Family Practice Vol. 17, No. 3, 254-260
© Oxford University Press 2000
Adherence to the guidelines of a regional formulary
Department of General Practice of Groningen University, Ant. Deusinglaan 4, 9713 AW Groningen, The Netherlands.
Kamps GB, Stewart RE, van der Werf GTh, Schuling J and Meyboom-de Jong B. Adherence to the guidelines of a regional formulary. Family Practice 2000; 17: 254–260.
Received 23 June 1999; Revised 5 November 1999; Accepted 21 December 1999.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Background. Pharmacotherapeutical guidelines, called formularies, have been developed to facilitate effective, efficient and cost-conscious prescribing. Monitoring adherence to such guidelines may be a reasonable way of assessing prescribing practices.
Objective. The aim of this study was to assess how strictly the GPs participating in our department's registration network adhere to the guidelines of the regional formulary, and which indications and drugs the GPs used.
Methods. This is a descriptive study, concerning 1000 consecutive prescriptions from each of the 17 participating GPs. The third edition of the Groningen formulary (GFIII), published in 1995, was used. If the drug prescribed was advised in the formulary, we considered it to be global adherence. If the indication was mentioned in the formulary, and the drug prescribed was advised for that indication in the formulary, it was considered to be specific adherence. Both the medications prescribed and the health problems registered by the GPs, but not mentioned in the GFIII, were analysed.
Results. The 17 000 prescriptions chosen for analysis formed ~25% of all prescriptions written by the GPs in 1 year. The indications for only 24 prescriptions (0.14%) were missing. Among the 17 GPs, the number of different drugs prescribed varied between 167 and 219 per 1000 prescriptions. The global adherence varied from 76 to 89% among the GPs, and the specific adherence varied from 55 to 71%. Of the 17 000 prescriptions, 11 457 (67%) concerned indications mentioned in the GFIII. Prescriptions for indications not mentioned in the GFIII contained 4353 (78.5%) drugs advised in the formulary. Of the 251 medications mentioned in the GFIII, only 15 (6%) were not prescribed.
Discussion. The GPs in our study were neither representative, nor were they chosen at random. Their patient population was comparable in age, sex and insurance status. These findings are an example of what level of adherence is obtainable. The formulary covered approximately two-thirds of the indications registered by GPs, and did not contain many unnecessary medications (6%).
Keywords. Family physicians, formularies, guidelines adherence, practice guidelines.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
It is important to develop evidence-based guidelines and to set criteria for quality of care for busy GPs.1–4 This applies particularly to pharmacotherapy, because prescribing is the most frequent therapeutic intervention used by GPs. However, developing guidelines and setting criteria is only the first step. The second step involves an assessment of performance with respect to guidelines and criteria. The third step, a critical evaluation, closes the quality circle.
The goals of good prescribing or rational pharmacotherapy should maximize effectiviness, minimize risks and costs, and respect patient preferences.5,6 At the 1987 WHO conference in Nairobi, the following statement was agreed upon: "rational use of drugs requires that patients receive medications appropriate to their clinical needs, in doses that meet their own individual requirements, for an adequate period of time and at the lowest costs to the community".7 However, these sweeping statements are optimal theoretical criteria which offer no solution to the basic problem of assessing good prescription practices. The first question to be addressed is whether or not it is necessary to prescribe at all. Most studies advocate limited prescribing in order to prevent adverse reactions and extensive costs.5,8
Once the physician has decided to prescribe, he has to decide what to prescribe. Pharmacotherapeutical guidelines, called formularies, have been developed to facilitate effective, efficient and cost-conscious prescribing. Monitoring adherence to such guidelines may be a reasonable way of assessing prescribing practices. However, guidelines may vary greatly between regions. When six primary care guidelines were compared in The Netherlands, the number of advised drugs for one specified indication ranged from five to 12, and consensus about one drug varied from 35 to 100% depending on the diagnosis.9 This lack of consensus is confusing for GPs.
In addition, there is no general consensus about the aims of formularies, and it is often unclear to the physician which formulary applies to a specific case. Grant stated that 90% of the problems for which patients contact their GP should be mentioned in the formulary, and the GPs should adhere to the guidelines in 90% of the cases.10 Barber advocated the setting of standards, so that at least 80% of the prescribed medicines would meet the protocol.5 Bergman et al. advised that drugs accounting for 90% of drug use (DU90%) should serve as an indicator for the quality of drug prescribing.11
To assess the possibilities for improvement among the GPs participating in our department's regional registration network, we initiated a study to address the following question: how strictly are the GPs adhering to the guidelines of their regional formulary, and which indications and drugs from the formulary were used by the GPs?
