The influence of patients' hopes of receiving a prescription on doctors' perceptions and the decision to prescribe: a questionnaire surveyBMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7121.1506 (Published 06 December 1997) Cite this as: BMJ 1997;315:1506
- Nicky Britten, senior lecturer in medical sociology ()a,
- Obioha Ukoumunne, research associate in medical statisticsa
- a Department of General Practice, United Medical and Dental School of Guy's and St Thomas's Hospitals, London SE11 6SP
- Correspondence to: Dr Britten
- Accepted 3 September 1997
Objectives: To measure patients' expectations of receiving prescriptions and general practitioners' perceptions of these expectations and to determine the factors most closely associated with the decision to prescribe.
Design: Questionnaires were completed by patients waiting to see their general practitioners, and by their doctors immediately after the consultations.
Setting: Four non-fundholding group practices in southeast London.
Subjects: 544 unselected patients consulting 15 general practitioners.
Main outcome measures: Doctors' perceptions of patients' expectations; doctors' decisions to prescribe.
Results: 67% (354/526) of patients hoped for a prescription; doctors perceived that 56% (305/542) of patients wanted prescriptions; and doctors wrote prescriptions in 59% (321/543) of consultations. Despite the close agreement between patients' hopes and doctors' perceptions, 25% (89/353) of patients hoped for a prescription but did not receive one. In 22% (68/313) of consultations in which prescriptions were written, they were not strictly indicated on purely medical grounds, and in only 66% (202/306) of consultations in which prescriptions were written were they both indicated and hoped for. Doctors' perceptions of patients' expectations were the strongest predictor of the decision to prescribe, but the final regression model also included patients' hopes and ethnic group, and the doctor's feeling of being pressurised.
Conclusions: In an area of low prescribing and high expectations the decision to prescribe was closely related to actual and perceived expectations, but the latter was the more significant influence.
Evidence that patients' expectations influence general practitioners' prescribing is equivocal; in this study patients' hopes of receiving a prescription exceeded both doctors' perceptions and the level of prescribing
Over a quarter of patients who hoped for a prescription did not receive one
In a fifth of consultations in which a prescription was written, the prescription was not strictly indicated on purely medical grounds
Doctors' perceptions were the strongest determinant of the decision to prescribe
Doctors who felt pressurised were less likely to write a prescription if they perceived that the patient wanted one, and if they did write a prescription, it was less likely to be indicated than when the doctor did not feel pressurised
Drugs prescribed by general practitioners account for about one tenth of NHS expenditure and half the total cost of family health services.1 General practitioners' prescribing behaviour has been the focus of recent attention from policymakers anxious to reduce costs, educators wanting to improve the quality of prescribing, and pharmacists wishing to develop their professional role. Little attention has been given to the role of patients, although patients' expectations are sometimes held responsible for inappropriate prescribing.1 It is estimated that about 5% of prescriptions are not redeemed2 and that large quantities of prescribed drugs are not consumed.
Although doctors perceive a high level of demand, the evidence that patients' expectations influence prescribing decisions is equivocal.3 Most previous studies have not measured doctors' perceptions of patients' expectations but have shown that the proportion of prescriptions written often exceeds patients' expectations.4 5 6 7 8 9 This suggests that doctors overestimate patients' expectations for prescriptions.10
We aimed to measure both patients' expectations for prescriptions and general practitioners' perceptions of these expectations, and to determine the factors most closely associated with the decision to prescribe, in naturally occurring general practice consultations.
The study was carried out in four practices in the area covered by the Lambeth, Southwark, and Lewisham Family Health Service Authority in southeast London between September and December 1994. This authority had lower than average prescribing levels compared with the rest of England.11 The practices were chosen on the basis of their prescribing analysis and cost (PACT) data, to include two high prescribing and two low prescribing practices (relative to the other practices covered by the same family health service authority). All the practices were non-fundholding group practices, two of them receiving high levels and two of them low levels of deprivation payments.
We chose surgery sessions to represent different days of the week and times of day. At the beginning of each chosen session, the researcher gave a questionnaire to the doctor. This asked for the patient's name or computer number, the diagnosis or presenting problem, the doctor's perception of whether the patient wanted a prescription, whether the doctor felt pressurised to write a prescription, whether a prescription was written (and if so, the details of the drug(s)), and whether the prescription was “strictly indicated on purely medical grounds.” The 15 participating doctors completed a questionnaire for each patient immediately after the consultation.
