursing alone (non-users of the algorithm)
in the pre-operative clinic, there was no significant difference in the proportion of patients who
received cefazolin on a month-to-month basis between the two groups (Figure 11-4).
Similarly, when the proportion of patients with reported beta-lactam allergy who received
pre-operative cefazolin was stratified based on description of allergic reaction, there was a
uniform increase in cefazolin use regardless of the type of the reaction (Figure 11-5). The
increase in cefazolin use was most marked in patients reporting rash, unknown reaction and
hives. Patients who reported anaphylaxis were the least likely to receive cefazolin in both the
Baseline (2 out of 19, 10.5%) and Intervention (15 out of 32, 46.9%) periods but still increased
significantly (p=0.006).
11.3 Secondary Outcome Measure: Surgical Site Infection
Surgical site infection outcome data is summarized in Table 11-4. There was no
statistically significant difference in surgical site infections rates during the two time periods,
although there were numerically fewer SSIs in cardiac surgical procedures following
implementation of the intervention.
Page 99
86
11.4 Process Measure: Stamp Utilization
During the Intervention period, 122 of the 439 (27.8%) patients were seen by nursing
staff alone (non-users of the algorithm and stamp), with the remaining patients seen by either an
anesthesiologist, pharmacist or both. When accounting for all pre-operative patients with
reported beta-lactam allergy, 47.6% of patients had a stamp imprinted on the anesthetic record
(Figure 11-6). When the denominator was restricted to only patients seen by an anesthesiologist
or pharmacist (users of the stamp), this proportion increased to 65.9% (Figure 11-7). There was
no evidence of special cause variation during the Implementation and Intervention periods.
11.5 Balancing Measure: Peri-operative Allergic Reaction
There were no documented peri-operative allergic reactions in the Baseline period and
two peri-operative allergic reactions (2 out of 439, 0.5%) in the Intervention period among those
with a reported beta-lactam allergy (Figure 11-8). Given the rarity of these events, there was
insufficient data to assess for special cause variation. There was no statistically significant
difference in the proportion of patients developing peri-operative reactions during the Baseline
and Intervention periods (p = 0.26).
In the first case, the patient developed urticaria and wheezing intra-operatively 20
minutes after cefazolin administration which responded to epinephrine and corticosteroids. In the
second case, the patient developed hypotension and increased airway pressure 20 minutes after
administration of cefazolin which responded to epinephrine. In the both cases, blue dye also was
administered at the beginning of the surgical procedure. The development of the peri-operative
allergic reaction did not affect the surgeon’s ability to complete the procedure, or the patient’s
course of hospitalization. Both patients will be further investigated by an allergist to determine
the most likely drug culprit of the hypersensitivity reaction.
Page 100
87
Table 11-1: Baseline characteristics of patients with reported beta-lactam allergy during
baseline and intervention periods
Baseline (n=279) Intervention (n=439)
Age (years, IQR) 61 [49, 72] 62 [51, 72]
Gender (% male) 98 (35.1) 161 (36.7)
ASA Class
I
II
III
IV
V
6 (2.2)
45 (16.1)
124 (44.4)
102 (36.6)
2 (0.7)
23 (5.2)
85 (19.4)
210 (47.8)
119 (27.1)
2 (0.5)
Surgical Specialty
Burn
Cardiac Surgery
General Surgery
Gynecology
Gynecology-Oncology
Neurosurgery
Ophthalmology
Oral Dentistry
Orthopedics
Otolaryngology
Plastic Surgery
Urology
Vascular Surgery
14 (5.0)
20 (7.2)
52 (18.6)
18 (6.5)
27 (9.7)
21 (7.5)
1 (0.4)
5 (1.8)
28 (10.0)
11 (3.9)
18 (6.5)
38 (13.6)
26 (9.3)
4 (0.9)*
21 (4.8)
96 (21.9)
39 (8.9)
41 (9.3)
46 (10.5)
2 (0.5)
9 (2.1)
45 (10.3)
30 (6.8)
23 (5.2)
56 (12.8)
27 (6.2)
Wound Classification
Clean
Clean-Contaminated
Contaminated
Dirty
144 (52.4)
119 (43.3)
1 (0.4)
11 (4.0)
234 (53.5)
197 (45.1)
1 (0.2)
5 (1.1)
Surgery Duration 142 [84.5, 225.5] 115 [73.0, 181.5]*
Page 101
88
Table 11-2: Distribution of reported beta-lactam antibiotic allergy
Baseline (n=279) Intervention (n=439)
Reported Beta-lactam*
Amoxicillin
Amoxicillin-Clavulanate
Ampicillin
Cefaclor
Cefazolin
Cefprozil
Ceftazidime
Ceftriaxone
Cefuroxime
Cephalexin
Cephalosporin unspecified
Cloxacillin
Penicillin
Piperacillin-Tazobactam
27 (9.7)
3 (1.1)
5 (1.8)
7 (2.5)
5 (1.8)
1 (0.4)
1 (0.4)
1 (0.4)
0
13 (4.7)
1 (0.4)
3 (1.1)
218 (78.1)
0
33 (7.5)
6 (1.4)
7 (1.6)
14 (3.2)
8 (1.8)
5 (1.1)
1 (0.2)
2 (0.5)
1 (0.2)
12 (2.7)
7 (1.6)
5 (1.1)
341 (77.6)
2 (0.5)
Allergic Reaction*
Rash
Unknown
Hives
Angioedema or swelling
Anaphylaxis
Dyspnea or difficulty breathing
Dizziness or syncope
Gastrointestinal
Other
97 (34.8)
67 (24.0)
42 (15.1)
28 (10.0)
19 (6.8)
9 (3.2)
5 (1.8)
11 (3.9)
14 (5.0)
181 (41.2)
71 (16.2)†
77 (17.5)
43 (9.8)
32 (7.3)
8 (1.8)
11 (2.5)
26 (5.9)
32 (7.3)
* Cumulative reported allergen and reaction type exceeds cohort size as some patients reported
multiple allergens and reactions
† Statistical significance (p < 0.05) on univariable analysis
Page 102
89
Table 11-3: Pre-operative antibiotic administration during baseline and intervention
period
Pre-operative Antibiotic* Baseline (n=279) Intervention (n=439)
Cefazolin 119 (42.7) 376 (85.6)†
Metronidazole 45 (16.1) 61 (13.9)†
Clindamycin 122 (43.7) 44 (10.0)†
Gentamicin 11 (3.9) 2 (0.5)†
Piperacillin-tazobactam 1 (0.4) 2 (0.5)
Meropenem 1 (0.4) 0
Ertapenem 1 (0.4) 2 (0.5)
Vancomycin 19 (6.8) 7 (1.6)†
Ceftriaxone 3 (1.1) 0
Ciprofloxacin 9 (3.2) 5 (1.1)
* Cumulative reported allergen and reaction type exceeds cohort size as some patients received
multiple antibiotics
† Statistical significance (p < 0.05) on univariable analysis
Page 103
90
Figure 11-1: Statistical process control chart (P-chart) for outcome measure: Proportion of
patients with reported beta-lactam allergy receiving cefazolin as pre-operative prophylaxis
CL 0.4265
UCL
0.6323
LCL
0.2208
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
May June Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May June July Aug
Pro
po
rtio
n o
f p
atie
nts
wit
h s
elf-
rep
ort
ed b
eta
lact
am
alle
rgy
rece
ivin
g ce
fazo
lin in
ele
ctiv
e su
rger
ies
Month
Algorithm presented toAnesthesia business meeting
Implementation of Algorithm and Stamp
Algorithm Re-design
Algorithm posted in Operating Room
Page 104
91
The shaded months (December 2018 and January 2019) correspond to the Implementation period
which was censored from the regression analysis.
