Development of prescribing indicators for elderly medical inpatients. C. Alice Oborne, Gwenno M. Batty, Vivienne Maskrey, C. G. Swift & S. H. D. Jackson.
B r j Clin Phormocol 1997: 43: 9 1-97
Development of prescribing indicators for elderly medical inpatients C. Alice Oborne, Gwenno M. Batty, Vivienne Maskrey, C. G. Swift & S. H. D. Jackson Clinical Age Researih Unit, Department ofHealth Care ofthe Elderly, King’s College School ofMedicine and Dentistry, London, UK
Aims To identie and improve suboptimal prescribing for elderly patients we have developed a number of prescribing indicators which focus on areas of concern and allow evaluation of the benefit of interventions. We report here on fourteen indcators. Methods The indicators are of three types: a) purely descriptive with no attempt to define optimal values, e.g. number of items prescribed per patient; b) based on unnecessary or potentially harmfd prescribing, e.g. duplication; c) assessing the appropriateness of prescribing specific drugs or combinations e.g. digoxin and warfarin/aspirin in atrial fibrillation (AF). Appropriateness was defined on the basis of objective research findings and involved comparing individual patient clinical records to criteria for appropriate prescribing. Prescribing and personal data were collected for medical inpatients aged 65 years or over in 19 hospitals in England and Wales. A total of 1686 patients were included, median age 81 years, 41% were male. Results Patients were prescribed 11 475 items, mean 4.6 regular items per patient. Completion of drug allergy/sensitivity statements varied from 3 to 93% between units. Use of generic name and specification of a maximum frequency of administration for ‘as required’ medicines were more consistent, ranging from 76-94% and 52-81% respectively. Little duplication of therapy was seen. Benzohazepines were prescribed for 22% patients, but were appropriate in only approximately one third of these. O f the 2% patients prescribed an angiotensin convecting enzyme inhibitor with a potassium-sparing duretic or potassium supplement, prescription of the combination was appropriate in 84%. Coprescription of steroids with p2-adrenoceptor agonists appeared excessive in 67% patients receiving a P2-adrenoceptor agonist, as only 51% had documented evidence of steroid responsiveness or another indcation for steroids. Stroke prophylaxis in AF was inadequate: 22% patients prescribed digoxin also received warfarin or aspirin 300 mg whereas 64% should have received the coprescription. Conclusions These prescribing indicators are sensitive to inappropriate prescribing for elderly medcal inpatients and cover a wide range of therapeutic areas. They should enable changes in prescribing quality to be measured objectively. Interhospital variation in casemix resulted in substantial hfferences in the proportion of patients in whom it would have been appropriate to prescribe specific drugs or combinations and prevented derivation of reference ranges of optimal prescribing for four indicators. Keywords: elderly patients, drug use evaluation, hospital, prescribing indicators, appropriateness of prescribing
Introduction Prescribing for elderly patients is an increasing component of NHS activity. People are surviving longer and the over 75 year olds represent a rising proportion of the community. The expanding range of medicines available, greater emphasis on preventative prescribing and multiple pathologies all contribute to elderly patients being prescribed three times more medicines than younger patients [l]. Irrational prescribing for elderly patients encompasses both overtreatment, increasing the risk of adverse drug reactions [2] and undertreatment [3]. Correspondence: Ms Alice Oborne, Clinical Age Research U n i t Department of Health Care of the Elderly, King’s College School of Medicine and Dentistry, Bessemer Road, London SES 9RS. UK
0 1997 Blackwell Science Ltd
In 1991, Beers et al. published criteria of inappropriate medication use in elderly nursing home residents as defined by a panel of 13 nationally recognised experts [4]. Decisions were based on drugs that should be avoided entirely in elderly patients and excessive doses or durations of therapy. Application in 12 American nursing homes found 40% residents received at least one inappropriate medicine and 10% were prescribed two or more inappropriate medicines [5]. These criteria have also been applied to prescribing for a non-institutionalised population aged 65 years or over [6]. Of an estimated population of 28 million, 23% received at least one contraindicated medicine during 1987, of whom 20% received two or more inappropriate prescriptions. Significant undertreatment of five conditions: hypertension, angina, hearing impairment, shortness of breath on 91
C.A. Oborne et ol.