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The data concerning the indication-related medication were collected by the 17 GPs participating in the Morbidity and Medication Registration Network of the Department of General Practice at Groningen University (RNG), in The Netherlands.12,13 This registration network was established in 1989 for the collection of reliable data for research and education, and for the development of policies for care provided by GPs.
The GPs were recruited from among active GPs involved in teaching and research. These GPs are a selected group, which meets regularly about once in every 3 months to discuss registration issues. All GPs use the same computerized health information system (HIS). HIS fulfils the criteria set by the Dutch College of General Practitioners and is commercially available. Until now, it did not contain an electronic formulary. Two adaptations were made for the RNG to make it possible for diagnoses to be recorded in an episode-linked manner, and to ensure that prescriptions could only be printed if an International Classification of Primary Care (ICPC)-coded indication was filled in.
Of the 7000 GPs in The Netherlands, 
90% own a computer, 88% have a HIS, and 80% use the medical module for the daily recording of referrals and prescriptions.14 The HIS used by the network GPs is actually used by ~20% of the GPs in The Netherlands.
During all face to face encounters, the diagnosis, referrals and prescribed medications were systematically registered and electronically recorded. All prescriptions written out after telephone contact with the practice assistant were also electronically registered. Using these methods, a large database of indication-related prescriptions has been developed.
A total of 1000 consecutive prescriptions from each of the 17 participating GPs were analysed. These 17 000 prescriptions were a mix of new and repeat prescriptions. We chose the last 1000 consecutive prescriptions given before December 15th 1997, in order to avoid the period around Christmas, as precription behaviour is often influenced by holidays and locum situations. Health problems were registered using the ICPC.15 The prescribed medications were coded according to the Anatomical Therapeutical Classification (ATC).16
Regional formulary
Groningen's regional formulary first appeared in 1991, through the consensus of three GPs, two specialists and seven pharmacists. While this group stated that they based their decisions on evidence from publications, they gave no references in the formulary. The third edition of the Groningen Formulary (GFIII), published in 1995, was used in this study.17 It contained 186 health conditions and 251 medications. No GP participating in the registration network took part in the development of the regional formulary.
The formulary is distributed free of charge to all GPs in the region by the Sick Fund, and is used in the medical training of students and vocational training of GPs.
Analysis
For each prescription, we checked if the prescribed drug matched the advice in the GFIII. If the prescribed drug was advised in the formulary, we considered it to be global adherence (the prescription of a drug mentioned in the GFIII). If the indication was mentioned in the formulary, and the prescribed drug was advised for that indication in the formulary, it was considered to be specific adherence (the prescription of a drug advised for the indication from the formulary). By calculating confidence intervals, real differences were assessed.
Finding unsuitable prescriptions for specific health problems was considered a ‘mismatch’: a measure to assess the reliability of the registering system.
To assess the extent to which the formulary covered the GPs' prescriptions, both the medications and the health problems registered by the GPs, but not mentioned in GFIII, were analysed.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
There were 17 GPs participating in the network: 14 from three group practices and three from solo practices belonging to one locum group. All practices were located in towns of two northern provinces of The Netherlands. Among the 14 GPs from the group practices, there were eight women (two full-time) and six men (three full-time). The three solo GPs were all men and all practised full time. In one of the group practices, the GPs were salaried; the others worked on a fee for service basis.
A total of 148 628 prescriptions for 30 109 patients were registered in the RNG in 1997. Of these prescriptions, 68 026 were prescribed by the 17 participating GPs, 4261 prescriptions by other doctors (trainees or locums) and 76 341 prescriptions by practice assistants.
The 17 000 prescriptions chosen for analysis formed ~25% of all prescriptions written by the GPs in 1 year. Those were written out for 7189 patients of whom 63.4% were women; the mean age for all patients was 43.6 years (SD 22.4); 66.6% were insured by government health programmes.
The mismatch, i.e. unsuitable prescription behaviour, e.g. eyedrops for arrythmia, or vaginal miconazole for anxiousness, was <1% for all 17 000 prescriptions, and was due primarily to clerical inaccuracy.
Since the system required a diagnostic code before the prescription could be printed, the indications for only 24 prescriptions (0.14%) were missing. These were repeat prescriptions, which were initiated before the adaptation of the electronic system was realized.
