Addressing Inpatient Diabetes: A QI Perspective Special Focus on Hypoglycemia Reduction Greg Maynard M.D., CQO at UC Davis (Former Director, UCSD CIIS.

Addressing Inpatient Diabetes: A QI PerspectiveSpecial Focus on Hypoglycemia Reduction

Greg Maynard M.D.,

CQO at UC Davis

(Former Director, UCSD CIIS and SHM CMO)

Kristi Kulasa M.D

Director UCSD Inpatient Glycemic Control

Safe Glycemic Control Team Leads

Greg Maynard MD, MScHospital Medicine

Kristen Kulasa MDEndocrinology, Inpatient DM lead

Conflict of Interest Statement

– AHRQ grants to improve glycemic control and reduce hypoglycemia

– Mentors in SHM Glycemic Control programs

Objectives- The participant in this activity will:

1. Appreciate the epidemiology of inpatient diabetes, hyperglycemia, and hypoglycemic adverse drug events

2. Understand the key ingredients for successful inpatient glycemic control / insulin management programs, including high quality ‘glucometrics’, inter-professional teams, and tips on design and implementation of order sets / protocols

3. Be able cite the most common preventable sources of iatrogenic hypoglycemia and how to reduce them with a hypoglycemia reduction bundle

Why Glycemic Control?(It’s about more than infusion insulin glycemic targets!)

• DM / Hyperglycemia Very Common • Association with poor outcomes and glycemic excursions• Opportunity to identify and intervene

– poorly controlled DM, previously undiagnosed DM, stress hyperglycemia (pre-diabetes)

• Hypoglycemia and extreme hyperglycemia– Safety problem and a Quality problem

• Public reporting, regulatory guidelines etc.• Inpatient Care - Complex w/ unique challenges

– Education alone insufficient, need systems change• Huge Implementation Gap - Chaotic baseline

Society of Hospital Medicine. http://www.hospitalmedicine.org/ResourceRoomRedesign/ pdf/GC_Workbook.pdf.

Michelangelo's famous statue, David, returns to Italy after US Tour

Global Prevalence of DM to double by 2030!

Focus on the non-ICU Wards:

Implementation Gap• > 1/3 with mean glucose > 180 mg/dL• 60%-70% of insulin regimens sliding scale only (even if

horrible control)• >15% with hypoglycemic episodes during their stay• 5.7% of patient-days hypoglycemic in non-critical care units• 40% of patients with hypoglycemia have more events• Uneven training / performance amongst staff• Poor coordination of tray delivery, monitoring, and

insulin• Inconsistent transitions• Patients often confused or angry• 40 – 50% of hospitals have no reliable measures

How do we overcome barriers?

• Physician buy-in / leadership

• Fear of hypoglycemia• Time and Resources• Workflow change• Information / reporting• Measurement /

feedback• Multiple teams and

hand-offs

• Skepticism of benefits• Pre-existing orders• Coordination • Staff turnover • Competing priorities• Unpredictable / varied

caloric intake• Steroids • Training failures

We know it’s broken……

How do we fix it?

My First Algorithm for Process Improvement

Essential Elements

Successful PI Efforts

• Institutional support – buy in• Teams and Culture of Improvement• Identify best practices• Understand Current Process / Practice• Defined goals• Willingness to Redesign process• Implementation of best practices / reliable

interventions• Metrics – reliable, practical, rapid feedback

– At least some measures are “real time”

• Ongoing informed improvement• Educational programs

Some designs don’t make any sense………

Even if they’ve been there a long time.

UCSD Team Structure

• Inpatient Glycemic Consult Team– 1.5 Endo’s– 3 APN/CDE’s (2 at 400 bed hospital, 1 at 200 bed hospital)

• Multidisciplinary Glycemic Control Steering Committee– Representatives from Endo, Hospital Medicine, Nursing, Pharmacy,

Surgery, Nutrition Services, IT, Nursing Education, POC Lab– Meets monthly

• Diabetes Initiative Group (Diabetes Nurse Champions)– 1-2 representatives from each unit– Meets monthly

Enhancing insulin-use safety in hospitals:Practical recommendations from an

ASHP Foundation Expert Consensus PanelCobaugh D, Maynard G, et al. Am J Health-Syst Pharm 2013;70:1404-13.