Figure 11-2: Segmented regression analysis on the primary outcome measure (monthly
cefazolin use)
Page 105
92
8
54
32
36.4%
59.1%
77.3%
90.9%
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
100.0%
0
5
10
15
20
Concern aboutcross-reactivity
Appropriatereason (based on
algorithm)
Other beta-lactam given
Antibioticalready running
Patient refused
Freq
uen
cy
Reason
Figure 11-3 Pareto chart: Reasons for avoiding cefazolin when alternate antibiotic
administered during intervention period (n=22)
Page 106
93
Error bars represent one standard deviation for each monthly calculated proportion
Figure 11-4: Pre-operative cefazolin use during intervention period stratified by algorithm
use in pre-operative clinic
Page 107
94
Figure 11-5: Cefazolin use in the baseline and intervention periods, stratified by
description of allergic reaction
Page 108
95
Table 11-4 Surgical site infection outcomes during study period
IPAC: Infection Prevention and Control, NSQIP: National Surgical Quality Improvement
Program
Surveillance
Method
July – Dec
2018
Jan – June
2019
Risk
difference
(%)
95%
Confidence
Interval
p-value
NSQIP-sampled
procedures
40 out of 747
(5.4%)
37 out of 665
(5.6%)
+0.2 -2.2, +2.6 0.86
IPAC: All cardiac
bypass and valve
surgeries
8 out of 186
(4.3%)
6 out of 223
(2.7%)
-1.6 -5.2, +2.0 0.37
Page 109
96
Figure 11-6: Statistical process control chart (P-chart) for process measure: Proportion of
patients with reported beta-lactam allergy with stamp imprinted on anesthetic record (All
patients assessed in pre-operative clinic)
CL 0.4761
UCL0.6819
LCL 0.2703
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Feb Mar Apr May June July Aug
Pro
po
rtio
n o
f p
atie
nts
wit
h r
ep
ort
ed
be
ta-l
acta
m a
llerg
y w
ith
st
amp
imp
rin
ted
in a
ne
sth
eti
c re
cord
Month
Page 110
97
CL 0.6593
UCL0.8963
LCL0.4223
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Feb Mar Apr May June July Aug
Pro
po
rtio
n o
f p
atie
nts
wit
h r
ep
ort
ed b
eta
-lac
tam
alle
rgy
wit
h
stam
p im
pri
nte
d in
an
esth
etic
reco
rd
Month
Figure 11-7: Statistical process control chart (P-chart) for process measure: Proportion of
patients with reported beta-lactam allergy with stamp imprinted on anesthetic record
(Patients assessed by anesthesiologist of pharmacist only)
Page 111
98
Figure 11-8: Statistical process control chart (T-chart) for balancing measure: Time
between peri-operative allergic reactions during study period
3038.041
144.0020
500
1000
1500
2000
2500
3000
3500
01-May-18 19-Mar-19 09-May-19
Tim
e B
etw
een
Eve
nts
(D
ays)
Date
Page 112
99
Chapter 12 Discussion
Discussion
12.1 Summary of Results
Standardization of the allergy assessment and pre-operative antibiotic selection in
patients with a reported beta-lactam allergy undergoing elective surgery resulted in a marked and
sustained increase in the use of pre-operative cefazolin. This increase was associated with a
reduced proportion of SSIs in cardiac surgeries during the Intervention period but the limited
time horizon and sample size resulted in this study being underpowered to detect a statistically
significant difference.
The intervention was safe, with only two peri-operative allergic events occurring during
the Intervention Period. In both cases, a blue dye was administered as part of the surgical
procedure. Use of blue dyes are associated with a high risk of anaphylactic reactions. A recent
national prospective study of peri-operative reactions occurring within the National Health
System hospitals in the United Kingdom found that blue dye was the fourth most common cause
of peri-operative anaphylaxis with an estimated incidence rate of 1 in 6863 administrations. It is
therefore possible that these peri-operative allergic reactions were secondary to blue dye rather
than cefazolin. It is also well established that patients with a history of allergy to one drug have
an increased risk of developing an allergy to a second structurally unrelated compound (19).
Therefore, development of an allergic reaction to cefazolin does not necessarily imply an allergic
cross-reaction.
Page 113
100
12.2 Comparison with Previous Interventions
In comparison to published studies describing interventions aimed at improving pre-
operative cefazolin use, our intervention utilized a standardized algorithm without the need for
additional diagnostic testing such as skin testing or direct drug provocation. This was achieved
by designing an algorithm which incorporated current evidence demonstrating the lack of
allergic cross-reactivity between cefazolin and other beta-lactam antibiotics. Doing so resulted in
a significantly simplified algorithm which primarily focused on whether patients reported a
severe delayed hypersensitivity reaction to any beta-lactam antibiotic or a Type I (IgE-mediated)
hypersensitivity reaction specifically to cefazolin, both of which were extremely uncommon.
