Quality at general practice consultations: cross sectional surveyBMJ 1999; 319 doi: https://doi.org/10.1136/bmj.319.7212.738 (Published 18 September 1999) Cite this as: BMJ 1999;319:738
- John G R Howie, professor ()a,
- David J Heaney, research fellowa,
- Margaret Maxwell, research fellowa,
- Jeremy J Walker, research fellowa,
- George K Freeman, professorb,
- Harbinder Rai, research assistantb
- a Department of Community Health Sciences-General Practice, University of Edinburgh, Edinburgh EH8 9DX
- b Department of Primary Health Care and General Practice, Imperial College School of Medicine, Chelsea and Westminster Hospital, London SW10 9NH
- Correspondence to: J G R Howie
- Accepted 29 July 1999
Objectives: To measure quality of care at general practice consultations in diverse geographical areas, and to determine the principal correlates associated with enablement as an outcome measure.
Design: Cross sectional multipractice questionnaire based study.
Setting: Random sample of practices in four participating regions: Lothian, Coventry, Oxfordshire, and west London.
Participants: 25 994 adults attending 53 practices over two weeks in March and April 1998.
Main outcome measures: Patient enablement, duration of consultation, how well patients know their doctor, and the size of the practice list.
Results: A hierarchy of needs or reasons for consultation was created. Similar overall enablement scores were achieved for most casemix presentations (mean 3.1, 95% confidence interval 3.1 to 3.1). Mean duration of consultation for all patients was 8.0 minutes (8.0 to 8.1); however, duration of consultation increased for patients with psychological problems or where psychological and social problems coexisted (mean 9.1, 9.0 to 9.2). The 2195 patients who spoke languages other than English at home were analysed separately as they had generally higher enablement scores (mean 4.5, 4.3 to 4.7) than those patients who spoke English only despite having shorter consultations (mean 7.1 (6.9 to 7.3) minutes. At individual consultations, enablement score was most closely correlated with duration of consultation and knowing the doctor well. Individual doctors had a wide range of mean enablement scores (1.1-5.3) and mean durations of consultation (3.8-14.4 minutes). Doctors' ability to enable was linked to the duration of their consultation and the percentage of their patients who knew them well and was inversely related to the size of their practice. At practice level, mean enablement scores ranged from 2.3 to 4.4 and duration of consultation ranged from 4.9 to 12.2 minutes. Correlations between ranks at practice level were not significant.
Conclusions: It may be time to reward doctors who have longer consultations, provide greater continuity of care, and both enable more patients and enable patients more.
38% of practices approached on a random basis in four areas of the United Kingdom collected data for 2 weeks showing the feasibility of surveying the content and outcome of routine consultations in general practice
At consultation level, enablement correlates best with the duration of consultations and how well the patient knows the doctor
These correlates apply at doctor level as well—more enabling doctors work in smaller practices than less enabling doctors
Case mix does not seem to be a determinant of enablement scores, but patients with more complex problems require longer consultations to achieve equal enablement
Patterns of duration of consultation and enablement in patients who speak languages other than English are different and require further study
The definition and measurement of quality of care in general practice is set to maintain a high profile as issues relating to quality assurance and clinical governance in primary care groups and local healthcare cooperatives1–3 engage management and professions.
Within general practice, work on quality and the development of performance indicators4–6 is in hand on issues of practice organisation.7 8 care of continuing health problems,9–12 and achievement of public health targets.13 14 However, the core activity of general practice remains the consultation. Two areas of work in this discipline are particularly relevant to this paper: the use of time in consultations and its relation to “enablement,” an outcome measure that seems related to, but different from, satisfaction15–19; and continuity of care.20–22
Our two principal aims were to see if survey methods developed and used to study quality of care at consultations in volunteer practices in Scotland17 could be developed further and used on a large scale with randomly selected practices elsewhere in the United Kingdom; and to explore the correlates of enablement and to see whether these could be modelled.
Participants and methods
A preconsultation questionnaire was completed by all patients aged 12 and over. One set of questions asked about problems in general terms (acute or urgent, new or continuing, social, psychological, administrative, wish for general health advice, need for a prescription, returning on request), and a further set of questions asked which of these problems they wished to discuss. The 12 question version of the general health questionnaire and a set of five questions on social wellbeing were added to develop the psychological and social case-mix components.
