Clin Audit of start anticoagulation treatment in inpatients

I Clin Pathol 1993;46:67-71

Audit of start of anticoagulation treatment ininpatients

G Tan, H Cohen, F Taylor, J Gabbay

AbstractAims: To develop a method for evaluatingthe start of anticoagulation treatment ininpatients.Methods: One hundred case notes wereaudited using a proforma based on localguidelines in accordance with BritishSociety for Haematology recommenda-tions.Results: Confirmatory investigationswere done in 93% and 79% of patients withsymptomatic deep venous thrombosis andpulmonary embolism, respectively. Iden-tification of patients' risk factors for anti-coagulation by history taking and labo-ratory tests was often inadequate: baselinecoagulation screen, platelet count, liverfunction and renal function tests weredone in 52%, 95%, 70% and 87% of cases,respectively. There was a tendency toundertreat patients: 33% of the activatedpartial thromboplastin times (APTT) and58% of the International NormalisedRatios (INR) were subtherapeutic. Theheparin-warfarin crossover period wasparticularly problematic: 37% stoppedheparin without an INR that day, or hadan INR ofless than 2. Microscopic haema-turia was monitored occasionally. Of the62 patients continuing anticoagulation,72% were discharged with the final INR inthe therapeutic range. At discharge, only74% of patients had documented appoint-ments for the anticoagulant Clinic, theperiod between discharge and appoint-ment ranging from 0 to 12 days. Of the 25cases with an appointment exceeding fourdays after discharge, only six (24%) hadarrangements for an interim INR check.Conclusions: The experience allowed theproforma to become streamlined to amore practical, reliable, and valid tool foruse elsewhere. Findings will be fed back tothe hospital staff to promote practiceimprovements before closing the auditloop by re-evaluating practice. Furtherstudies are in progress to identify barriersexperienced by doctors in implementingthe guidelines and problems in the processof referral to the anticoagulant clinic.

(3 Clin Pathol 1993;46:67-71)

Anticoagulants are commonly used in cardiol-ogy and in the prevention and treatment ofvenous thrombosis and pulmonary embolism. 'In 1982, a quarter of a million people in the

United Kingdom were estimated to be receiv-ing oral anticoagulants each year; since thenthe numbers have increased.2 Not only doesthe quality of patient care depend on effectiveanticoagulation, but if clinical research on suchpatients cannot rely on effective anticoagulantcontrol, the research findings may be mislead-ing. Fortunately the technical control of anti-coagulation is now good, achieved mainlythrough use of the International NormalisedRatio (INR) system and careful standardisa-tion of thromboplastins.The study site is a 580 bed acute general

hospital, with anticoagulation services compa-rable with that of most district general hospi-tals. Inpatients may start anticoagulationunder the care of any physician, surgeon, orgynaecologist. Hence, although there areguidelines on anticoagulant treatment from thehospital haematologist, in reality there is widevariation in anticoagulation management. Ananticoagulant clinic supervised by a consultanthaematologist is available for monitoring andadvising outpatients. Referral of patients to theanticoagulant clinic occurs either before dis-charge from a ward, direct from anotheroutpatient clinic, or on discharge from anotherhospital where anticoagulation was started butwould not be followed up.The British Society for Haematology issued

guidelines on the use and monitoring ofheparin treatment in 19873 and revised theguidelines on oral anticoagulation in 1990.2 Atthe study hospital, guidelines on the inpatientmanagement of anticoagulant treatment fordeep venous thrombosis and pulmonaryembolism were drawn up in accordance withthose issued by the British Society for Haema-tology by a consultant haematologist afterdiscussion at the hospital physicians' auditmeetings.4 These have been circulated to alljunior medical staff since 1989, and includespecific recommendations for confirmatoryinvestigations, heparin and warfarin initiation,timing and frequency of tests for anticoagula-tion control, enquiry of risk factors such asinteractive medication, duration of anticoagu-lant courses, and referral to the anticoagulantclinic.The aim of this study was to develop a

method for evaluating the management ofinpatients starting heparin. Specific objectiveswere to design an audit proforma for use oninpatient medical case notes, to pilot its appli-cability and usefulness at an acute generalhospital and to conduct an audit of the practiceat this hospital.