Only 15 patients (1.9%) during the study period met any of the two criteria. Our algorithm also
recommended the use of cefazolin in patients with an unknown reaction history. Vaisman et al.
also utilized this strategy in their published intervention of 485 patients undergoing a
standardized allergy assessment by a pharmacist (41). Of these 485 patients, 23 (4.7%) patients
could not recall the reaction, and cefazolin was recommended in these patients without any
adverse reactions (41). It is our experience that unknown reactions typically reflect a remote,
non-severe allergic reaction. These features combined with the lack of cross-reactivity between
cefazolin and other beta-lactam antibiotics (in the event of a true allergy) forms the basis for
recommending cefazolin in patients with an unknown allergic reaction. These two unique
strategies resulted in an extremely simple algorithm that could be quickly used by healthcare
providers in the pre-operative setting. Avoiding the need for additional testing also allowed the
intervention to be quickly scaled to involve the entire pre-operative population. This is in
contrast to most previously published studies which were limited to a specific surgical cohort
(52,53,53,55,56) or relied on physician referral for testing (52,53,55–57,62).
Page 114
101
Secondly, our intervention did not require external physician assessment during the pre-
operative period. Previously published studies utilized an allergist and specialized nursing
support (52–57,62) or infectious diseases physician (41) external to the pre-operative clinic to
perform the intervention. In contrast, our intervention empowered anesthesiologists (with the
support of pharmacists) already working in the pre-operative clinic (and operating rooms) to use
the algorithm. The success of this strategy was dependent on the simplicity of the algorithm
which had minimal impact on the pre-existing workflow for pharmacists and anesthesiologists.
Paradoxically, users of the algorithm commented that the algorithm saved time, as the various
permutations of reported beta-lactam allergy and allergic reaction types could be quickly
addressed using the algorithm. Margallo et al. implemented a structured allergy assessment for
nurses working in the pre-operative clinic, but their intervention did not target prescribers or
provide a pre-operative antibiotic recommendation(58). It is therefore not surprising that their
intervention only modestly improved pre-operative beta-lactam antibiotic use.
Despite a significant increase in the use of cefazolin (from 42.7 to 85.6%), only 47.6% of
patients with reported beta-lactam allergy had a stamp imprinted on the anesthetic record. Based
on this, it is highly unlikely that the stamp was the primary driver of change and likely a
suboptimal process measure. This is supported by the observation that patients seen by nursing
only (non-users of the algorithm or stamp) in the pre-operative clinic had a parallel increase in
cefazolin use (Figure 11-3). Furthermore, in the 22 cases where an alternate pre-operative
antibiotic was administered and an anesthesiologist was interviewed, the absence of a stamp was
not cited as a reason.
The finding of improved cefazolin use despite suboptimal stamp utilization can be
attributed to two reasons. First, the simplicity of the algorithm allowed anesthesiologists to
Page 115
102
comfortably use the algorithm on the day of surgery instead of relying on the stamp imprint for
guidance. Second, because this intervention empowered the end-prescriber (anesthesiologists) to
use the algorithm rather than relying on a recommendation from an external consultant, the
barriers associated with provider-to-provider communication were significantly less. Even if a
stamp was missing from the anesthetic record, anesthesiologists were equipped with the
appropriate tool to carry out the allergy and pre-operative antibiotic assessment on the day of
surgery. This is the first intervention described to date to utilize the strategy of targeting the end-
prescriber instead of utilizing an external consultant. Prior studies have utilized various methods
to facilitate provider-to-provider communication but have invariably encountered problems with
uptake of recommendations (35,57).
12.3 Study Limitations
Due to the low frequency of SSIs and reported beta-lactam allergies at our institution, this
study was underpowered to detect a significant difference in SSI rates during the study period.
We also did not have the infrastructure to measure SSI outcomes in all patients with reported
beta-lactam allergy.
Without a control group, we could not completely exclude concurrent interventions
aimed at improving pre-operative antibiotic prophylaxis during the study period. However, we
did not encounter any existing quality improvement interventions during the diagnostic phase of
this project. Moreover, the timing of the change in outcome measure occurred at the same time
as the launch of our intervention which provides compelling evidence that the measured effect
was a result of our intervention.
In addition, this intervention was limited to patients undergoing elective surgery. It is
unclear whether the same effect can be seen in non-elective surgeries, but given the data
Page 116
103
demonstrating increased cefazolin use in patients who bypassed the algorithm in the pre-
operative clinic, we believe that this intervention could be safely scaled to emergency surgeries,
or to institutions who do not have a formal pre-operative clinic.
There was also a component of detection bias during the Intervention period, as there
were a greater number of patients with a reported beta-lactam allergy compared to the Baseline
period. We suspect that improved awareness of the impact of reported beta-lactam allergies led
health providers to document more diligently whenever an allergy was encountered. This was
also supported by the fact that there were fewer patients with a documented unknown reaction
during the Intervention period, likely as a result of improved allergy history taking. Despite this,
the overall distribution of allergic reactions was similar between the Baseline and Intervention
period, suggesting that improved documentation of reported beta-lactam allergy during the
intervention period did not favour more mild reactions. Moreover, the proportion of patients with
reported beta-lactam allergy receiving cefazolin during the Baseline period closely approximated
the proportion calculated during the 18-month retrospective cohort study (Table 5-2).
Considering the “worst case scenario” where an additional group of un-documented patients with
reported beta-lactam allergy (equaling the difference between the Baseline and Intervention
period) all received cefazolin during the Baseline period, the proportion of patients receiving
cefazolin would still be only 63.6%. Therefore, the finding of detection bias unlikely affected the
impact of this intervention.
Finally, while our intervention was successful at improving pre-operative use of cefazolin
specifically, the algorithm did not address whether a patient would be safe to receive other beta-
lactam antibiotics. The process of allergy de-labelling, and the determination of which beta-
Page 117
104
lactam antibiotics can be safely used in patients with a reported allergy is significantly more
complex and was felt to be beyond the scope of this quality improvement project.
Finally, this quality improvement study was conducted at a single institution with
engaged anesthesiologists where baseline use of perioperative cefazolin was higher compared to
other institutions(6,41). Whether or not this intervention would have the same uptake and impact
on perioperative cefazolin in a different clinical context is unknown and further evaluation
should assess how easily this tool can be scaled to other institutions.
12.4 Cost Considerations
The costs associated with implementation of this intervention were minimal. Costs
included the procurement of four stamps to be used in the pre-operative clinic, and printing and
lamination of 30 algorithms displayed in the pre-operative clinic and operating rooms.
