Family Practice Vol. 16, No. 5, 510-514
© Oxford University Press 1999
Randomized controlled trial of the effect of medical audit on AIDS prevention in general practice
Research Unit of General Practice, Odense University, Winslowparken 17, DK-5000 Odense C, Denmark.
Sandbæk A and Kragstrup J. Randomized controlled trial of the effect of medical audit on AIDS prevention in general practice.Family Practice 1999; 16:510–514.
Received 22 June 1998; Revised 29 March 1999; Accepted 13 May 1999.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and methodsStatisticsResultsDiscussionReferences |
|---|
Objective.We aimed to evaluate the effect of a medical audit on AIDS prevention in general practice.
Methods.We conducted a prospective randomized controlled study performed as ‘lagged intervention’. At the time of comparison, the intervention group had completed 6 months of audit including a primary activity registration, feedback of own data and a meeting with colleagues and experts, and had received brief summaries of the meetings and reminders about the project (a full ‘audit circle’). The participants were from general practices in Copenhagen and the Counties of Funen and Vejle, Denmark. One hundred and thirty-three GPs completed the project. The main outcome measures were the number of consultations involving AIDS prevention and the number of talks about AIDS initiated by the GP, and some elements of the content were registered on a chart.
Results. No statistically significant difference was observed in the frequency of consultations involving AIDS prevention between the intervention group (1.2% of consultations) and the control group (1.4%). Furthermore, no significant differences were observed regarding the content of these consultations or regarding the fraction of such consultations initiated by the GPs.
Conclusions.Medical audit had no observed effect on AIDS prevention in general practice.
Keywords. AIDS prevention, medical audit, randomized controlled trial.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and methodsStatisticsResultsDiscussionReferences |
|---|
In a previous study we used questionnaires and prospective activity registration to describe the role of Danish GPs in AIDS prevention.1 The study showed that GPs wanted to play a central role in the prevention of AIDS and felt they had sufficient knowledge to give advice, but the study also showed that AIDS prevention did not take up a lot of the GPs' time, and that the GPs did not comply with the guidelines from the Danish Board of Health concerning working routines. Initiatives to improve the GP's role in the prevention of HIV infection were therefore relevant.
Our knowledge about the methods most suitable for obtaining changes in the practice patterns of the GPs is, however, incomplete.2,3 In a previous study, we found no significant effect on AIDS prevention in general practice of a reference programme produced by the Danish Board of Health in collaboration with GP organizations.4 Intervention based on the GPs' registrations of their own activities (medical audit) appeared to be an attractive alternative, but little evidence for the effectiveness of the method existed.5 As for most other methods for intervention in practice patterns, very few controlled studies have been published.3
The objective of the present study was to evaluate in a randomized controlled trial the effect of a medical audit on AIDS prevention in general practice.
|
Material and methods |
|---|
|
TopAbstractIntroductionMaterial and methodsStatisticsResultsDiscussionReferences |
|---|
Participants
All GPs in the County of Copenhagen and the Counties of Funen and Vejle were invited to participate in the project. Initially, 146 GPs accepted to participate, 54 from Copenhagen (20% of the GPs in the municipality), 41 from the County of Vejle (20%) and 51 from the County of Funen (18%). After randomization, 13 GPs dropped out for various reasons: illness (2), closing down of practice (1), on leave (1), lack of interest (4) or unknown causes (5). A total of 133 GPs completed the 4-week registration period (1–28 November 1994) used to evaluate the effect of intervention: 64 in the intervention group (12% drop-out) and 69 in the control group (5% drop-out). The mean age of the participants was 49 years; 37% were women and 39% worked in single-handed practice.
Evaluation design
The effect of the intervention (medical audit) was evaluated in a prospective randomized controlled design. GPs who accepted to participate in the project were randomized into an intervention and a control group. The two groups took part in medical audit with a 6-month interval (lagged intervention) (Fig. 1
). The intervention group finalized the medical audit by carrying out their final registration during the period 1–28 November 1994. In the same period, the control group started the audit by doing the first registration. These registrations were used for comparisons between the two groups. The GPs registered the number of working days in the period, the total number of consultations and the number of consultations in which AIDS prevention was mentioned. An activity registration form was used to collect data on all consultations in which AIDS prevention was discussed (Fig. 2). The project followed the recommendations of the Helsinki Declaration and was approved by the local Scientific Ethics Committee.
|
|
The evaluated audit
The intervention comprised:
- Primary registration in a 4-week period of all consultations where AIDS prevention was mentioned.
- Individualized written feedback about the results of the primary registration of the GP and the average of the group.
- Meeting with colleagues, an HIV-positive patient and experts. Results from the primary registration were used as a basis for discussions with the aim of obtaining consensus about ‘good practice’ for prevention of AIDS. Thirty-four of the GPs in the intervention group participated in the meeting.