67

Research Unit, RoyalCollege of Physiciansand AcademicDepartment of PublicHealth, St Mary'sHospital MedicalSchool, LondonG TanF TaylorThe Health-CareDevelopment Unit,Academic Departmentof Public Health, StMary's HospitalMedical School,LondonJ GabbayDepartment ofHaematology, StMary's HospitalMedical School andCentral MiddlesexHospital, LondonH CohenCorrespondence to:Dr Grace Tan, ClinicalResearch Fellow,Health-Care DevelopmentUnit, The Bungalow, CentralMiddlesex Hospital, ActonLane, London NWIO 7NS

Accepted for publication8 July 1992

on February 21, 2022 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.46.1.67 on 1 January 1993. Dow

nloaded from

Tan, Cohen, Taylor, Gabbay

MethodsAn audit proforma was designed, based on thehospital guidelines. The questions weregrouped under the key headings of patientparticulars, admission/discharge information,initial decision to anticoagulate, evaluation ofrisk factors for anticoagulation, initiation ofanticoagulant treatment, management ofanticoagulant treatment, documentation ofdecisions, discharge communication and infor-mation to the patient. A pilot study of bothproforma and method was conducted on 20cases. This included a reliability study of theproforma questions by crosschecking two inde-pendent auditors (the research assistant andmedically trained research fellow) using Kanalyses. This test is useful for comparingagreement between two observers classifyingnominal categories, taking into account theagreement that might be expected by chancealone. We accepted questions as reliable whenthe K score achieved a significance level ofp < 0-01.At this hospital the tests for anticoagulant

control are the activated partial thromboplas-tin time (APTT) for heparin and the Inter-national Normalised Ratio (INR) for warfarin.The normal range for the APTT is 30 to 40seconds and the recommended range for con-trol of therapeutic intravenous heparin is 50 to80 seconds. A common therapeutic range forwarfarin (of INR between 2-0 to 4 5) isadopted during the induction phase of anti-coagulation, and the INR ranges advised bythe British Society for Haematology for differ-ent conditions applied after the patient hasstabilised (for example, INR of 2-0-3-0 fortreatment of deep vein thrombosis or in atrialfibrillation; INR of 3-0-4 5 for recurrent deepvein thrombosis or mechanical prosthetic heartvalves).2

Patients were first identified from thehaematology laboratory worksheet if they hadhad either an isolated APTT or an APTTsimultaneous with an INR. They were thenentered as cases if it was confirmed whencrosschecked with the medical notes that thepatient had been treated with heparin. Patientswho had heparin prophylactically, includingpatients already receiving longterm warfarin,were excluded from the study. Inpatient medi-cal notes were retrieved retrospectively for allidentified as cases from April to July 1991, andprospectively for such patients from August toOctober 1991. Retrospective retrieval of inpa-tient medical notes was achieved by requestingcase notes from the medical records depart-ment up to five times. Eighty three case noteswere requested, for which 79 (95%) wereretrieved. Of these, 57 (72%) cases were valid

Table I Investigations done to confirm indication to anticoagulate (percentages inparentheses)

cases and entered into the audit study. Pro-spective audit of case notes was achieved byobtaining these from medical records, inter-cepting them when with medical secretaries,on the wards, or at the anticoagulant clinic. Allthe 63 case notes searched for were success-

fully retrieved: eight from medical records (sixentered the study) and 55 looked for by theresearch assistant (37 entered the study).

ResultsSTUDY POPULATIONThe 100 audited cases comprised 58 women

and 42 men. The age range was 20 to 90 years(median 57 years). The diagnoses for whichheparin was given included deep venous

thrombosis (n = 40), pulmonary embolism(n = 47), deep vein thrombosis with pulmo-nary embolism (n = 5), cerebrovascular acci-dent (n = 1), systemic embolus other thancerebrovascular accident (n = 4), pulmonaryembolism with systemic embolus (n = 1),cerebrovascular accident with systemic embo-lus (n = 1), and intracardiac thrombus(n = 1). A subgroup of 38 patients discon-tinued anticoagulation, 25 (66%) cases

because the confirmatory investigation resultwas negative, 11 (29%) because the patientdied, and two (5%) because the clinical deci-sion changed on review.For the 23 cases of deep vein thrombosis

who were symptomatic on admission (asopposed to developing the symptoms whilealready an inpatient for other reasons), thelength of stay ranged from four to 17 days(median nine days). For the 18 cases admittedwith symptoms or signs of pulmonary embo-lism, the length of stay ranged from five to 33days (median nine days).