Although this study was underpowered to detect a decrease in SSIs, use of this
intervention could potentially result in cost savings through the prevention of SSIs. Blumenthal
et al. estimated that between 112 and 124 patients would need to undergo an allergy assessment
to prevent one surgical site infection (6). Based on these estimates, approximately 4 surgical site
infections may have been prevented during the intervention period. Based on a study of Veterans
Affairs hospitals in the United States, a surgical site infection was associated a 1.43 times greater
use of healthcare costs compared to patients without infection (64).
Reductions in alternate antibiotic use represent an additional source of cost savings. For
example, the acquisition cost of vancomycin at our institution is $2.89 per 1 gram vial compared
to $2.54 per 1 gram vial of cefazolin. Considering the volume of surgical procedures which
Page 118
105
occur annually and the need for re-dosing of antibiotics during surgical procedures, cost savings
from improved antibiotic utilization can become significant.
Finally, potential indirect savings include improved efficiency in the pre-operative clinic
and operating room due to a simplified allergy assessment algorithm. Due to the risk of
histamine-related infusion reactions with rapid administration of vancomycin (an alternate
antibiotic to cefazolin), vancomycin is recommended to be administered 60-120 minutes before
surgical incision (1). Use of vancomycin therefore may result in delays in surgical incision time
and operating room efficiency. Moussa et al. at McGill University demonstrated that in surgical
procedures where vancomycin was used, there was on average a 22-minute delay in surgery
initiation (35) compared to procedures which used cefazolin. Therefore, reducing the use of
vancomycin as an alternate antibiotic may also result in improved efficiency of operating room
function.
12.5 Sustainability and Spread
The effect of increased cefazolin use as pre-operative prophylaxis remained sustained
over a 9-month period. As this intervention required no significant resource requirements, we are
hopeful that this practice change will be preserved. There are plans to transition to an electronic
anesthetic record documentation system in the upcoming year, and we plan to implement this
algorithm into their system to hardwire the decision-making process. Furthermore, existing pre-
printed order sets which contain pre-operative antibiotic orders, have been updated to include
guidance based on the content of the algorithm.
There has been significant interest in adapting this algorithm for use in other surgical
programs within and outside our institution. As part of the Centre for Quality Improvement and
Patient Safety Certificate Course Program, an anesthesiology fellow implemented this algorithm
Page 119
106
for women undergoing elective Caesarian sections at SHSC. These patients are not assessed in
the pre-operative clinic and may have other indications for antibiotic prophylaxis. The
intervention was well received, with a similar increase in the proportion of patients with reported
beta-lactam allergy receiving cefazolin (Figure 12-1). This practice change was facilitated by the
fact that anesthesiologists who provide care for the obstetric population are the same physicians
providing peri-operative care in the main operating rooms at our institution.
Outside of our institution, Michael Garron Hospital has modified their standardized
allergy history assessment to align with our algorithm and have demonstrated similar results and
safety. University Health Network is also in the process of implementing an adapted version of
the algorithm into their pre-operative assessment program. We have also been contacted by
surgeons at Mount Sinai Hospital and St. Michael’s Hospital who are interested in implementing
this algorithm into their pre-operative process.
12.6 Future Directions
As more patients with reported beta-lactam allergy undergo elective surgery receive
cefazolin based on our algorithm, we hope to accumulate a large enough sample size to study
whether this intervention results in a reduction in surgical site infection. A 10-year follow-up
study would be required to have adequate statistical power to assess the impact of increased
cefazolin use on SSI rates at our institution. Since this would unlikely be feasible, a multicenter
study of this intervention would be the next step in evaluation.
The demonstration of safety with this intervention also confirms the theory of non-cross
reactivity between cefazolin and penicillin. Doing so paves the way for other quality
improvement interventions aimed at improving cefazolin use in the treatment and prevention of
other infections where this agent is preferred. Examples include the use of cefazolin in
Page 120
107
Staphylococcus aureus bacteremia, uncomplicated skin and soft tissue infections and Group B
streptococcus prophylaxis in pregnant women.
Finally, as more data accumulates regarding the safety of this intervention, we hope that
more anesthesiologists will change their prescribing practice to align with the algorithm. As
illustrated in Figure 11-3, a small number of anesthesiologists remained concerned for allergic
cross-reactivity during the study period. We hope to reinforce the safety of this intervention by
presenting at a future Quality Improvement Rounds to share our successes and address any
remaining concerns. Ultimately, the results of our study will add to the literature on the safety of
using cefazolin as perioperative prophylaxis for patients reporting a beta-lactam allergy and will
contribute to future surgical guideline updates(1).
12.7 Conclusions
Implementation of an anesthesiologist-led standardized algorithm for allergy assessment
and pre-operative antibiotic selection in patients with reported beta-lactam allergy undergoing
elective surgery resulted in a pronounced and sustained increase in the use of cefazolin. Due to
its simplicity and minimal resource requirements, this intervention has potential to be scalable to
other institutions and surgical settings with potential for significant impact on reducing surgical
site infection rates across a broad number of surgeries.
Page 121
108
Implementation of Algorithm
CL 0.3955
UCL
1.0000
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Jan
-17
Feb
-17
Mar
-17
Ap
r-17
May
-17
Jun
-17
Jul-
17
Au
g-17
Sep
-17
Oct
-17
No
v-1
7
Dec
-17
Jan
-18
Feb
-18
Mar
-18
Ap
r-18
May
-18
Jun
-18
Jul-
18
Au
g-18
Sep
-18
Oct
-18
No
v-1
8
Dec
-18
Jan
-19
Feb
-19
Mar
-19
Ap
r-19
Pro
po
rtio
n o
f P
atie
nts
wit
h R
epo
rted
Be
ta-L
acta
m A
llerg
y u
nd
ergo
ing
Elec
tive
Cae
sare
an S
ecti
on
re
ceiv
ing
Ce
fazo
lin
Month
Figure 12-1: Statistical process control chart (P-chart): Proportion of patients with
reported beta-lactam allergy undergoing elective Caesarean section receiving cefazolin as
pre-operative prophylaxis
Page 122
109
Glossary of Terms
Allergy A hypersensitivity of the immune system to specific
substance
Allergy skin testing A test used in the diagnosis of allergies which involves
injection of the substance under the skin to provoke a
small, controlled allergic response.