- A brief summary of the meeting including guidelines drawn up in consensus was sent to the intervention group.
- Reminders about the project were sent to the intervention group after 4 months.
- Final registration similar to the first registration was carried out in a 4-week period starting 6 months after the beginning of the ‘audit circle’.
|
Statistics |
|---|
|
TopAbstractIntroductionMaterial and methodsStatisticsResultsDiscussionReferences |
|---|
Differences in number of consultations between the intervention and the control groups were tested by the non-parametric Mann–Whitney U test. Differences between the first and second registration of the intervention group were tested by the non-parametric Wilcoxon Rank Sum test for matched pairs. Negative binominal regression6 was used to control the estimates of activity levels for number of working days and total number of consultations in the registration period. The statistical significance of all other differences was tested by chi-square test. A 5% level of statistical significance was used in all tests. The power of the test for difference in number of consultations was calculated by means of a poisson model using a variance stabilizing transform.7
|
Results |
|---|
|
TopAbstractIntroductionMaterial and methodsStatisticsResultsDiscussionReferences |
|---|
The median number of consultations in the 4-week study period in November 1994 involving AIDS prevention was four in the intervention group (range 0–16) as well as in the control group (0–18), corresponding to 1.2 and 1.5% of all consultations, respectively (non significant, n.s.). The power of the study was 80% for detecting a difference of just one patient. In the intervention group, the GP had taken the intiative in 35% of the consultations in which AIDS prevention was discussed, compared with 38% in the control group (n.s.). No statistically significant change in the prevalence of consultations involving AIDS prevention was observed in the intervention group between the registrations at the beginning and at the end of the audit period (1.4% compared with 1.2% of all consultations).
No statistically significant differences regarding the characteristics (age, gender and primary reason for encounter) of the patients given advice about AIDS prevention were observed between the two groups. The fraction of AIDS-related consultations including an HIV test did not differ statistically significantly between the intervention group (66%) and the control group (64%), and registrations of topics discussed were similar in the two groups (n.s.). GPs participating in the study did not differ significantly from an average of GPs in Denmark8 with regard to age, gender or type of practice (Table 1).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and methodsStatisticsResultsDiscussionReferences |
|---|
In the present study, no changes in practice patterns were observed among GPs who took part in an audit about AIDS prevention. The prevalence of consultations in which AIDS was discussed was unaffected, as were the selection of patients and the registered content of consultations. Participation in the audit process may have led to more subtle effects, e.g. change in attitudes, but the lack of effect on the registered activities was a surprise to the authors.
It is important to evaluate whether the result is due to design problems, including inability to measure an effect. Only approximately one-fifth of the GPs invited took part in the audit. The participating GPs did not differ from the non-participating GPs with regard to demographic data. Participation in quality improvement work is not mandatory for Danish GPs, and participation rates in audits are usually as in the present study.9 It may be expected that GPs take part in audits in which they have a special interest. However, we do not know anything about the participating GPs' interest in or knowledge of AIDS, nor about their experiences in caring for patients with HIV or AIDS compared with non-participants, as we did not ask them. Probably the study group did not differ dramatically in their management of prevention activities, as they recorded a similar number of talks about prevention using the same registration as a larger group of less selected GPs participating in a previous study.4
Self-registration may not be the optimal method for evaluation of the intervention effects, as it could act as an intervention itself (Hawthorne effect). We considered validating the self-registrations using data collected in some other ways, but no solutions were found. Bearing the results in mind, the registered prevalences are probably valid, as the prevalences found in the present study are similar to those demonstrated in previous studies,4 where the registration was not done for the purpose of personal feedback. In order to minimize the Hawthorne effect, the participating GPs were not informed as to what kind of effect parameters the research project was using.
Most studies of the effects of medical audits have not included a control group,9–11 and the few controlled studies have not been randomized.12,13 A controlled study has shown the effects on GPs' clinical behaviour with regard to some common childhood conditions (acute cough, recurrent wheezy chest, bedwetting, acute vomiting and itchy rash).13 The GPs taking part in the study produced in small groups a set of standards for one selected childhood condition. In addition to this activity, the intervention included a feedback of tabulated data either comparing clinical performance with that of all other groups or of only the GP's own group. The study was designed so that the various activities could be evaluated, and the conclusion was that involvement in setting clinical standards was the activity which caused a change in behaviour. The study also tried to measure the effect on the health of the children involved.12 Only in the group of children with recurrent wheezy chest did the study show an improvement in health as a result of setting standards. The outcome of audits may depend on a number of factors such as the topic for the intervention and elements in the audit process.