Patients were asked what languages they spoke at home, what languages they expected to use at consultations, how well they knew the doctor they were going to see, and whether the doctor was the one of their choice.
Doctors completed information on the time the consultation started and finished, whether the consultation was booked, open, or an emergency session, whether the patient was a “fit in”—that is, added to the surgery session without having an appointment—or a temporary resident, and whether a student was present.
After the consultation, the patient completed the patient enablement instrument (figure). The original instrument has been developed from previous work by the addition of a “not applicable” option.19 Its conceptual basis has been described elsewhere17 and derives from the assumption that what is important in predicting outcome is how the respondent feels and perceives life.23 Responses of “much better,” “better,” and “same or less” or “not applicable” were scored 2, 1 and 0 respectively, giving a score range of 0-12. Patients also indicated whether a prescription was given and whether the consultation was interrupted
All information was recorded on a single form. The patient's preconsultation information was hidden from the doctor on completion, and the outcome questions were hidden from the patient before the consultation, by the use of seals.
Some help for patients who had difficulty completing the questionnaire was available in the waiting room. When patients were unable to provide information or declined to take part or when receptionists did not hand out forms, note was still made of when consultations started and finished. A template in Punjabi was prepared for practices in west London and Coventry, and help was available with this and other South Asian languages in many of the practices.
The intention to recruit 50 doctors in 10 practices in each of the four participating regions (Lothian, Coventry, Oxfordshire, and west London) was achieved by inviting a random sample of about twice the necessary size to attend a series of briefing meetings. In practices of up to three partners all had to agree to take part, but three out of four partners, four of five partners, or all but two partners in larger practices were regarded as sufficient. Trainees and assistants were included where appropriate. Practices were asked to collect consultation data for two consecutive weeks during March and April 1998. Ethical approval was obtained in all four regions.
Doctors were asked to complete an attitude survey instrument24 to identify their levels of patient centredness as defined in previous work.16 Practices provided information on their demography, and doctors provided information about their training and the languages they spoke at home.
Information was analysed with SAS. We carried out multiple regression analysis with enablement as the outcome variable, and correlation coefficients were computed where appropriate to assess the strength of associations between variables.
Overall, 221 doctors in 53 practices collected information on 25 994 consultations with adult patients. The mean enablement score for English speaking patients was 3.1 and the mean consultation length was 8.0 minutes. The 2195 patients who indicated they spoke languages other than English at home were categorised as “other language” patients. About a quarter of these expected to consult in their own language. Their consultations were more enabling (mean enablement score 4.5) and shorter (mean 7.1 minutes) irrespective of whether they received help to complete responses. Except where indicated, all analyses in this paper are based on the 23 799 consultations with English speaking patients. Mean duration of consultation for a further 7338 patients who did not complete a questionnaire was 0.37 minutes longer than for those who did.
Needs hierarchy and reason for consultation
Overall, 10 758 (45.2%) patients reported an acute illness, 10 011(42.1%) reported chronic health problems, 9855 (41.4%) reported social problems, and 7062 (29.7%) reported psychological problems (including general health questionnaire-12 scores of 5 or above). Administrative issues (including wishes for general health advice and for prescriptions, without identification of an acute or chronic problem) occurred in a further 2007 (8.4%) patients. Less than 1.0% of patients could not be categorised (237 patients), and 4760 (20.0%) of patients wanted to discuss more than one problem.
Patients with an acute or chronic problem, or both, but neither social or psychological problems were categorised as “biomedical.” Those with or without biomedical problems but with either a social or a psychological problem (but not both) were described as “social” or “psychological” respectively Those with any combination of problems, including both social and psychological, were classed as “complex.” Those with an administrative need but no other need described above were categorised as “administrative.” With this approach, 9413 (39.6%) patients had biomedical needs, 12 142 (51.0%) had a mix of social and psychological problems, and 2007 (8.4%) had administrative needs.
Tables 1 and 2 show the principal correlates with enablement at consultations. Table 1 shows that enablement was similar across case mix. Mean duration of consultation increased when consultations had a psychological component. High mean duration of consultation was associated with a smaller number of short consultations and a greater number of long and very long consultations.