American Society of
Anesthesiologists (ASA)
Physical Classification Status
A classification system used by physicians to
categorize a patient's fitness for surgery and predicts
their risk of surgical complications. The score ranges
from 1 (most fit) to 5 (least fit).
Anesthesiologist A medical specialist who provides peri-operative care
to patients before, during and after surgery.
Anesthetic record A medical document that catalogues the vital signs,
procedures undertaken and medications administered
during the course of a surgical procedure.
Beta-Lactam allergy An allergy to one (or more) antibiotics which contain a
beta-lactam molecular ring structure. Examples
include penicillins, cephalosporins and carbapenems.
Direct drug provocation A test used in the diagnosis of allergies which involves
giving a small dose of a drug to a patient to assess for
an allergic response.
Elective surgery Surgery that is scheduled in advance and does not
involve a medical emergency.
Page 123
110
Mediation analysis A method used to understand the mechanism by which
one variable affects another variable through the
inclusion of an intermediary (mediator) variable.
National Surgical Quality
Improvement Program
(NSQIP)
A nationally benchmarked program that hospitals can
voluntarily participate in to measure, monitor and
improve their surgical quality of care.
Pre-operative prophylaxis An antibiotic given to a patient before surgery to
decrease the risk of developing a surgical site
infection.
Severe delayed
hypersensitivity reaction
A potentially life-threatening subset of T-cell (Type
IV) hypersensitivity reactions that can cause severe
organ damage. Examples include Steven-Johnson
Syndrome, Toxic Epidermal Necrolysis, and Drug
Rash with Eosinophilia and Systemic Symptoms
Type I (IgE-mediated)
Hypersensitivity Reaction
A type of allergic reaction whereby an antibody
(Immunoglobulin E) directed against the allergen
triggers mast cells to release granules of histamine and
other compounds resulting in the development of
urticaria (hives), angioedema (swelling of the face and
lips), bronchospasm (closing of the airway) and
potentially anaphylaxis.
Surgical site infection An infection that takes place after surgery in the part
of the body where the surgery took place.
Wound classification A four-category variable ranging from clean (I) to
dirty (IV) used to identify patients at risk of surgical
site infection
Page 124
111
References
1. Bratzler DW, Dellinger EP, Olsen KM, Perl TM, Auwaerter PG, Bolon MK, et al. Clinical
practice guidelines for antimicrobial prophylaxis in surgery. Am J Health Syst Pharm. 2013
Feb 1;70(3):195–283.
2. Zagursky RJ, Pichichero ME. Cross-reactivity in β-Lactam Allergy. The Journal of Allergy
and Clinical Immunology: In Practice. 2018 Jan 1;6(1):72-81.e1.
3. MacFadden DR, LaDelfa A, Leen J, Gold WL, Daneman N, Weber E, et al. Impact of
Reported Beta-Lactam Allergy on Inpatient Outcomes: A Multicenter Prospective Cohort
Study. Clin Infect Dis. 2016 Oct 1;63(7):904–10.
4. Macy E, Contreras R. Health care use and serious infection prevalence associated with
penicillin “allergy” in hospitalized patients: A cohort study. Journal of Allergy and Clinical
Immunology. 2014 Mar 1;133(3):790–6.
5. Huang K-HG, Cluzet V, Hamilton K, Fadugba O. The Impact of Reported Beta-Lactam
Allergy in Hospitalized Patients With Hematologic Malignancies Requiring Antibiotics.
Clin Infect Dis. 2018 Jun 18;67(1):27–33.
6. Blumenthal KG, Ryan EE, Li Y, Lee H, Kuhlen JL, Shenoy ES. The Impact of a Reported
Penicillin Allergy on Surgical Site Infection Risk. Clin Infect Dis. 2018 18;66(3):329–36.
7. Turner NA, Moehring R, Sarubbi C, Wrenn RH, Drew RH, Cunningham CK, et al.
Influence of Reported Penicillin Allergy on Mortality in MSSA Bacteremia. Open Forum
Infect Dis [Internet]. 2018 Feb 19 [cited 2019 Aug 23];5(3). Available from:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5861429/
8. Kong K-F, Schneper L, Mathee K. Beta-lactam Antibiotics: From Antibiosis to Resistance
and Bacteriology. APMIS. 2010 Jan;118(1):1–36.
9. Shenoy ES, Macy E, Rowe T, Blumenthal KG. Evaluation and Management of Penicillin
Allergy: A Review. JAMA. 2019 Jan 15;321(2):188–99.
10. Bigby M, Jick S, Jick H, Arndt K. Drug-induced cutaneous reactions. A report from the
Boston Collaborative Drug Surveillance Program on 15,438 consecutive inpatients, 1975 to
1982. JAMA. 1986 Dec 26;256(24):3358–63.
11. Shah NS, Ridgway JP, Pettit N, Fahrenbach J, Robicsek A. Documenting Penicillin
Allergy: The Impact of Inconsistency. Lai H-C, editor. PLoS ONE. 2016 Mar
16;11(3):e0150514.
12. Patel BM. Skin rash with infectious mononucleosis and ampicillin. Pediatrics. 1967
Nov;40(5):910–1.
Page 125
112
13. Blanca M, Torres MJ, García JJ, Romano A, Mayorga C, de Ramon E, et al. Natural
evolution of skin test sensitivity in patients allergic to beta-lactam antibiotics. J Allergy
Clin Immunol. 1999 May;103(5 Pt 1):918–24.
14. Wanat M, Anthierens S, Butler CC, Wright JM, Dracup N, Pavitt SH, et al. Patient and
Prescriber Views of Penicillin Allergy Testing and Subsequent Antibiotic Use: A Rapid
Review. Antibiotics (Basel). 2018 Aug 6;7(3).
15. Macy E, Poon K-Y T. Self-reported antibiotic allergy incidence and prevalence: age and
sex effects. Am J Med. 2009 Aug;122(8):778.e1-7.
16. Burrows JA, Nissen LM, Kirkpatrick CMJ, Bell SC. Beta-lactam allergy in adults with
cystic fibrosis. J Cyst Fibros. 2007 Jul;6(4):297–303.
17. Pleasants RA, Walker TR, Samuelson WM. Allergic reactions to parenteral beta-lactam
antibiotics in patients with cystic fibrosis. Chest. 1994 Oct;106(4):1124–8.
18. Imlay H, Krantz EM, Stohs EJ, Lan KF, Zier J, Kim HN, et al. Reported Beta Lactam and
Other Antibiotic Allergies in Solid Organ and Hematopoietic Cell Transplant Recipients.