exertion and depression was demonstrated in a sample of 1226 non-institutionalised 65 to 71 year olds [3]. Deficiencies in diagnosis and drug treatment ranged fiom 26% for hypertension to 80% for depressed patients. A hrther survey found substantial regional variation in implementation of myocardial infarction (MI) clinical trial results [7].Variation in practice behveen the 6 3 ) hocpitals was not explained by casemix or post MI complications. Given the extent of suboptimal prewribing demonstrated, assessing and improving prescribing for elderly patients should be a priority. Development of xandardised methodology, tools and reference standards to aid this is essential. Volume and cost of prescribing is regularly fed back to each general practice in England and Wales using PACT (prescribing analyses and cost), which describes dispensing for all patients in the practice. PACT data have been used to ascribe numeric standards to consensus criteria of prescribing quality, these standards were then applied to PACT data from other general practices [8]. However, PACT contains no personal, clinical or coprescription data and as such. analyses using PACT data alone are not able to evaluate appropriateness of prescribing or develop indicators of appropriateness of prescribing. A study attempting to determine appropriateness of coprescribing $,-adrenoceptor agonists and steroids for asthma from PACT data alone concluded a more sophisticated approach, which takes into account the many factorc which influence prescribing decicionc \vas required to assess and improve prescribing qualit)- [9]. The World Health Organisation (WHO) has derived indicators to describe key areas of outpatient drug use in developing countries 110. 111. These indicators have also been applied to inpatient drug use [12]. Prescribing indicators include mean number of drugs per encounter, percent of encounters with an injection prescribed and percent of encounters resulting in prescription of an antibiotic. Whilst these indicators are intended to be objective measures of prescribing behaviour allowing coniparison between prescriber 4 g/24 h)** 8. Patients prescribed H2-receptor blocker and/or proton pump inhibitor duplication** 9. Patients prescribed inhaled antimuscarinic duplication (excluding regular 81 prn)** lopatients prescribed short acting inhaled P2-adrenoceptor agonist duplication (excluding regular and pm)**
85% 41%
76.3-93.9% 3.4-93.2%
63%
51.8-81.0%
0.7%
0-7.7%
4.6%
1.3-15.4%
2.6%
0-1 7.4%
0.8%
0-9.1%
2.6%
0-25.0%
v r u g s under survedlance by the Committee for Safety of Medicines. *Optimal prescribing is 100%. +This includes acceptable use of brand name (see text). **Optimal prescribing is 0%. Indicators 7-10 report kequency of duplication as a proportion of all patients prescribed the key drug.
to be administered once only (stat) were included if they were to be administered on the day of the data collection. Two investigators obtained clinical data for patients prescribed benzodiazepines, angiotensin converting enzyme inhibitor with a potassium-sparing diuretic or potassium supplement (ACEI/K), P2-adrenoceptor agonists or digoxin for group (c) indicators. All sections of the medlcal notes were reviewed including the current hospital admission, previous admissions, GP referrals, clinic attendances and letters and discharge summaries. The kappa measure of agreement was used to assess interobserver variabhty in interpretation of clinical information (designation of appropriateness of prescribing) for 30 cases [20]. Results
Prescribing indicators Fourteen prescribing indicators were developed, ten assessed directly from drug chart data and four requiring collection of clinical data (Tables 1 and 2). As group (c) indicators were developed and piloted during the early data collection period, results reported are based on data collected in 16 to 18 of the total 19 units visited. 0 I997 Blackwell Science Ltd Br Uin Pharmacol, 43,9 1-97
The mean number of items prescribed per patient was 6.6 (range 0-18) and mean number of regular items was 4.6 (range 0-14). Around 40% of drug charts had completed medicine allergy/sensitivity statements, this varied widely between units from 3% to 93%. Generic prescribing (prescribing by generic name or items where use of a brand name was necessary) was less variable at 76 -94%, as was specification of a maximum fi-equency which 'as required' items could be dispensed, ranging from 52% to 81%. Drugs for which no maximum frequency was specified included opiates and P2-adrenoceptor agonists: 42% of all patients prescribed morphine/diamorphine and 43% patients prescribed P2-adrenoceptor agonists had no maximum fiequency documented. Duplication of prescribing and use of black triangle drugs was rarely seen in this sample of patients (Table 1).