PRE-ANTICOAGULATION PRACTICE

The guidelines recommend objective investiga-tion to diagnose deep vein thrombosis or

pulmonary embolism as soon as possible. Ofthe 45 cases clinically diagnosed as deep veinthrombosis, 93% had such investigation,usually a venogram performed within one day(range 0-7 days). Of the six cases which hadboth venogram and ultrasound scan, two hadboth tests performed the same day; two hadvenograms (results positive) before the ultra-sound scan; and two had the ultrasound scan

(results negative) before the venogram. For the52 cases diagnosed as pulmonary embolism,41 (79%) had objective investigation by perfu-sion scan. Thirty three of these 52 (63%) cases

also had ventilation scans, amounting to 80%of the 41 cases who had perfusion scans. Therewas a delay between diagnosis and perfusionscan ranging from 0 to 12 days (median threedays), and ventilation scan ranging from 0 tosix days (median three days)(table 1).The guidelines state that before starting

anticoagulants a check should be made of thepatient's current medication (particularly aspi-rin and non-steroidal anti-inflammatorydrugs), age, weight, baseline coagulationscreen, platelet count, liver and renal functiontests. The drug history was fully documented(complete drug names and doses) in only 69 of

Deep vein thrombosis (n = 45) Pulmonary embolism (n = 52)

Venogram Ultrasound Both Q scan V/Q scan

Investigation done 28 (62) 8 (18) 6 (13) 41 (79) 33 (63)Delay range (days) 0-7 0-3 1-11 0-12 0-6Delay median (days) 1 1 3 3 3

68

on February 21, 2022 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.46.1.67 on 1 January 1993. Dow

nloaded from

Audit of start of anticoagulation treatment in inpatients

Table 2 APTT and INR test results (percentages in parentheses)

APTT (days 1-10) INR (days 1-I0) INR (days 4-10)

Total tests done 439 272 186Below therapeutic range 145 (33) 158 (58) 82 (44)In therapeutic range 167 (38) 103 (38) 93 (50)Above therapeutic range 127 (29) 11(4) 11 (6)

the 100 cases, and partially documented in 30cases (the relevant pages were missing in one

case). Specific enquiry of aspirin use was madeonly in 18 cases, and non-steroidal anti-inflammatory drugs in five cases. The patient'sage and weight were noted in 97 and 70 cases,

respectively. Baseline coagulation screen (pro-thrombin time, AP'IT, and thrombin time),platelet count, liver function tests and renalfunction tests were done in 52, 95, 70 and 87cases, respectively.

INITIATION OF ANTICOAGULATION

According to the guidelines, all patients start-ing anticoagulant treatment should receive a

bolus dose of intravenous heparin which, withfew exceptions, should be 5000 IU. Thisshould be immediately followed by a con-

tinuous intravenous infusion of heparin30 000 IU/24 hours, with dose modificationfor extremes of body weight. Only 29 of the100 cases received an intravenous bolus doseas recommended. In 10 patients who were

maintained on intravenous heparin by con-

tinuous infusion, anticoagulation was initiatedwith a dose of subcutaneous heparin. Of these39 cases, 34 (87%) received 5000 IU heparin,but one received 3200 IU, one 12 500 IU andthree 10 000 IU. Because of the difficulty ofinterpreting entries in the notes, the auditorscould not assess the appropriateness of thedoses in these latter five patients. For the 75patients on whom there was sufficient informa-tion, it was calculated that heparin was startedbetween two hours to seven days after clinicaldiagnosis (median six hours). Warfarin was

initiated over a period ranging from the day ofclinical diagnosis to 25 days later (median twodays).

MONITORING ANTICOAGULANT CONTROL

The guidelines recommend that the APTT be

APTT results by individual case (days 1-10; n = 60)200 - .