Clin Infect Dis [Internet]. [cited 2019 Oct 18]; Available from:
https://academic.oup.com/cid/advance-article/doi/10.1093/cid/ciz1025/5588939
19. Strom BL, Schinnar R, Apter AJ, Margolis DJ, Lautenbach E, Hennessy S, et al. Absence
of Cross-Reactivity between Sulfonamide Antibiotics and Sulfonamide Nonantibiotics.
New England Journal of Medicine. 2003 Oct 23;349(17):1628–35.
20. Shojania KG, Duncan BW, McDonald KM, Wachter RM, Markowitz AJ. Making health
care safer: a critical analysis of patient safety practices. Evid Rep Technol Assess (Summ).
2001;(43):i–x, 1–668.
21. Kirkland KB, Briggs JP, Trivette SL, Wilkinson WE, Sexton DJ. The impact of surgical-
site infections in the 1990s: attributable mortality, excess length of hospitalization, and
extra costs. Infect Control Hosp Epidemiol. 1999 Nov;20(11):725–30.
22. Zimlichman E, Henderson D, Tamir O, Franz C, Song P, Yamin CK, et al. Health care-
associated infections: a meta-analysis of costs and financial impact on the US health care
system. JAMA Intern Med. 2013 Dec 9;173(22):2039–46.
23. Centers for Disease Control (CDC)/National Healthcare Safety Network (NHSN). Protocol
for definition of surgical site infection. [Internet]. [cited 2019 Aug 23]. Available from:
https://www.cdc.gov/nhsn/pdfs/pscmanual/9pscssicurrent.pdf.
24. Owens CD, Stoessel K. Surgical site infections: epidemiology, microbiology and
prevention. J Hosp Infect. 2008 Nov;70 Suppl 2:3–10.
25. Fore J. A review of skin and the effects of aging on skin structure and function. Ostomy
Wound Manage. 2006 Sep;52(9):24–35; quiz 36–7.
Page 126
113
26. Kaye KS, Schmit K, Pieper C, Sloane R, Caughlan KF, Sexton DJ, et al. The effect of
increasing age on the risk of surgical site infection. J Infect Dis. 2005 Apr 1;191(7):1056–
62.
27. Sørensen LT. Wound healing and infection in surgery: the pathophysiological impact of
smoking, smoking cessation, and nicotine replacement therapy: a systematic review. Ann
Surg. 2012 Jun;255(6):1069–79.
28. Martin ET, Kaye KS, Knott C, Nguyen H, Santarossa M, Evans R, et al. Diabetes and Risk
of Surgical Site Infection: A systematic review and meta-analysis. Infect Control Hosp
Epidemiol. 2016 Jan;37(1):88–99.
29. Haubner F, Ohmann E, Pohl F, Strutz J, Gassner HG. Wound healing after radiation
therapy: review of the literature. Radiat Oncol. 2012 Sep 24;7:162.
30. Bootun R. Effects of immunosuppressive therapy on wound healing. Int Wound J. 2013
Feb;10(1):98–104.
31. Culver DH, Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG, et al. Surgical
wound infection rates by wound class, operative procedure, and patient risk index. National
Nosocomial Infections Surveillance System. Am J Med. 1991 Sep 16;91(3B):152S-157S.
32. National Nosocomial Infections Surveillance System. National Nosocomial Infections
Surveillance (NNIS) System Report, data summary from January 1992 through June 2004,
issued October 2004. Am J Infect Control. 2004 Dec;32(8):470–85.
33. Hawn MT, Richman JS, Vick CC, Deierhoi RJ, Graham LA, Henderson WG, et al. Timing
of surgical antibiotic prophylaxis and the risk of surgical site infection. JAMA Surg. 2013
Jul;148(7):649–57.
34. Vorobeichik L, Weber EA, Tarshis J. Misconceptions Surrounding Penicillin Allergy:
Implications for Anesthesiologists. Anesth Analg. 2018;127(3):642–9.
35. Moussa Y, Shuster J, Matte G, Sullivan A, Goldstein RH, Cunningham D, et al. De-
labeling of β-lactam allergy reduces intraoperative time and optimizes choice in antibiotic
prophylaxis. Surgery. 2018 May 8;
36. Picard M, Bégin P, Bouchard H, Cloutier J, Lacombe-Barrios J, Paradis J, et al. Treatment
of patients with a history of penicillin allergy in a large tertiary-care academic hospital. J
Allergy Clin Immunol Pract. 2013 Jun;1(3):252–7.
37. American College of Surgeons, User Guide for the 2017 ACS NSQIP Participant Use Data
File (PUF) [Internet]. [cited 2019 Aug 23]. Available from: https://www.facs.org/-
/media/files/quality-programs/nsqip/nsqip_puf_userguide_2017.ashx?la=en
38. Elligsen M, Walker SAN, Simor A, Daneman N. Prospective Audit and Feedback of
Antimicrobial Stewardship in Critical Care: Program Implementation, Experience, and
Challenges. Can J Hosp Pharm. 2012;65(1):31–6.
Page 127
114
39. Steen J, Loeys T, Moerkerke B, Vansteelandt S. medflex: An R Package for Flexible
Mediation Analysis using Natural Effect Models. Journal of Statistical Software. 2017 Feb
27;76(1):1–46.
40. Leis JA, Palmay L, Ho G, Raybardhan S, Gill S, Kan T, et al. Point-of-Care β-Lactam
Allergy Skin Testing by Antimicrobial Stewardship Programs: A Pragmatic Multicenter
Prospective Evaluation. Clin Infect Dis. 2017 01;65(7):1059–65.
41. Vaisman A, McCready J, Hicks S, Powis J. Optimizing preoperative prophylaxis in patients
with reported β-lactam allergy: a novel extension of antimicrobial stewardship. J
Antimicrob Chemother. 2017 01;72(9):2657–60.
42. Haslam S, Yen D, Dvirnik N, Engen D. Cefazolin Use in Patients who Report a Non-IgE
Mediated Penicillin Allergy: A Retrospective Look at Adverse Reactions in Arthroplasty.
Iowa Orthop J. 2012;32:100–3.