Patient's age over 65 was associated with high enablement and long consultations, and consultations for women lasted longer than those for men (table 2). Patients who wanted, but did not get, a prescription reported lower enablement for equal duration of consultation, and this was particularly so for biomedical problems. Patients with social problems alone had the lowest enablement score of any case mix category. However, patients with a social problem combined with a high general health questionnaire-12 score achieved high enablement; this was associated with greater duration of consultation. The more problems a patient wanted to discuss, the longer the consultation and the greater the enablement. “Fit in” consultations were shorter than routine consultations whether booked or open; this was most profound for psychological problems, where enablement achieved was also lower Knowing the doctor well resulted in considerably increased enablement. Enablement was greatest in single handed practices and lowest in practices of six or more doctors. The proportion of doctors known well increased with the doctor's experience, but less experienced doctors compensated by spending more time with unfamiliar patients than did more experienced doctors and therefore achieved similar overall enablement scores.
Interruptions increased consultations by an average of 2 minutes but reduced enablement only in biomedical consultations. Being vocationally trained did not affect enablement but duration of consultation was shorter. Non-members of the Royal College of General Practitioners consulted slightly faster than members, but enablement scores were similar. Doctors categorised as high for patient centredness consulted more slowly than others but had similar scores for enablement, the percentages of disadvantaged patients they saw, and patients who knew them well.
Mean enablement scores and mean durations of consultations were calculated for individual doctors. Enablement scores at doctor level (for English speaking patients) ranged from 1.1 to 5.3 and durations of consultations ranged from 3.8 to 14.4 minutes. Both were approximately normally distributed. The Spearman's rank correlation between doctors' order for mean enablement and mean duration of consultation was 0.38 for all 171 doctors with 50 valid enablement scores, 0.66 for 52 doctors with over 120 valid enablement scores, and 0.93 for seven doctors with over 170 valid enablement scores.
The percentage of patients seen with social and psychological problems varied by doctor from less than 10% to more than 70%. The percentage of patients indicating that they knew the doctor very well ranged from 0% (typically for locums, registrars, or very new partners) to 68%. The percentage of patients speaking other languages ranged from 0% to 89%. None of these influenced scores for mean enablement or mean duration of consultation at doctor level. This was confirmed by comparing doctor rank order for enablement for biomedical against social, psychological, and complex problems (r=0.55 for 30 in each category), patients who knew the doctor very well versus those who did not (r=0.60 for 40 in each category), and English speaking versus other language speaking patients (r=0.72 when restricted to doctors with at least 25 valid enablement scores in each language group).
Language of patients and doctors—We analysed data for both English speaking and other language speaking patients, seen by English speaking and other language speaking doctors, split by case mix for duration of consultation, enablement, and the percentage of patients who knew the doctor very well (data not shown). Other language speaking patients knew their doctors better than did English speaking patients (they generally attended smaller practices). They also reported significantly higher enablement (mean scores 4.5 versus 3.1) and shorter consultations (mean duration 7.1 versus 8.0 minutes) than did English speaking patients. This was particularly noticeable for biomedical presentations. Other language speaking doctors (all of whom spoke Asian languages) seemed to enable other language speaking patients with psychological problems less well than expected, probably because their consultations in this area were significantly shorter.
Sex of patients and doctors—We matched patients and doctors by sex (data not shown). Overall, 75.3% of patients seen by female doctors were female (5104 of 6779) compared with 59.1% seen by male doctors (7931 of 13 415). Case mix was similar for male and female doctors. Patients of male doctors knew them better. Enablement values for male and female doctors were comparable overall, but female doctors spent more time with their patients than did male doctors, particularly when the patient did not know the doctor well.
High and low enabling doctors—Doctors were divided into fourths on the basis of their enablement scores. Table 3 shows that doctors in the highest as against the lowest fourth had longer consultations, had more patients who knew them very well, worked with smaller lists, and enabled more in every analysis carried out.
List size and knowing the doctor well—Table 4 shows that the proportion of patients who knew the doctor well (a possible proxy for continuity) decreased as total list increased. The three largest practices seemed to go against that trend and were examined separately. In two of these practices the proportion of patients knowing the doctor well decreased as list size increased. The third practice operated an individual list system and had the same continuity as the best groups in the table. Their mean durations of consultations were, however, short—particularly for patients who did not know them well—and their mean enablement was low. Mean duration of consultation and mean enablement scores for patients who both did and did not know the doctor well show the persistent benefit of this attribute in terms of enablement (table 4). Consultations where the patient knew the doctor well were generally slightly longer than those where this was not the case.