The researcher also approached each adult patient in the waiting room and asked them to complete a questionnaire while waiting to see the doctor. An oral explanation of the study was accompanied by a written explanation for the patient to keep. The questionnaire asked for the patient's name or computer number, the symptom or presenting problem, prior drug treatment for this or other conditions, expectations of receiving a prescription that day, exemption status for prescription charges, and demographic characteristics. Expectations of a prescription were assessed with two questions to identify both ideal expectation and actual expectation.7 The first asked if the patient was hoping that the doctor would give them a prescription that day, and the second asked if the patient expected that the doctor would actually give them a prescription. Questionnaires were not given to the patients who had already completed one, to those who could not speak English, to those who saw the doctor immediately, or to those who refused. Completed questionnaires were returned to the researcher in sealed envelopes.
At the end of surgery the researcher matched the patients' and doctors' questionnaires, and the responses were then coded. Symptoms and diagnoses were coded according to the International Classification of Primary Care.12 A 10% sample of matched questionnaires was checked for accuracy of coding. Information about the doctors' qualifications was added to the database from the medical register.
The response rate was calculated by dividing the number of matched pairs of questionnaires by the total number of consultations in the chosen surgery sessions, having excluded those who did not attend, those who had already completed a questionnaire, and, in one practice, those who were known by the practice staff not to speak English. This is likely to be an underestimate as it was not possible to confirm apparent non-attenders in one practice, and some non-English speakers may not have been identified. Characteristics of responding patients were compared with the age and sex of non-respondents and with the demographic characteristics of the resident population of the inner London boroughs of Lambeth, Southwark, and Lewisham from the 1991 census.
Analysis of variance and χ2 tests were used as appropriate to identify the correlates of the two outcome measures: doctors' perceptions of their patients' expectations and the decision to prescribe. A significance level of 5% was used as the criterion for identifying correlates. Once these correlates were identified, they were used as independent variables in the forward stepwise logistic regressions of the outcome variables. The variables of the fitted models were estimated again with the multilevel modelling package MLn13 14 to separate the variation due to factors relating to patients from the variation due to factors relating to doctors.15
The response rate was 64.8% (544 completed pairs of questionnaires from 839 consultations). Comparison with non-respondents showed that men and patients aged over 44 years were less likely to respond than women and younger people. In addition, comparison with census data showed that the patients in the study were less likely to be single or economically active. Housing tenure, ethnic group, and social class were similar in the study group and census population.
The 15 participating doctors were all the doctors in the study practices. Eight were women, four had qualified outside the United Kingdom, and five had qualified before 1980. Nine of them had further qualifications in addition to the first bachelor of medicine (or its equivalent).
Patients' hopes and expectations were closely related (67% (354/526) of patients hoped for a prescription; 65% (328/504) expected one). Despite this close agreement, 4% (20/495) of patients expected to receive a prescription when they had not expressed hope for one, and 6% (29/495) expected not receive one even though they hoped for one. In the rest of the results section we refer only to patients' hopes for a prescription (not actual expectations). Patients exempt from prescription charges were significantly more likely to hope for a prescription than those who paid for their prescriptions (73% (203/277) v 61% (146/241), P<0.01). Doctors perceived that 56% (305/542) of patients wanted prescriptions, and their perceptions were strongly associated with patients' hopes (Table 1). Doctors wrote prescriptions in 59.1% (321/543) of consultations. Despite the close agreement between patients' expectations and doctors' prescribing decisions, a quarter of those hoping for a prescription did not receive one, and a similarproportion of those not hoping for a prescription did receive one (Table 1). Patients received a repeat prescription in 42% (103/245) of consultations in which a prescription was written.
Patients in the high prescribing practices were more likely both to hope for a prescription (72.2% (174/241) v 63% (180/285), P<0.05) and to receive one (67% (168/249) v 52% (153/294), P<0.001) than those in low prescribing practices.
Doctors considered that prescriptions were not strictly indicated in 22% (68/313) of consultations in which they were written. This was not associated with patients' hopes. Of the 68 non-indicated prescriptions, however, 61 were perceived as wanted, and in only 7 cases were the doctors uncertain of the patient's hopes. This association was not significant because of small cell sizes (only 15 prescriptions were written during consultations in which doctors perceived that patients did not want a prescription). Table 2 shows the logistic regression of the variable referring to clinical indication. Non-indicated prescriptions were more likely to be written during consultations in which the doctor felt pressurised, the doctor was female, or during morning surgeries, but only the first of these variables was significant. Overall, 3% (8/306) of prescriptions written were neither hoped for nor indicated, and 31% (96/306) of prescriptions written were either not hoped for or not indicated. Thus only two thirds of the total were both indicated and hoped for.
Table 3 shows the significant relations between doctors' perceptions of whether patients wanted prescriptions and other variables measured in the study. Patients whose consultation was an “extra” to the main appointment list were more likely to be perceived as wanting prescriptions, as were those who had already used self medication, those exempt from prescription charges, and those without further education. Doctors' perceptions also varied according to the patient's age and reason for encounter. Doctors' perceptions varied according to their own qualifications: patients consulting doctors with minimum qualifications or doctors who had qualified abroad were the most likely to be perceived as wanting prescriptions.