The topic of the audit described in this article may be a difficult one. GPs may have found it difficult to talk about AIDS with their patients. In particular it may be difficult to bring up the subject in consultations where the reason for encounter was something quite different, e.g. a medical examination needed for a driving licence. Also, the topic may not have been considered essential to the work of a GP. In a previous study we found that a reference programme sent to GPs had no effect on their handling of AIDS prevention.4 Audits (and other quality assurance methods) may have more effect when used for other less sensitive topics.
The reported audit included most of the elements recommended for a complete ‘audit circle’,14 for example registrations of the participants' own activities, feedback of results, meeting for participants with participation of other key persons, reminders and re-registrations. The audit evaluated did not produce changes in AIDS prevention in general practice, but we do not know whether changes to the content of the audit circle or any alternative methods would have produced results.2
Furthermore, we do not know whether the audit has affected the attitude of the participating GPs. Some reviews in this field15 have concluded that changing behaviour is a result of a process including various factors such as time, motivation, barriers and some CME activities. The actual audit may have started this process, but not completed it.
In Denmark, several audit and other quality improvement activities are in progress, and the National Health Service has given high priority to these activities, including financial support, but at present we do not know what to expect from these activities nor much about the costs involved. Our study points to the need for randomized controlled evaluations of the various methods used for quality improvement in general practice and in other fields. Further cost-effectiveness and cost-benefit analyses are necessary to make quality improvement activities more effective in the future.
Key messages
- This study describes the effect of medical audit including GPs tested in a randomized controlled design.
- Medical audit had no measurable effect on the behaviour of the GPs concerning prevention of AIDS.
- No statistically significant differences were observed in the frequency of consultations involving AIDS prevention between the intervention group and the control group.
- No statistically significant differences regarding the content of the AIDS prevention or regarding the talks initiated by the GPs were observed.
|
Acknowledgments |
|---|
We thank Professor Jes Søgård, Centre for Health and Social Policy, Odense University, for his statistical support.
|
References |
|---|
|
TopAbstractIntroductionMaterial and methodsStatisticsResultsDiscussionReferences |
|---|
1 Sandbaek A, Kragstrup J. Role of Danish general practitioners in AIDS prevention. Scand J Prim Health Care 1995; 13: 307–315.[Web of Science][Medline]
2
Wensing M, Grol R. Single and combined strategies for implementing changes in primary care: a literature review. Int J Qual Health Care 1994; 6: 115–132.
3
Davis DA, Thomson MA, Oxman AD, Haynes RB. Changing physician performance. A systematic review of the effect of continuing medical education strategies. JAMA 1995; 274: 700–705.
4 Sandbaek A, Kragstrup J. The effect of a reference program and a media campaign on the prevention of AIDS in general practice. Effekten af et referenceprogram og en mediekampagne på forebyggelse af AIDS i almen praksis. Ugeskr Laeger 1998, 160: 6348–6352.[Medline]
5
Robinson MB. Evaluation of medical audit. J Epidemiol Community Health 1994; 48: 435–440.
6 Cameron A, Trividi P. Regression based tests for overdispersion in the Poisson model. J Economet 1990; 46: 347–364.
7 McCullagh P, Nelder JA. Generalized Linear Models. London: Chapman & Hall, 1989.
8 Andersen F. Information about general practice as of 31 December 1992. Praksisoplysninger per 31.12 1992. Ugeskr Laeger 1993; 155: 3608–3619.
9 Munck A, Søgaard P, Gahrn-Hansen B. The effect of medical audit on treatment of respiratory tract infections in general practice. Ugeskr Laeger 1995; 157: 2851–2855.[Medline]
10 Fleming DM, Lawrence MS. Impact of audit on preventive measures. Br Med J Clin Res Educ 1983; 287: 1852–1854.
11 Anderson CM, Chambers S, Clamp M et al. Can audit improve patient care? Effects of studying use of digoxin in general practice. Br Med J 1988; 297: 113–114.
12 Russell IT, Addington-Hall JM, Avery PJ et al. Medical audit in general practice. II: Effects on health of patients with common childhood conditions. North of England Study of Standards and Performance in General Practice. Br Med J 1992; 304: 1484–1488.
13 Russell IT, Addington-Hall JM, Avery PJ et al. Medical audit in general practice. I: Effects on doctors' clinical behaviour for common childhood conditions. North of England Study of Standards and Performance in General Practice (see comments). Br Med J 1992; 304: 1480–1484.
14 Fowkes FG. Medical audit cycle. A review of methods and research in clinical practice. Med Educ 1982; 16: 228–238.[Web of Science][Medline]
15
Grol R. Personal paper. Beliefs and evidence in changing clinical practice. Br Med J 1997; 315: 418–421.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||