Prescribing: Recommendations 1 - 3

Recommendation 1

Develop protocol-driven and evidence-based order sets for specific uses of insulin:– IV to Subcutaneous insulin transitions, DKA, etc– Include decision-support to guide insulin use based on patient’s nutritional

status and for appropriate monitoring

Recommendation 2

Eliminate the routine administration of correction / sliding scale insulin doses as the primary strategy to treat hyperglycemia

Recommendation 3

Eliminate the use of “free text” insulin orders in electronic and paper records.

Replace them with protocol-driven and evidence-based order sets that allow for the prescribing of complex insulin regimens.

Cobaugh D, Maynard G, et al. Am J Health-Syst Pharm 2013;70:1404-13.

Integrate Best Practice into protocols, order

sets, documentation

• Actionable glycemic target • Consistent carbohydrate / dietary / consult• A1c • Patient education plan• Hypoglycemia protocol• Guidance for transitions (linked protocols)• Coordinated monitoring / nutrition / insulin• DC oral agents, insulin preferred• Insulin regimens for different conditions• Dosing guidance

Glycemic Targets in Non-Critical Care Setting

1. Premeal BG target of <140 mg/dl and random BG <180 mg/dl for the majority of patients.

2. Glycemic targets should be modified according to clinical status. – For patients who achieve and maintain glycemic control without

hypoglycemia, a lower target range may be reasonable. – For patients with terminal illness and/or with limited life expectancy or

at high risk for hypoglycemia, a higher target range (BG <200 mg/dl) may be reasonable.

3. For avoidance of hypoglycemia, we suggest that antidiabetic therapy be reassessed when BG values are 100 mg/dl). Modification of glucose-lowering treatment is usually necessary when BG values are <70 mg/dl.

Umpierrez et al Endo Society Clin Practice Guideline, 2012 J Clin Endocrinol Metabol 97(1):16-38

200

mg

/ dL

70 mg / dL

180

mg/

dL100 m

g / dL

Key things to know about SC insulin management

in the hospital - What we stress to staff

• Just do it! (when glucose over target)• Basal / Nutritional (prandial) / Correctional • What do I do when the nutrition stops? NPO p MN?• Giving that first dose (how do I do this)?• 50:50 rule –• Steroids, tube feeds, TPN• Best strategies to reduce iatrogenic hypoglycemia?• How should we manage at transitions?

Default is order set use, prompt to DC oral agents

Glycemic target, prompt for education, Diets all CHO limited.

For eating patient:Dosing guidanceBasal / Bolus defaultLast glucose / A1c displayedCorrection scale matches TDD

Admonition to avoid sliding scale. Dosing guidance for transition from infusion.Different SQ regimens for different intake.

Hypoglycemia protocolA1c order checked off.

Why measure?

• Assess local baseline• Assure the team and medical staff protocols are safe and

effective• Track progress over time• Compare like units to each other• Prioritize efforts• Benchmark – compare performance to others• Assess trade-offs between glycemic control and

hypoglycemia

Clinical Informatics and Glycemic Control

• Clinical Decision Support – Order sets with embedded CDS– Computerized insulin dosing algorithms

• Month to Month reporting• Flow sheets - Visual cues / graphics / trends • Real time reports (hyper- and hypo- glycemic outliers,

reports that capture patients “off protocol”)

An important part of a larger framework for improvement!

Glucometrics -

Measurement Challenges and Decisions

• Huge volumes of data• Inclusion - Patients with DM dx, OR hyperglycemic • Exclusion - ? ED, first hospital day?, OB-GYN• Critical Care vs non-ICU• Critical Care – measure those on infusion or all comers?• Unit of measurement

– (glucose reading? Patient day? Patient stay?)