The distribution of the 17 000 prescriptions over the different ICPC chapters is shown in Table 1
with the global and specific adherence per chapter. Five ICPC chapters contained 64% of all prescriptions, for respiratory, skin, cardiovascular, musculoskeletal and psychological disorders, respectively. Over the different chapters, global adherence ranged from 67 to 95% and specific adherence ranged from 32 to 90% per chapter. The global adherence correlated 0.36 (P = 0.17) with the number of drugs per ICPC chapter.
|
The lowest global and specific adherence concerned the chapters hearing (H), psychological (P) and male genital (Y) problems. In chapter H, 41% of the indications concerned otitis media. Although the guidelines advised prescribing antibiotics only for children younger than 2 years old, the GPs prescribed antibiotics in 57% of cases of otitis media for all patients. Low global adherence in chapter P is due to several factors which include the time which has passed since the guidelines were published, the heterogeneity of chapter P and the lack of symptom diagnoses. The guidelines on depression contained mainly old tricyclic antidepressants, whereas, in 57% of the cases, GPs tend to prescribe modern antidepressants. This is illustrated in Table 2
, where the variation in adherence for the 13 most frequent indications is shown. Here, adherence to the guidelines is one of the lowest for the diagnosis of depression. The explanation for low adherence in chapter Y is not clear. This may be due to the low number of prescriptions in the chapter.
|
The highest global and specific adherence was seen in the chapter on family planning (W). Though specific adherence was low for the chapter on respiratory diseases (chapter R), the nonetheless high global adherence may be explained by the relatively high number of respiratory drugs (n = 33) mentioned in the formulary. For infectious diseases, such as bronchitis, and symptoms such as cough, antibiotics were prescibed relatively often, contrary to the advice in the formulary. In chapter T, which has as its main topic diabetes mellitus, the specific adherence was very high, due to the limited number of antidiabetic drugs.
Adherence
Among the 17 GPs, the number of different drugs prescribed varied between 167 and 219 per 1000 prescriptions. The global adherence varied from 76 to 89% among the GPs (Table 3 or Fig. 1). The only significant difference among the GPs was that one GP from group practice 2 differed from the six GPs with the highest global adherence.
|
|
The specific adherence varied from 55 to 71% (Table 3
or Fig. 2). The differences were as follows: six GPs showed fewer adherences than the two GPs with the highest adherence; only one GP differed from the majority. All of the five GPs from group practice 1 showed high specific adherence.
|
There was no clear correlation between global and specific adherence. The GP with the lowest specific adherence showed high global adherence, and two partners in a ‘one man’ practice within group practice 3 showed the same adherence pattern (Table 3
).
Coverage of the formulary
From the 17 000 prescriptions, 11 457 prescriptions (67%) concerned indications mentioned in the GFIII. The remaining indications (n = 5543) were not mentioned in the GFIII. Prescriptions for indications that were not mentioned in the GFIII contained 4353 (78.5%) drugs advised in the formulary.
For the 17 000 prescriptions written, 549 different medicaments were prescribed. From the 251 medications mentioned in the GFIII, only 15 (6%) were not prescribed. The frequently prescribed drugs were usually advised in the formulary, while the less frequently prescribed drugs were not mentioned in the formulary. On the one hand, among the 45 drugs prescribed in at least 0.5% of the 17 000 prescriptions, totalling 57% of all prescriptions, only one drug (paroxetine) was not mentioned in the formulary. On the other hand, among the 207 drugs prescribed only once, twice or three times, 170 drugs were not mentioned in the formulary.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
These findings are an example of what level of adherence is obtainable. It probably represents best practice; however, there may even be room for improvement when explicit attention is paid to prescribing according to the formulary guidelines, and if the collective data are used by individual GPs to gauge their prescription behaviour.18 There is a large range in adherences for different diagnoses and among different GPs.
The adherences to the regional formulary approached the standard set by Barber, although 80% global and specific adherence was not reached by all the GPs studied.5 Another study involving 63 GPs in The Netherlands found that global adherence had a mean of 69% (range 56–88) and specific adherence a mean of 41% (range 20–60), compared with 85% for global and 63% for specific adherence in our study.19
The GPs in our study were neither representative, nor were they chosen at random. These GPs chose to participate in the Department's Regional Morbidity and Medication Network and were recruited among the most active GPs. Participating in the RNG did not limit the choice of medications available to the GPs. They were free to choose whatever drug they felt their patient required.
Although the GPs were a selected group, the patient population was comparable in age, sex and insurance status with patients from the same region, with the exception of a 0.3 higher percentage of women. There was also a lower percentage of the age group of 0–19 years and of the age group <65 years (1.7 and 0.3%, respectively) in the network than found generally in the region.12
The department staff organized regular meetings with network GPs concerning registration issues and feedback on their behaviour in practice. There were no set sessions dedicated to the regional formulary, nor were GPs encouraged to prescribe specifically according to the guidelines. Until now, no electronic formulary or prescription system has been implemented for GPs in The Netherlands.
The formulary covered approximately two-thirds of the indications registered by GPs, and did not contain many unnecessary medicines (6%). Interestingly, one-third of the registered indications were not listed in the formulary, although drugs mentioned in the formulary were prescribed for the majority of these indications. This is understandable since the same painkiller may be prescribed for low back pain, which is a formulary indication, and for shoulder pain, which is not mentioned in the formulary.