150 -:a

He 100

80

50

0

a

* * a

* *

*

_ *

_^- -,w--w--.

.... w^. ..*-..* :i* **: - - -: - - - -t::.+-- :- - 9+..t- -.,--.,,.:::; 4 s:; *' .

I - ;-0- * - 6 - -:

3 *0 O. * --*S*Iw-0

.a. .1

2 3 4 500 1 0 20 30 40 50

Case numberTherapeutic range

Figure 1 APTT results by individual case (days 1-10; n = 60).

measured four hours after starting the heparininfusion and daily thereafter. More frequenttesting may be necessary, detailed in theguidelines, if heparin control is unsatisfactory.Over the first 10 days of anticoagulation, atotal of 439 APTT tests were done on 100cases, 38% of which were in the therapeuticrange (50 to 80 seconds), 33% were below 50seconds, and 29% above 80 seconds (table 2,fig 1). The mean APTT calculated for eachcase who continued anticoagulation rangedfrom 36 to 109 seconds (mean (SD) 64 (15)seconds).The hospital guidelines advise doing the first

INR check on day 3 of treatment with war-farin. Of the 60 (out of a possible 62) caseswhere relevant information was not missing orambiguous, 32 (53%) cases had an INR donebefore this (10 cases had INR tests on the firstday, 29 cases on the second day). Forty four(73%) cases did have the INR tested on thethird day of warfarin treatment. During thefirst 10 days of anticoagulation, 272 INR testswere done for the 62 cases continuing anti-coagulation. Only 38% fell in the therapeuticrange, 58% below, and 4% above (table 2, fig2). Even if we exclude the results of the firstthree days before warfarin has achieved fulleffect, of the 186 INR tests done from days 4 to10 of warfarin treatment, only 50% of tests layin the therapeutic range, 44% below, and 6%above (table 2, fig 3). Table 2 summarises theAPTT and INR results. Omitting one patientwho had an INR of more than 8 on oneoccasion, the remaining mean INR results fellin the range of 1-0 to 6-0 (mean (SD) 2-2(0-8)).

It is recommended in the guidelines thatheparin be stopped only after the INR equalsor exceeds 2-0. Of the 49 cases where relevantinformation was not missing or ambiguous, 18(37%) cases stopped heparin without an INRtest that day, or had an INR below 2-0.Another recommendation in the guidelines isthat the patient should not be discharged untilthe INR is stable between 2-0 to 4-5. We foundthat although 72% of the cases were dischargedwith the final INR in the therapeutic range,21% were under 2-0 and 7% over 4-5.Monitoring the side-effects of overanticoa-

gulation is important, and it is recommendedin the guidelines that this should include thepresence of microscopic haematuria, bruising,and bleeding from venepuncture sites orwounds. Dipstix urinalysis was recorded atleast once after starting heparin in only 19 ofthe 62 (31%) cases continuing anticoagula-tion. No major bleeds requiring therapeuticintervention were documented. Minor bleeds,frank haematuria which resolved with antico-agulant adjustment, were documented in twocases.

DISCHARGE ARRANGEMENTSAt discharge, an appointment should be madefor the earliest anticoagulant clinic, whichwould normally not exceed seven days, barringbank holidays. Appointments were documen-ted in 74% of the cases, and in these, theperiod between discharge and anticoagulant

69 on F

ebruary 21, 2022 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.46.1.67 on 1 January 1993. D

ownloaded from

Tan, Cohen, Taylor, Gabbayr-...:::.:.:...... 1:A ..................:::.:.,:::I- .....*

