Introduction

Top Abstract Introduction Subjects and methods Results Discussion References

Menorrhagia is an important healthcare problem for women of reproductive age.¹ About 5% of these women attend their general practitioner annually.² A high proportion of all gynaecological referrals are for menstrual problems,³ which results in a high cost to health services. In 1993 about 822 000 prescriptions for treatment of menorrhagia were written, costing £7m.⁴ The number of operations performed in the United Kingdom rose by 5000 in the 5 years to 1993,⁵ exposing more women to the risks of surgery. ^{6 7}

Inappropriate prescribing persists across the medical spectrum. ^{8 9} In one survey only 4.5% of patients received tranexamic acid, the most effective first line treatment for menorrhagia. ^{4 10-12} Ineffective treatment will lead to referral and a high chance of surgery, with 60% of referred women undergoing hysterectomy.³ Influencing the knowledge of and therapeutic approach to menorrhagia with an educational package could rationalise prescribing and reduce referral rates from primary to secondary care.

Postgraduate education is an established commitment for general practitioners in the United Kingdom.¹³ There is little evidence that traditional lectures, direct mailings, or guidelines result in a sustained change in physicians' prescribing and referral practice¹⁴ without effective development, implementation, and dissemination strategies.^15-17 Consequently there is a need for methods that facilitate learning and change in prescribing and referral practice.

Educational methods that have a sustained effect on physician behaviour exist. ^{18 19} These adapt "social marketing" techniques that are based on the principles of "academic detailing" (see box).²⁰ They involve the use of multifaceted interventions with clear educational and behavioural objectives. By using credible independent academic resources, referenced, authoritative, and unbiased sources of information are presented with concise graphic and written materials. Follow up meetings have been shown to double the effect. ^{21 22}

We describe the effect of an educational package on the reported treatment of and referral for menorrhagia in primary care in East Anglia.

	Subjects and methods

Top Abstract Introduction Subjects and methods Results Discussion References

Education package
We used an intervention in the form of an educational package, incorporating many of the effective features of academic detailing. ^{18 19} All general practices in East Anglia were approached and invited to participate. One of the research team visited each practice to meet partners, as an informal group, for an hour. The educational and behavioural aims and our independent funding were stated at the outset. The educational package consisted of four elements: a visual presentation of evidence taken from the literature, ^{4 23} a printed referenced summary, a flow chart for management of menorrhagia, and a follow up visit at 6 months.

Randomisation and follow up
Randomisation was at practice level. We identified 306 practices from family health service authority lists. A copy of the contents of the educational package was sent to the regional postgraduate education office for approval for educational allowance (2 hours' disease management). All practices were invited to participate by letter. After 2 weeks a phone call was made to each practice to establish their intention to participate, nominate a link partner, and arrange a meeting. This occurred from October to November 1995. Of the 306 practices, 100 were recruited. Before recruitment all practices were given an identifying number and were randomised to receive the educational package or act as controls by using a computer generated randomisation table. Randomisation was declared after recruitment. Fifty four practices were randomised to educational intervention. A presentation to the intervention groups of the educational package was given, and the control practices received an introduction to the study. Demographic features of the practices were recorded: fundholding status, training practice status, dispensing status, urban or rural location, list size, branch surgery, and proportion of male partners and partners on the obstetric list.

Analysis
The effect of intervention was assessed at the level of practices rather than patients, although individual patient data were collected. Analysis was performed on an intention to treat basis by using logistic regression to estimate the effect of intervention on the odds of the main outcome variables. These were the proportions of referrals for each practice and the prescribing of tranexamic acid, non-steroidal anti-inflammatory drugs, and norethisterone, unadjusted and adjusted for the effects of practice attributes. Figure 1 shows a time line of the study structure. Ethical approval was obtained from all eight local research ethics committees.

View larger version (33K): [in this window] [in a new window]   Fig 1.   Recruitment, randomisation, and retention of practices in study over time

	Results

Top Abstract Introduction Subjects and methods Results Discussion References

Ninety five practices (52 intervention and 43 control) returned at least one data sheet, of which 74 (40 intervention and 34 control) had patients with regular heavy menstruation (menorrhagia). Sixteen of the 100 recruited practices withdrew from data gathering, and five practices failed to return any sheets but still participated in the meetings. The reasons were pressure of work, staff shortage, illness, and break up of partnerships.

In the practices recruited there were 245 male and 103 female partners. The mean list size was 6371 in the intervention group and 5368 in the control group (P=0.2). In the intervention group 30% of the partners were women compared with 29% in the control group (P=0.6). No significant demographic differences between the two practice groups were observed.