Undoubtedly, the coverage could increase if more prescriptions were analysed, but the emerging picture would not change fundamentally. Although the GFIII did not cover the morbidity completely in its present form, it is certainly a reasonable standard. However, there is room for improvement in the formulary. It should be based on the incidences and prevalences of health problems in general practice, and the evidence on which the formulary is based should be referred to and accounted for in the formulary.
If the guidelines of the formulary are developed according to set criteria, Barber's ideal of 80% could be obtained. About 80% of health problems encountered in general practice would be contained in the formulary, and at least 80% of prescribing practices would meet the formulary specifications. Grol et al. found that guideline recommendations were followed in an average of 61% of cases.20 The Barber rule of 80% seems more realistic than the 90% rule of Grant, or the DU90% of Bergman. In the near future, an electronic formulary will be implemented in the GPs' health information system in The Netherlands, which holds much promise, and should allow the 80% rule to be attained. In further studies, patient preferences should also be taken into account
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
1 Grol R. National standard setting for quality of care in general practice: attitudes of general practitioners and reponse to a set of standards. Br J Gen Pract 1990; 40: 361–364.[Web of Science][Medline]
2
Giuffrida A, Gravelle H, Roland M. Measuring quality of care with routine data: avoiding confusion between performance indicators and health outcomes. Br Med J 1999; 319: 94–98.
3
West E, Newton J. Clinical guidelines. Br Med J 1997; 315: 324.
4
Baker R, Fraser RC. Development of review criteria: linking guidelines and assessment of quality. Br Med J 1995; 311: 370–373.
5
Barber N. What constitutes good prescribing? Br Med J 1995; 310: 923–925.
6 Parish PA. Drug prescribing—the concern of all. J R Society Health 1973; 4: 213–217.
7 WHO Conference of Experts on the Rational Use of Drugs, Nairobi. Geneva: WHO, 1987.
8 Rucker TD, Schiff G. Drug formularies. Myths-in-formation. Med Care 1990; 28: 928–942.[Web of Science][Medline]
9 Kamps GB, Meyboom-de Jong. Regionale formularia voor huisartsen vergeleken (Regional formularies for GPs compared). Ned Tijdschr Geneeskd 1997; 141: 1002–1007.[Medline]
10 Grant GB, Gregory DA, van Zwanenberg TDA. Basis Formulary in General Practice. Practical Guidelines for General Practice. Oxford: Oxford University Press, 1987.
11 Bergman U, Popa C, Tomson V, Wettermark B, Einarson TR, Aberg H, Sjoqvist F. Drug uitilization 90%—a simple method for assessing the quality of drugs prescribing. Eur J Clin Pharmacol 1998; 54: 113–118.[Web of Science][Medline]
12 Werf GTh van der, Smith RJA, Stewart RE, Meyboom-de Jong B. Spiegel op de huisarts; over registratie van ziekte, medicatie en verwijzingen in de geautomatiseerde huisartspraktijk. Disciplinegroep huisartsgeneeskunde, Rijksuniversiteit Groningen, 1998.
13 Wieringa N, Graeff P de, Werf GTh van der, Vos R. Cardiovascular drugs: discrepancies in demographics between pre- and post-registration use. Eur J Clin Pharmacol 1999 55: 537–544.[Medline]
14 Althuis TP. Nut III, een studie naar automatisering van Nederlandse huisartsen 1997 in (A survey of automatization of Dutch General Practitioners in 1997). Utrecht: NHG (Dutch College of General Practitioners), 1999.
15 Lamberts H, Wood M. International Classification of Primary Care. Oxford: Oxford Medical Publications, 1987.
16 Guidelines for ATC Classification and DDD Assignment. Oslo: WHO Collaboration Centre for Drug Statistics Methodology, 1995.
17 Groninger Formularium, 3rd edn. Groningen: Stichting Farmaceutische Dienstverlening, 1995.
18
Grol R. Implementing guidelines in general practice care. Qual Health Care 1997; 1: 184–191.
19 Kamps GB, Meyboom de Jong B. Voorschrijven volgens de regels. Compliantie van jtwee groepen huisartsen met hun regionale formularia. (Prescribing according to guidelines. Adherence of two groups of GPs to their own regional formulary.) Huisarts Wet 1998; 41: 416–420.
20
Grol R, Dalhuijsen J, Thomas S, Veld C in't, Rutten G, Mokkink H. Attributes of clinical guidelines that influence use of guidelines in general practice: observational study. Br Med J 1998; 317: 858–861.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A H.-V. de Wal, R. Smith, G. van der Werf, and B Meyboom-De Jong Towards improvement of the accuracy and completeness of medication registration with the use of an electronic medical record (EMR) Fam. Pract., June 1, 2001; 18(3): 288 - 291. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||