At practice level, mean enablement scores ranged from 2.3 to 4.4 and mean durations of consultations ranged from 4.9 to 12.2 minutes. Correlations between ranks at practice level were not significant.
A multiple regression analysis was carried out at consultation level, with enablement as the outcome variable. Several covariates were identified as significant predictors of enablement, including knowing the doctor very well and duration of consultation. The overall predictive power of the model was, however, low (adjusted r2=0.037).
Feasibility and methods
The first purpose of our study was to develop previously used survey methods and to test their utility and acceptability in diverse geographical areas. We achieved a 38% take-up rate from our random sample of practices, this being higher for larger practices. The practices that declined had a similar demographic profile to those that accepted, but we cannot comment on whether the patterns of care they offered would also have been similar. We achieved a cross section of small, large, deprived, and non-deprived practices across the four participating areas, and these practices also differed in the ethnic mix of both patients and doctors.
By devising a single instrument that combined data both before and after consultation, we achieved an 80% completion rate overall. Because samples of 50 completed consultations per doctor were found to give generally stable enablement scores, we included in our main analyses all 171 of our 221 doctors who collected at least 50 completed consultations for English speaking patients. Doctors who fell below that target were mainly those with limited commitments or with larger than average numbers of patients speaking other languages.
As with any outcome measure it is hard to know whether reported enablement reflects true enablement The desire to please a familiar doctor and differences between cultural groups could create artificial differences. We found, however, that some doctors consistently enabled better than others, irrespective of all the aspects of case mix we studied. The finding of differences between English speaking and other language speaking patients was of particular interest and requires to be researched further with both quantitative and qualitative methods
Correlates with enablement
The second purpose of our study was to measure enablement and to identify its principal correlates. Doctors seemed to discriminate well about the needs of their patients, achieving similar overall enablement scores for most case mix presentations. Predictable differences in outcome and duration of consultation were related to age and sex of patients, to being added to surgery sessions without having appointments, and to having a consultation that was interrupted. The negative effect of not receiving a prescription when one was wanted was confirmed. The effect of ethnicity—reflected by languages spoken at home—in producing high enablement and short duration of consultation confirms the difficulty of judging quality in a multicultural society. The absence of an overall effect of social and psychological problems (a possible proxy for deprivation) on enablement or duration of consultation may be surprising, but it seems to be explained by the tendency for patients with social problems alone to receive shorter and less enabling consultations being compensated for by the fact that those with added psychological problems received longer and more enabling consultations. The substantial benefit of knowing the doctor very well gives an important contemporary message as does the fact that this benefit is progressively less likely to be found as practice size increases. The difficulty of modelling these variables into a single explanation for enablement—reflecting the large degree of variation within the enablement scores of individual doctors—confirms the complex nature of consultation in general practice. The ability of doctors to allocate time efficiently compounds the problem. Some doctors are, however, clearly less effective than others.
Work is currently in hand to develop a consultation quality measure at doctor and practice level and to compare results with other practice level performance indicators, which are based on routinely available NHS data.
If our ideas and methods gain broad support an incentive should be introduced into general practitioner contracts to reward doctors who spend more time at consultations, provide greater continuity of care, and both enable more patients and enable patients more. The fact that this triad of attributes, which both the public and the profession profess to value, are more likely to come together in smaller than larger practices should be heeded by those advocating bigger organisations as the way to improve general practice services.
We thank Dr T Jones (Oxfordshire Health Authority), Dr M Stern (Coventry Health Authority), Dr P Berrey (Lothian Health Board), Dr R Elton (University of Edinburgh), Dr M Pierce (Imperial College School of Medicine), and Dr S Campbell (National Primary Care Research and Development Centre) for regular discussions about the planning and implementation of this project, the doctors and practices who participated, and their managers and reception staff.
Contributors: JGRH, DJH, and MM initiated the study and wrote the protocol, along with GKF, and led the design of the analyses, which were carried out by JJW. All authors contributed to the design of research instruments, to the recruitment and briefing of practices, and to the interpretation of the results JGRH, DJH, MM, and JJW wrote the paper, which GF helped to edit and develop JGRH will act as guarantor for the paper.
Funding The work was supported by grants from the chief scientist's office at the Scottish Office Home and Health Department and the research and development directorates of Anglia and Oxford NHS, West Midlands NHS, and North Thames NHS.
Competing interests None declared.