In total 1001 completed data sheets were returned from the 95 practices. Five data sheets did not have a practice identifier code and were eliminated from analysis. The median number of forms returned per practice was seven. There were 607 data sheet returns from the intervention group (mean 12; range 1-36) and 394 returns from the controls (9; 1-24); the difference in the means was not significant. Of the data sheets, 563 were for regular heavy periods (377 intervention and 186 control); any with intermenstrual, postcoital, or irregular heavy bleeding were excluded. There was no significant difference in the proportion of consultations reported between the two groups (P=0.96).

Referral rates from reported consultations for menorrhagia are shown in table 1. The education package had a significant effect with referral rates 31% (9/29) lower in the intervention group, adjusted for demographic attributes in the practices. Patients in the intervention group had an increase in odds of receiving tranexamic acid of 2.4, an unadjusted increase of over 63% (56.8/34.9) compared with controls (table 1). Treatment with norethisterone did not differ significantly between the two groups, with 22-24% of all patients receiving it (table 1). Treatment with mefenamic acid was monitored specifically as it is commonly prescribed for menorrhagia or dysmenorrhoea, unlike other non-steroidal drugs. Intervention practices used mefenamic acid less commonly than controls but this was not significant (table 1). Both groups prescribed other non-steroidal anti-inflammatory drugs at the same rate.

Consultations with either no treatment prescribed or after one or more treatments could result in referral. Table 2 shows the proportion of patients treated with tranexamic acid, mefenamic acid, and norethisterone or combinations of treatments who were referred to hospital during the period of the study, for intervention and control groups separately.

A logistic model of the effect of practice attributes on the likelihood of treatment with differing therapies independent of randomisation showed that fundholding or branch surgery status had no significant association with treatment or referral. There was, however, a significant positive association between training practices and prescription of norethisterone (odds ratio 1.71; 95% confidence interval 1.15 to 2.55), also noted with list sizes greater than 5000 patients (1.74; 1.13 to 2.68) and when the proportion of male partners in practices exceeded 75% (1.51; 1.02 to 2.25). Significant associations were also noted between dispensing practices and prescription of mefenamic acid (1.77; 1.23 to 2.54) and urban practice location with prescription of tranexamic acid and mefenamic acid (1.81; 1.3 to 2.53 and 0.57; 0.4 to 0.82, respectively).

There were equal numbers of consultations in practices returning 10 or more data sheets and in those returning fewer than 10 (figure 2). There was a substantially greater effect of intervention on both referral and treatment in those practices returning fewer than 10 data sheets.

View larger version (17K): [in this window] [in a new window]   Fig 2.   Cases of menorrhagia reported against proportion referred to hospital from each practice. Equal numbers of cases were reported from practices returning more or fewer than 10 data sheets, thus allowing balanced comparison of reporting behaviour in terms of number of data sheets received and referral and treatment behaviour

	Discussion

Top Abstract Introduction Subjects and methods Results Discussion References

There is clear evidence that changes in practice were effected. We have shown a lower referral rate and a proportionately higher level of appropriate prescribing with the use of this educational approach. The implications for patients, general practitioners, and the health service in terms of cost and outcome are considerable.

Changes in referral
A 31% lower referral rate could result in fewer therapeutic surgical procedures as 60% of women referred with menstrual disorders will have a hysterectomy.³ We have shown a short term outcome that should be monitored in follow up studies.

Decision making
The referral process is complex. The patient is influenced by many factors: severity of symptoms, other problems, relatives' and their own preconceptions and attitudes. Influences on the doctor include training, attitudes, interviewing techniques, and confidence in managing the particular symptom pattern.²⁵ Within a group of doctors with differing educational needs a reluctance to change treatments may occur for economic reasons, lack of interest, or doubts about the relevance of the advice.²⁶ A "research-practice gap" is suggested by Dowie to explain this difference between perception of information and its value to practitioners.²⁷ Educators must bridge this gap as research findings are often reflected belatedly or not at all in the prescribing behaviour of physicians. ^{9 28 29} The challenge faced by our study group was to deal with all these points.

Changes in treatment
Despite published randomised controlled trials, tranexamic acid has remained a little used drug.¹⁰ Our educational package emphasised the low risk of antifibrinolytic drugs and encouraged their use as the rational first line treatment. Consequently an increase of 63% in the prescription of tranexamic acid was observed in the intervention group and, when the drug was given as a single agent, 45% (17 v 31) fewer referrals were made. The education package seemed to give participants the confidence to use tranexamic acid, which may be due to "demystification" of the drug, observed benefits, and patient feedback.

Study design
The prevalence of objectively measured menorrhagia is half that expressed by patients subjectively.³³ The use of the subjective complaint of "regular heavy loss that is a problem for the patient" allowed general practitioners to record and manage consultations that pragmatically reflect the real clinical problem as objective measurements of menstrual loss are not routinely available.