43. Hoang H, Remtulla S, Zurek K, Degenhardt O. Covenant Health Antimicrobial
Stewardship E-Newsletter: Update on Penicillin Allergies [Internet]. 2018 [cited 2019 Sep
2]. Available from:
http://extcontent.covenanthealth.ca/20180116_education_CHASEIssue17Penallergies.pdf
44. American Academy of Allergy, Asthma & Immunology | Choosing Wisely [Internet]. [cited
2019 Sep 2]. Available from: http://www.choosingwisely.org/societies/american-academy-
of-allergy-asthma-immunology/
45. Leis JA, Evans GA, Ciccotelli W, Garber G, Gregson D, Lee TC, et al. Choosing Wisely
Canada – top five list in infectious diseases: An official position statement of the
Association of Medical Microbiology and Infectious Disease (AMMI) Canada. Official
Journal of the Association of Medical Microbiology and Infectious Disease Canada
[Internet]. 2017 Sep 5 [cited 2019 Sep 2]; Available from:
https://jammi.utpjournals.press/doi/abs/10.3138/jammi.1.1.02
46. Savic LC, Khan DA, Kopac P, Clarke RC, Cooke PJ, Dewachter P, et al. Management of a
surgical patient with a label of penicillin allergy: narrative review and consensus
recommendations. Br J Anaesth. 2019 Jul;123(1):e82–94.
47. Petz LD. Immunologic cross-reactivity between penicillins and cephalosporins: a review. J
Infect Dis. 1978 May;137 Suppl:S74–9.
48. Dash CH. Penicillin allergy and the cephalosporins. J Antimicrob Chemother. 1975;1(3
Suppl):107–18.
49. Pichichero ME. A review of evidence supporting the American Academy of Pediatrics
recommendation for prescribing cephalosporin antibiotics for penicillin-allergic patients.
Pediatrics. 2005 Apr;115(4):1048–57.
50. Staicu ML, Soni D, Conn KM, Ramsey A. A survey of inpatient practitioner knowledge of
penicillin allergy at 2 community teaching hospitals. Ann Allergy Asthma Immunol.
2017;119(1):42–7.
Page 128
115
51. Puchner TC, Zacharisen MC. A survey of antibiotic prescribing and knowledge of
penicillin allergy. Ann Allergy Asthma Immunol. 2002 Jan;88(1):24–9.
52. Li JT, Markus PJ, Osmon DR, Estes L, Gosselin VA, Hanssen AD. Reduction of
vancomycin use in orthopedic patients with a history of antibiotic allergy. Mayo Clin Proc.
2000 Sep;75(9):902–6.
53. Park M, Markus P, Matesic D, Li JTC. Safety and effectiveness of a preoperative allergy
clinic in decreasing vancomycin use in patients with a history of penicillin allergy. Ann
Allergy Asthma Immunol. 2006 Nov;97(5):681–7.
54. Frigas E, Park MA, Narr BJ, Volcheck GW, Danielson DR, Markus PJ, et al. Preoperative
evaluation of patients with history of allergy to penicillin: comparison of 2 models of
practice. Mayo Clin Proc. 2008 Jun;83(6):651–62.
55. Cook DJ, Barbara DW, Singh KE, Dearani JA. Penicillin skin testing in cardiac surgery. J
Thorac Cardiovasc Surg. 2014 Jun;147(6):1931–5.
56. McDanel DL, Azar AE, Dowden AM, Murray-Bainer S, Noiseux NO, Willenborg M, et al.
Screening for Beta-Lactam Allergy in Joint Arthroplasty Patients to Improve Surgical
Prophylaxis Practice. J Arthroplasty. 2017;32(9S):S101–8.
57. Savic L, Gurr L, Kaura V, Toolan J, Sandoe J a. T, Hopkins PM, et al. Penicillin allergy de-
labelling ahead of elective surgery: feasibility and barriers. British Journal of Anaesthesia.
2019 Jul 1;123(1):e110–6.
58. Margallo J-EP, Smith EA, Marks G, Ben-Aderet M, Yang H, Madhusudhan M, et al.
Optimizing utilization of beta-lactam surgical prophylaxis through implementation of a
structured allergy assessment tool in a presurgical clinic. Infect Control Hosp Epidemiol.
2019 Sep 30;1–3.
59. Fox S, Park MA. Penicillin skin testing in the evaluation and management of penicillin
allergy. Ann Allergy Asthma Immunol. 2011 Jan;106(1):1–7.
60. Kuruvilla M, Thomas J. Direct oral amoxicillin challenge without antecedent penicillin skin
testing in low-risk patients. Ann Allergy Asthma Immunol. 2018;121(5):627–8.
61. Kuruvilla M, Shih J, Patel K, Scanlon N. Direct oral amoxicillin challenge without
preliminary skin testing in adult patients with allergy and at low risk with reported
penicillin allergy. Allergy Asthma Proc. 2019 01;40(1):57–61.
62. Moussa Y, Sullivan A, Matte G, Goldstein RH, Baldini G, Shuster J, et al. Impact of
persistent β-lactam allergy documentation despite delabeling in the perioperative setting. J
Allergy Clin Immunol Pract. 2019 Jul 8;
63. Lam P. Improving Cefazolin Use as Surgical Prophylaxis in Penicillin-Allergic Patients
[Internet]. Sunnybrook Health Sciences Centre; 2019 [cited 2019 Sep 24]. (Anesthesia
Quality Improvement Rounds). Available from:
https://www.youtube.com/watch?v=0Ga4UU_uLos
Page 129
116
64. Schweizer ML, Cullen JJ, Perencevich EN, Vaughan Sarrazin MS. Costs Associated With
Surgical Site Infections in Veterans Affairs Hospitals. JAMA Surg. 2014 Jun;149(6):575–
81.