St'' '1 ts: :~'s, s'! ''l; *; :'jt *; **

~~~~~~~~~ i lIIi

S 3

: ;- -'

Figure 2 INR results by individual case (days 1-10; n = 60).

clinic appointment ranged from 0-12 days(median five days). For six (23%) of the 26cases with an appointment of four or moredays after discharge, a specific arrangementwas made to recall the patient to the ward tocheck the patient's INR during this period.

DiscussionThe purpose of anticoagulant treatment is toprevent the development and complications ofthrombosis. Ideally, direct outcome measuresof practice would include incidence of throm-boembolic events and bleeding episodes(which may indicate under- and overanticoa-gulation, respectively). In designing the auditproforma, however, we encountered great diffi-culty in incorporating these as criteria, partic-ularly in extracting this information from theinpatient case notes because of inadequatedocumentation. Hence we had to use proxymeasures, such as APTT and INR ranges foranticoagulation control, and microscopic hae-maturia for over anticoagulation. It is recog-nised that microscopic haematuria may be oflittle clinical importance, but we feel that itmay alert the clinician to make a more detailedassessment.

K~~~~~~~~~~~~~~~~~~4

,.'*

i *

-;i.8.^_LL.... 1 t I i .1 1._-

,.,.,.| ier-ap-euti(,- a .f

Figure 3 INR results by individual case (days 4-10; n = 60).

The value of inpatient medical notes in theaudit of clinical practice depends on thecompleteness and accuracy of these as recordsof relevant events. In the ideal situation,auditors extracting data from case notes wouldconfidently assume that an unrecorded eventdid not occur. In reality, we found that manyevents did occur but were poorly recorded, andtherefore could not be audited. Examplesinclude putting thrombo-embolic deterrent(TED) stockings on the patient, switching theheparin pump off temporarily, and discussionswith the patient. This has been discussed morefully in other studies,4 and reinforces theimportance of meticulous note-keeping, sothat audit using medical records is both fea-sible and reliable.Another limitation we encountered in audit-

ing medical notes was the difficulty in inter-preting certain documented entries. Hence formany interesting aspects of anticoagulanttreatment which we had originally incorpo-rated into the audit proforma, the repeatabilityamong auditors was very poor. For example, abarium meal report "crater in greater curva-ture" was perceived by the medically trainedauditor but not by the research assistant asindicative of a previous peptic ulcer. Otherexamples of audit criteria we were obliged toreject because of this poor reliability as indica-tors of quality included pre-anticoagulationconsideration of other risk factors such aspatient compliance, history of oesophagealvarices, history suggestive of a bleedingdisorder and heparin hypersensitivity); mon-itoring of bruising and bleeding from vene-puncture sites; and the quality of informationgiven to the patient.

Clinical practice should be periodically eval-uated against established guidelines. Theguidelines themselves should be regularlyreviewed and updated in the light of changes inepidemiology, science, and technology. Guide-lines on anticoagulation would be useful only ifsuccessfully implemented in day-to-day prac-tice by junior hospital doctors. It is funda-mental that the guidelines be distributedwidely and explained clearly, and reinforcedwith sufficient educational impetus. This auditof inpatient notes we conducted has shownthat several recommendations of the guidelinesare poorly carried out in clinical practice. Whatremains unclear are the reasons behind thismismatch. Possible barriers faced by doctorsinclude the guidelines being unacceptable orconfusing, obstacles in the hospital system forcarrying out the recommended procedures, orthe lack of support from peers and superiorsregarding use of the guidelines. To increaseimplementation of the guidelines, it is impor-tant to verify these barriers, and specifically tocounteract them. We are keen to explore thisissue more deeply, and are currently surveyingthe non-haematology doctors who manageinpatients on anticoagulation for their viewsand perspectives.Although confirmation of clinical diagnosis

by objective investigation is ideal, it may bedifficult to achieve. Diagnostic tests areunavailable at nights and weekends, and avail-

0.

.l_I

70

-.F

0 9

..l

'a.1 .. lb .,.

-:O:O.. .1, .--'

on February 21, 2022 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.46.1.67 on 1 January 1993. Dow

nloaded from

Audit of start of anticoagulation treatment in inpatients