These variables were entered into a logistic regression analysis (Table 4). Of all these variables, patients' hopes for a prescription were most closely associated with doctors' perceptions. Differences associated with the patient's age, further education, prescription exemption status, and doctor's level of qualification were accounted for by the remaining variables.
The variables associated with the decision to prescribe are shown in Table 5. Doctors' perceptions and their decisions to prescribe were congruent in the case of several variables. For example, patients exempt from prescription charges were perceived as being more likely to want prescriptions and also were more likely to receive prescriptions than those who paid for their prescriptions. In other cases there was a lack of congruence; for example, doctors with two or more qualifications underprescribed in relation to their perceptions of patients' hopes, whereas doctors with fewer qualifications overprescribed in relation to their own perceptions.
Table 6 shows the logistic regression of the decision to prescribe. Doctors' perceptions' of patients' hopes for a prescription exerted the strongest influence. However, patients' hopes of receiving a prescription were independently associated with the decision to prescribe, as was the patients' ethnic group. Doctors who felt pressurised to write a prescription were less likely to write one than those who did not feel pressurised, once doctors' perceptions had been taken into account. This was because the doctors who perceived a patient as wanting a prescription were less likely to prescribe if they also felt pressurised than those who did not feel pressurised. Differences associated with the patient's age, further education, reason for encounter, self medication, prescription exemption status, and the doctor's place and level of qualification were explained by the variables remaining in the equation.
As this study was carried out in an inner city area characterised by high levels of deprivation and low levels of prescribing,16 the results may not be generalisable to other areas. The results are closely similar, however, to those of an Australian study that used comparable methods.17
This study is one of the first to measure directly the influence of both patients' hopes and doctors' perceptions on prescribing decisions. It provides quantitative estimates of the role of both these factors and shows that doctors' perceptions are the stronger influence. Thus any intervention aimed at changing prescribing behaviour should be directed at doctors' perceptions as much as at patients' hopes. Causality remains unclear. Patients' hopes were significantly higher in the high prescribing practices than in the low prescribing practices, suggesting that they were influenced, at least in part, by the doctors' previous prescribing habits.
Patients' hopes for prescriptions exceeded both doctors' perceptions of these hopes and their levels of prescribing. The level of expectations in this study was higher than in any of the previous studies using similar methods, with previous estimates ranging from 51%9 to 60%.18 The level of prescribing was comparable to that in the more recent studies.8 9 Patients in this part of London might have particularly high expectations. It is also possible that patients' expectations have risen as a result of recent initiatives such as the patient's charter, although little evidence yet supports this supposition.
A third of the prescriptions written in this study were either not indicated or not hoped for, with 3% being neither indicated nor hoped for. This suggests that scope exists for reducing prescribing without depriving patients of drugs that they either need or want. The writing of non-indicated prescriptions was primarily associated with the doctor's sense of feeling pressurised. Doctors feeling pressurised were less likely to write a prescription in consultations in which they perceived that the patient wanted one, and if they did write a prescription, it was less likely to be indicated than in consultations in which the doctor did not feel pressurised. It would require a larger sample or a qualitative study to disentangle fully the effects of doctors' perceptions of patients' hopes, their sense of feeling pressurised, and the writing of non-indicated prescriptions.
Despite the close relation between patients' hopes and doctors' prescribing decisions, the hopes of over a quarter of the sample were not met. Even in this area of high expectations, 13% of patients who doctors perceived as wanting a prescription were not hoping for one. If a prescription is not considered to be clinically necessary but is given simply because the doctor thinks the patient wants one, although the patient does not, the prescription is, by all parameters, unnecessary. Thus scope seems to exist for further exploration of patients' expectations within consultations. Bradley's work showed that doctors' assessments of patients' expectations were based on an intriguing variety of cues.19 More accurate estimates of patients' expectations could perhaps be ascertained if doctors asked patients directly if they were hoping for a prescription. This may lead to a discussion of why the patient thinks a prescription is appropriate. Patients' expectations may be based on past experience of doctors' prescribing20 or their perceptions of what is medically legitimate.21 A recent study has shown that 31% of patients consulting with sore throats used prescriptions for antibiotics when given the option of not using them.20 This compared with 99% usage in patients who were simply given a prescription for antibiotics. The finding in this study that the doctors who were more highly qualified had lower perceptions of patients' hopes and lower prescribing rates suggests that patient centred care (a focus of much postgraduate training for general practitioners) is not necessarily synonymous with higher prescribing.
We thank Professor Roger Jones and Dr Graham Calvert for their comments on an earlier draft of this paper. We are very grateful for the cooperation of patients and staff in the participating practices.
Funding: This study was funded by the NHS Executive South Thames project grant scheme.
Conflict of interest: None.