• Hypoglycemia attribution -• Increased testing around excursions • Glucose data source - POC, serum, ABGs, all? • Comparing across sites / benchmarking • How to summarize all this?

“Glucometrics”

• Separate analysis for ICU and non-ICU units desirable• Unit of analysis –

– the individual reading (not recommended)– the patient-day– the patient-stay

• No consensus on best methods yet, but SHM offers a variety of measures

Society of Hospital Medicine: https://www.studydata.net/qgen/LoginSecure.php

Data / Reporting for Glucometrics, Community, and More

Critical Care Benchmarking212 Adult ICUs in 65 hospitals

Mean

Median

Range

Top 25th 

percentile

Patient day-weighted mean BG 160.0 156.6 120.6 – 196.3 ≤ 150.7

% patient-day BG means  ≥ 180 mg/dL   26.5% 23.1% 7.5 – 46.7%  ≤ 19.8%

% stays with BG mean (day-weighted) ≥ 180 mg/dL 26.7% 24.2% 8.6 – 53.6% ≤ 18.4%

% patient-days with any BG > 299 mg/dL 9.4% 8.2% 2.6  - 22.7% ≤ 5.7%

%  patient-days with any BG < 70 mg/dL 5.2% 4.6% 1.1 – 17.2% ≤ 3.8%

% patient-days with any BG < 40 mg/dL 0.7% 0.6% 0.1 – 2.1% ≤ 0.4%

% hypoglycemic patients with recurrence 29.3% 33.2% 15.0 - 52.7% ≤ 24.2%

Mean time- resolution of hypoglycemia  (minutes) 112 97 32 - 289 ≤ 78 

Maynard G, Schnipper JL, Messler J, Ramos P, Kulasa K, Nolan A, Rogers K. Design and Implementation of a Web-Based Reporting and Benchmarking Center for Inpatient Glucometrics. J Diabetes Science Tech 2014:published online May 12, 2014.

Non-ICU “Core” Unit Benchmarking476 non-ICU units in 76 hospitals

Mean

Median

Range

Top 25th 

percentile

Patient day-weighted mean BG 162 164.4 128.4 – 187.5 ≤ 157.0

% patient-day BG means  ≥ 180 mg/dL   29.5% 30.5% 12.0 - 45.8%  ≤ 21%

% stays with BG mean (day-weighted) ≥ 180 mg/dL 27.5% 28.4% 6.8 – 43.3% ≤ 24%

% patient-days with any BG > 299 mg/dL 10.5% 10.9% 2.7  - 21.5% ≤ 6.9%

%  patient-days with any BG < 70 mg/dL 5.0% 4.9% 1.7 - 13.1% ≤ 3.3%

% patient-days with any BG < 40 mg/dL 0.6% 0.5% 0.1 - 1.6% ≤ 0.3%

% hypoglycemic patients with recurrence 32.4% 33.2% 7.0 - 52.7% ≤ 27.3%

Mean time- resolution of hypoglycemia  (minutes) 127 120 39 - 245 ≤ 78 

Maynard G, Ramos P, Kulasa K, Rogers KM, Messler J, Schnipper JL. How Sweet is It? The Use of Benchmarking to Optimize Inpatient Glycemic Control. Diabetes Spectrum 2014:27(3):212-217.

Benchmarking Ranking Bar Chart

Hypoglycemia Rates

Glycemic control – y axis: Hypoglycemia – x axis: Note extreme variability UCSD – Top quartile performance in hypoglycemia AND glycemic control -

Top academic center: 94 hospitals, 1030 units, 305K stays, 1.1 Million patient-days.