ability may be limited at other times. Inaddition, there may be unacceptable delays inobtaining test results. The former is the mainreason for the unnecessary starting of anti-coagulation and calls to question avoidableinconvenience to the patient and waste ofhospital resources. On the other hand, heparinis a cheap drug tolerated by most patientswhen well controlled. Hence the availability ofinvestigations to confirm clinical diagnoses ofthromboembolic episodes should be period-ically reviewed in the light of changingdemand. This should include cost-benefitevaluations of whether increasing radiologicalfacilities and staffing levels would reduceunnecessary inconvenience and costs topatients and hospital. In addition, liaisonbetween ward doctors and radiologists mayneed to be improved to ensure appropriatetests are obtained promptly.A lack of appreciation of the pharmacoki-

netic properties of anticoagulants may accountfor several of the suboptimal practices. Forexample, the patient's risk to being anti-coagulated is often inadequately assessed inregard to history taking, weight measurement,enquiry of concurrent medication and inves-tigation of baseline coagulation screen, plateletcount, liver function and renal function tests.The regimen for initiating anticoagulants usinga bolus dose of heparin was inconsistent, andthe monitoring of anticoagulation was oftenillogical (for example, testing the INR beforethe third day of warfarin treatment) andinadequate (for example, failure to do urinaly-sis). It may be that the importance of thesepractices is not adequately appreciated bymedical staff, perhaps from inadequate edu-cation.The control of anticoagulation, as indicated

by mean APTT and INR values within thetherapeutic range, appears reasonable. How-ever, the high percentage of tests, both APITand INR, falling below therapeutic rangesuggests a tendency for underanticoagulationby junior hospital doctors, a finding also notedby Doble and Baron.5 This may also accountfor the significant proportion of patients dis-charged with a final INR below therapeuticrange. A particularly important period foranticoagulation control is the crossover ofheparin with warfarin. Disappointingly, in athird of the cases heparin was stopped withoutan INR test that day, or when the INR was

below 2-0. Schulman et al advocate stoppingheparin only when the INR is within thetherapeutic range for two consecutive days,indicating that a stable warfarin effect has beenachieved.6 Hospital guidelines may need toincorporate a similar specific recommenda-tion.The handover of patients from the wards to

the anticoagulant clinic is a crucial link in thecontinuity of anticoagulation care. Havingcommitted a patient to anticoagulation whileon the ward, it is imperative that follow up asan outpatient be responsibly arranged. Antico-agulant clinic appointments and INR rechecksshould be appropriate and referral adequate.Our findings have provided insight to weak-nesses which exist at the point of ward dis-charge, and we plan to develop a more detailedmethod of auditing this important process ofpatient referral.

In this study the audit proforma has beensuccessfully applied at the pilot hospital toevaluate inpatient management of anticoagu-lant treatment using medical notes. The experi-ence has enabled us to streamline the originalproforma to a more practical, reliable, andvalid tool. We have begun to feed back theresults of this study to the hospital staff, andwill close the audit loop by re-evaluatinginpatient management with the audit proformaafter relevant policy changes to improve areasof substandard practice have been imple-mented. The audit proforma is available onrequest.

This work is funded by the Research Unit, Royal College ofPhysicians (UK). We are most grateful to Dr Anthony Hopkins,director of the Research Unit, for his invaluable intellectual andpractical support.

1 Davies GC, Salzman EW Cost effectiveness of prophylaxisof venous thrombeombolism. J Roy Soc Med 1981;74:177-80.

2 British Society for Haematology, British Committee forStandards in Haematology Haemostasis and ThrombosisTask Force. Guidelines on oral anticoagulation: Secondedition. J Clin Pathol 1990;43:177-83.

3 British Society For Haematology - Report prepared byBritish Committee for Standards in Haematology Hae-mostasis and Thrombosis Task Force. Guidelines on theuse and monitoring of heparin therapy. In Roberts B, ed.Standard Haematology Practice. Oxford: Blackwell Scien-tific, 1991:88-93.

4 Gabbay J, McNicol M, Spiby J, Davies SC, Layton AJ. Whatdid medical audit achieve? Lessons from the preliminaryevaluation of a year's medical audit. BMJ 1990;301:526-9.

5 Doble N, Baron JH. Anticoagulation control with warfarinby junior hospital doctors. J Roy Soc Med 1987;80:627.

6 Schulman S, Lockner D, Bergstrom K, Blomback M.Intensive initial oral anticoagulation and shorter heparintreatment in deep vein thrombosis. Thromb Haemostas1984;52:276-80.

71 on F

ebruary 21, 2022 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.46.1.67 on 1 January 1993. D

ownloaded from