Page 130
117
Appendices
Appendix 1: Sampling methodology and surgical procedures selected for NSQIP
• Sampling occurs on an 8-day cycle, with 60 cases sampled per cycle:
• The first 15 cases (sorted operative logs in chronologic order based on surgery date,
patient in room time, and operative room number) performed by General surgery or
Vascular surgery are included in the sample, then:
• Targeted surgical procedures denoted as “all cases” are included in the sample, then:
• The remaining targeted surgical procedures in the list are included in the sample (sorted
operative logs in chronologic order based on surgery date, patient in room time, and
operative room number) until 60 cases are reached
1. General Surgery
a Pancreatectomy (all cases)
b Proctectomy (all cases)
c Hepatectomy
d Colectomy
2. Vascular
a Carotid Endarterectomy (all cases)
b Abdominal Aortic Aneurysm Repair (AAA) (all cases)
c Aortoiliac Bypass Open (all cases)
d Lower Extremity Bypass –Open (all cases)
e Endovascular AAA (EVAR)
3. Neurosurgery
a Brain Tumor Procedure
a Spine Procedure (limited to 5 cases/cycle in conjunction with Orthopedic Surgery)
Page 131
118
4. Gynecology – Oncology Surgery
a Hysterectomy / Myomectomy
5. Urology
a Bladder Suspension (all cases)
b TURP
c Prostatectomy
d Nephrectomy
e Cystectomy
6. Orthopedic Surgery
a Hip Fracture
b Spine procedure (limited to 5 cases/cycle in conjunction with Neurosurgery)
7. Plastic Surgery
a Breast Reconstruction (all cases)
Page 132
119
Appendix 2: Definitions used for covariates included in multivariable logistic regression
model
Age (continuous variable): The time elapsed (in years) between the patient’s date of birth and
date of surgical procedure.
Gender (categorical variable): The patient’s sex as documented on the medical record.
Current Smoker (binary variable): Patient has smoked cigarettes within 12 months prior to
admission for the primary procedure.
Diabetes (binary variable): A diagnosis of diabetes mellitus requiring therapy with insulin or an
anti-diabetic agent.
Steroid or Immunosuppressant Use (binary variable): Patient has required the regular
administration of oral or parenteral corticosteroid medications or immunosuppressant
medications, for a chronic medical condition, within the 30 days prior to the primary procedure.
A one-time pulse limited short course, or a taper of less than 10 days duration would not qualify.
Long-interval injections of long-acting agents (e.g. monthly) that are part of an ongoing regimen
would qualify. Chemotherapy alone does not meet this criterion.
ASA Physical Status Classification (categorical variable): The patient’s physical condition as it
appears on the anesthetic record. The classification is as follows:
• ASA 1 – Normal healthy patient
• ASA 2 – Patient with mild systemic disease
• ASA 3 – Patient with severe systemic disease
• ASA 4 – Patient with severe systemic disease that is a constant threat to life
Page 133
120
• ASA 5 – Moribund patient who is not expected to survive without the operation
Wound Classification (categorical variable): The level of contamination of the surgical wound
as recorded on the operative record during the time of the procedure. This classification is based
on the NHSN definitions(23), and are as follows:
• Clean: An uninfected operative wound in which no inflammation is encountered and the
respiratory, alimentary, genital, or uninfected urinary tract is not entered.
• Clean/Contaminated: An operative wound in which the respiratory, alimentary, genital
or urinary tracts are entered under controlled conditions and without unusual
contamination.
• Contaminated: Open, fresh, accidental wounds. In addition, operations with major breaks
in sterile technique or gross spillage from the gastrointestinal tract, and incisions in which
acute, non-purulent inflammation is encountered including necrotic tissue without
evidence of purulent drainage (e.g., dry gangrene) are included in this category.
• Dirty/Infected: Old traumatic wounds with retained devitalized tissue and those that
involve existing clinical infection or perforated viscera. This definition suggests that the
organisms causing postoperative infection were present in the operative field before the
operation.
Page 134
121
Appendix 3: Literature review search strategy
Two databases (Ovid and Embase) were used to complete the literature review. In addition,
papers of interest were reviewed thoroughly to identify additional studies through their citations.
Ovid Search
1. pre?operative.mp. → 260260 results
2. surgical prophylaxis.mp. or exp Antibiotic Prophylaxis/ → 13304 results
3. (beta?lactam allerg* or penicillin allerg*).mp. [mp=title, abstract, original title, name of
substance word, subject heading word, floating sub-heading word, keyword heading
word, protocol supplementary concept word, rare disease supplementary concept word,
unique identifier, synonyms] → 1299 results
4. (1 and 3) or (2 and 3) → 82 results
Embase Search
1. pre?operative.mp. or exp preoperative evaluation/ → 413857 results
2. exp antibiotic prophylaxis/ or surgical prophylaxis.mp. → 29845 results
3. (beta?lactam allerg* or penicillin allerg*).mp. [mp=title, abstract, heading word, drug
trade name, original title, device manufacturer, drug manufacturer, device trade name,
keyword, floating subheading word, candidate term word]
4. (1 and 3) or (2 and 3) → 292 results
Page 135
122
Appendix 4: Anesthesiologist survey and responses
Question #1: In a c-section patient with a history of penicillin allergy, with what type of
reaction(s) would you be comfortable with giving cefazolin for antimicrobial prophylaxis?
Check all that apply.
Answered: 17 Skipped: 0
Answer Choices Responses %
Rash 13 76.5
Childhood reaction 13 76.5
Anaphylaxis 0 0
GI upset (nausea, diarrhea, etc) 16 94.1
Urticaria 7 41.2
Pruritus 13 76.5
Delayed reaction (> 1 hour) 3 17.7
Bronchospasm 1 5.9
Angioedema 0 0
Laryngeal edema 0 0
Unknown reaction (patient cannot recall) 12 70.6
Other 1 5.9
Page 136
123
Question #2: Your patient has a penicillin allergy. What additional questions on history would
you ask about? (check all that apply)
Answered: 17 Skipped: 0
Always Usually Sometimes Rarely Never Total
Type of reaction 85.71% 7.14% 0.00% 7.14% 0.00%
12 1 0 1 0 14
When last reaction occurred 56.25% 31.25% 12.50% 0.00% 0.00%
9 5 2 0 0 16
Any allergy testing done? 21.43% 35.71% 21.43% 21.43% 0.00%
3 5 3 3 0 14
Was patient on other
medications when allergic
reaction occurred?
7.14% 14.29% 35.71% 42.86% 0.00%
1 2 5 6 0 14
Never 0.00% 0.00% 0.00% 100.00% 0.00%
0 0 0 1 0 1
Question #3: What would help you to increase the use of cefazolin in patients with penicillin
allergy? Check all that apply.
Answered: 15 Skipped: 2
Answer Choices Response %
Arrange allergy testing prior to C-section 6 40
Book patient in pre-op clinic to discuss detailed allergy history 3 20
More time allocated for pre-op assessment on day of C-section 3 20
Algorithm or checklist tool (on anesthesia cart) to help with
decision-making about whether or not it is safe to give cefazolin
based on type of reaction
12 80
Attend rounds by an allergist or infectious disease physician about
this topic 8 53.3
Other 2 13.3