Iatrogenic HypoglycemiaA Top Source of Inpatient Adverse Drug Events

(ADEs)

Classen DC et al. Health Aff (Millwood) 2011;30:581–9.Agency for Healthcare Research and Quality, Rockville, MD, 2011 April. HCUP Statistical Brief #109.Classen DC et al. JAMA 997;277:301–6. Bates DW et al. JAMA 1997;277:307–11. Classen et al. Jt Comm J Qual Patient Saf. 2010;36:12-21

• ADEs are most common cause of inpatient complications– affecting 1.9 million stays annually– costing $4.2 billion / year– responsible for 1/3 of hospital acquired conditions (HACs).

• 50-60% of ADEs are preventable• 57% of ADEs are from hypoglycemic agents• > 10% of those on a hypoglycemic agent suffer at least

one hypoglycemic ADE

Hypoglycemia Risk Factors - Different Flavors

InherentLow BMI / cachexia / Advanced Malignancy / Age

Liver / Kidney disease / CHF

IatrogenicInsulin / oral agents

Some risk with appropriate use.

Risk magnified with inappropriate use or failure to react / anticipate preventable problems.

Overly aggressive targets, inappropriate prescribing

Improved Glycemic Control AND

Reduced Hypoglycemia possible.

Iatrogenic Hypoglycemia from Insulin

Most common failures

1. Inappropriate prescribing

2. Failure to respond to unexpected nutritional interruption

3. Poor coordination of nutrition delivery, monitoring, and insulin delivery

4. Failure to respond to a prior hypoglycemic day

Cobaugh DJ et al. Am J Health Syst Pharm;70(16):1404-13.Hellman R. Endocr Pract 2004;10 Suppl 2:100-8.Maynard GA, et. Diabetes Spectr 2008;21 241-247.

Other failures

• Monitoring and measurement deficiencies– Only 41% of hospitals utilize their glucose data to track glycemic

control and hypoglycemia rates – Concurrent monitoring to manage outliers and those at risk for

glycemic excursions often lacking

• Storing and Dispensing– Too many insulin concentrations leads to error

• Administering– Insulin pen errors– IV bolus and insulin infusions prepared outside of pharmacy prone

to error

Cobaugh DJ et al. Am J Health Syst Pharm;70(16):1404-13

New BPA for Tube Feedings on hold + Insulin

• Will appear for pt’s w/ “0” charted for TF rate + “on insulin”• Wording will be as follows:

• Potential Problems– RNs don’t consistently chart TF interruptions in I/O– Charting, if done, not always timely

Guidance for Scheduled Temporary NPOExample UC San Diego

• Basal / Nutritional / Correction insulin terms reinforced across orders, MAR, documentation venues

• New orders not required for temporary NPO

Insulin glargine (LANTUS) injection: “basal glargine insulin should still be administered even if the patient is temporarily NPO for a procedure, or has temporary interruption of procedure”

Nutritional RAA-I guidance for eating patients: “Give with first bite of food (or up to 30 minutes after first bite of food if patient is nauseated or has poor appetite). Give 0% if patient ate less than 50%, half if patient eats 50%, and full dose if they eat all / almost all of meal.”

And Why?

Step #5 Why was patient Hypoglycemic? Critical Thinking to prevent next episode!

Proposed CDS Display for hypoglycemia evaluation

– Federal Interagency Workgroup to prevent ADE

Active Surveillance• Identify patients with a potential deficit in care, who are in

the hospital right now.

• Triage tools to quickly determine if the patient is truly uncontrolled or “off protocol”.

• Intervene to bring onto protocol, reduce risk of glycemic excursions and continued deficits in care, provide ‘just in time’ education.

aka “measure-vention”

STROKE CODE - June 10

BG = 8

Recurrent hypoglycemia on same insulin doses for several days preceding stroke code

Flow sheets: Useful from primary team AND for “ Measure-Vention”

Triage report, investigation, and mitigation all within the EHR.