Appropriateness oj-prescribing indicators Appropriateness of prescribing data were collected for 298/308 (97%) patients prescribed benzodlazepines, 32/39 (82%)patients prescribed ACEI/K, 261/272 (96%) patients prescribed p2-adrenoceptor agonists and 224/230 (97%) patients prescribed digoxin. Appropriateness of prescribing data for the remainder of patients on these drugs was not assessed as their clinical notes were not available at the time of data collection. Benzodazepines were prescribed for 308/1395 (22%) patients, this ranged between units from 7% to 31% patients (Table 2). Appropriate prescription of a benzodiazepine also varied between units, from 0% to 15% patients (Figure 1). Patients initiated on a benzodiazepine for night sedation in hospital frequently had a pre-existing contraindlcation to benzodiazepines such as f d s or CNS depression (70/161, 43%). Angiotensin converting enzyme inhibitors and potassium-sparing diuretics or potassium supplements were infrequently coprescribed, for 39/1686 (2.3%) patients. Coprescription was appropriate in 27/32 (84%) of these patients (Figure 2). Steroids (intravenous, oral or inhaled) were prescribed for 183/272 (67%) patients using inhaled P2-adrenoceptor agonists, yet only 51% (1341261) patients had evidence of reversibility of airways obstruction or another indication for steroids documented in their clinical notes (Figure 3). Insuficient data were available to apply the P2-adrenoceptor agonist-steroid algorithm in 7.3% patients. A significant proportion of patients prescribed steroids had failed to respond to previous steroid trial (8/23, 35%). providing positive evidence of inappropriate prescribing (this figure is likely to be an underestimate as data collectors were not 95
C.A. Oborne et al. Table 2. Indicators for which validation is required Observed prescribing
Appropriate prescribing
~
Itiditator
~
.Ileati (“A)
Range (“Yo)
Mean (%)
Range (“Yo)
32.1 2.3 67.3 32.7
7.0-30.9 0-7.5 40.0-88.9 11.1-60.0 0.0-72.2 27.3-100
7.2 1.9 51.3 41.8 63.8 36.2
0-15.0 0-3.0 15.4-73.3 12.5-84.6 37.5-100 0-62.5
c. .Ippropriarenm of prcscnbitig i r i d i i d r m
11. Benzodiazepine uset 13. ACE inhibitor and potassiundpotasslum $paring diuretic uset 13. Sr-adrenoceptor agonist prescnhed with steroid# $-adrcnoceptor agonist prescnhed without steroid# 14. Digoxin prescnbed with anticoagulantiaspinn 300 mg# Digoxin prewibrd without antlcoaplantiaspirin 300 mg# ~
~~
77 7
74.1
~
fFigure reported IS a percentage of all patients included. #Figure reported IS a percentage of patients prescnbed the key drug @,-adrenoceptor agomst or ctgolan).
specifically required to document results of investigations). Not surprisingly in this patient group, the majority of prescriptions for $,-adrenoceptor agonists without steroids were appropriate: of 85 prescriptions for $2-adrenoceptor agonists alone, 80 (94%) were appropriate. Digoxin was prescribed with either an anticoagulant or aspirin 300 mg in 51/230 (22%) of patients prescribed digoxin. This represented underprescription of stroke prophylaxis in atrial fibnllation, as clinical data collected showed that a further 94 (42%) patients prescribed digoxin for AF with no contra-indication to anticoagulant/aspirin were not receiving either drug (Figure 4). O f those patients not receiving warfarin or aspirin 300mg, 36% (34194) were prescribed a lower dose of aspirin. As seen with benzodiazepine and ACEI/K prescribing, the proportion of patients in whom coprescription of anticoagulant/aspirin with digoxin for AF and steroid with inhaled ?2-adrenoceptor agonists would have been appropriate varied widely between units (Table 2). A correlation between observed and appropriate prescribing was seen for coprescription of steroids and ??adrenoceptor agonists (r=0.63, P