Iatrogenic Hypoglycemia from Insulin

Most common failures and strategies to address them• Inappropriate prescribing

– Standardized orders with embedded CDS – mandatory use– Ongoing monitoring for inappropriate prescribing, just in time intervention

• Failure to respond to unexpected nutritional interruption– Protocols and Education– Methods to reduce interruptions in tube feeding

• Poor coordination of nutrition delivery, monitoring, and insulin delivery– Clear directions in protocols and order sets– Regular education / competency training– Redesign process

• Failure to respond to a prior hypoglycemic day– Make sure ASSESSMENT is part of hypoglycemia protocol– Competency and case based-training – Monitor recurrent hypoglycemia rates

UCSD Results – Hypoglycemia Reduction Bundle and Other Interventions Non-ICU

22,990 non-ICU patients, representing 94,900 patient-days of observation were included over five year study.RR 2013 vs 2009-10 baseline hypoglycemic stay 0.71 (0.65,0.79)severe hypoglycemic stay 0.44 (0.34, 0.58)recurrent hypoglycemia 0.78 (0.64,0.94)hypoglycemic day 0.73 (0.66,0.79)severe hypoglycemic day 0.48 (0.37,0.62)Days with BG > 299 mg/dL 0.76 (0.73,0.80)

Maynard et al. Endocrine Practice 2014 Dec 12:1-34 epub ahead of print

Hypoglycemia Reduction at UCSDSimultaneous Improvement in Glycemic Control

UCSD - Secondary prevention of Hypoglycemia

A Series of Linked Protocols: Reinforce protocols by multiple methods, hardwire whenever possible

Basic Protocols Always More to Do

SC insulin SC Insulin Pumps

IV infusion insulin Monitoring

Periop management Coordination: CHO / BG test / insulin

Hypoglycemia Management Transitions

Patient Education Provider Education / competency

References Describing SHM Glucometrics and Glycemic Mentored ImplementationRodriguez A, Magee M, Ramos P, Seley JJ, Nolan A, Kulasa K, Caudell KA, Lamb A, MacIndoe J, Maynard G. Best Practices for Interdisciplinary Care Management by Hospital Glycemic Teams: Results of a Society of Hospital Medicine Survey Among 19 U.S. Hospitals. Diabetes Spectrum 2014:27(3):197-206.

 Maynard G, Ramos P, Kulasa K, Rogers KM, Messler J, Schnipper JL. How Sweet is It? The Use of Benchmarking to Optimize Inpatient Glycemic Control. Diabetes Spectrum 2014:27(3):212-217.

 Maynard G, Schnipper JL, Messler J, Ramos P, Kulasa K, Nolan A, Rogers K. Design and Implementation of a Web-Based Reporting and Benchmarking Center for Inpatient Glucometrics. J Diabetes Science Tech 2014:published online May 12, 2014. DOI: 10.1177/1932296814532237

Rogers K, Childers D, Messler J, Nolan A, Nickel WK, Maynard G. Glycemic Control Mentored Implementation: Creating a National Network of Shared Information. Joint Commission J Qual and Patient Safety 2014,40(3):111-118.

Maynard, Kulasa et (sent to you) Hypoglycemia Reduction Bundle, accepted in Endocrine Practice

UCSD TeamTeam members

Kristen Kulasa, MD Lead Inpatient Endocrinologist

Greg Maynard, MD Hospitalist

Diana Childers, MD Assistant Clinical Professor, Hospital Medicine

Pedro Ramos, MD Assistant Clinical Professor, Hospital Medicine

Aaron Field Senior Programmer Analyst, UCSD CIIS*

Charles Choe, MD Co-investigator, Endocrinologist

Ed Fink, MHSM Project Manager, UCSD CIIS*

Brian Clay, MD Epic / EMR liaison and expert

Meghan Sebasky, MD Assistant Clinical Professor, Hospital Medicine

Patricia S. Liao, MD Endocrinologist

Kevin Box, Pharm D Pharmacist

Diane Pearson, RN,BSN,MPH,PHN,CDE

Brittany Serences, MSN,RN,FNP-BC,BC-ADM

Suzanne Lohnes, MS,BSN,RN,CDE

……and many more

Safe Glycemic ControlA Team Sport

Questions and Comments?