Abstract
Background Recent research has shown that patients' expectations for prescriptions influence doctors' prescribing decisions, but little is known of the antecedents of these expectations.
Objectives To test earlier qualitative research about patients' views of medicines; to describe the demographic characteristics of those holding orthodox and unorthodox views of medicines; to investigate the relationship between patients' ideal and predicted expectations for prescriptions; and to determine the relative effects of attitudinal, demographic, organizational and illness variables on these expectations.
Design Questionnaire survey of patients consulting general practitioners.
Setting and participants A total of 544 patients and 15 doctors in four general practices.
Main variables studied Patients' attitudes to medicines; patients' demographic characteristics; organizational variables; aspects of patients' presenting problems.
Outcome measures Patients' ideal and predicted expectations for prescriptions.
Results Orthodox and unorthodox attitudes to medicines can be measured quantitatively, and ethnicity was the only demographic variable associated with both. Ideal and predicted expectations for prescriptions were closely related to each other but differed in their antecedents. Both types of expectations were associated with attitudinal, demographic, organizational and illness variables. Ideal expectations were influenced by orthodox and unorthodox attitudes to medicines, while predicted expectations were only influenced by orthodox attitudes.
Conclusions Future studies of patients' expectations for health services should distinguish between ideal and predicted expectations, and should consider the range of possible influences on these expectations. In particular, the effect of the organization and context of health services should be investigated.
Keywords: attitudes to medicines, expectations for prescriptions
Introduction
The literature on patients' expectations of health‐care has been more concerned with the links between expectations and outcomes than with the antecedents of these expectations. Studies of the relationship between expectations and satisfaction show inconsistent results, some suggesting that expectations influence satisfaction, while others do not. 1 However different studies have conceptualized expectations in different ways. In a review of the literature, Thompson and Sunol 2 identified four types of expectations: ideal, predicted, normative and unformed. Ideal expectations represent users' desired or preferred outcomes while predicted expectations refer to what they think will actually happen. Normative expectations represent what should or ought to happen. Unformed expectations occur when users are unable or unwilling to articulate their expectations, and highlight the fact that it may not always be possible or appropriate to try to measure users' expectations. This paper will employ Thompson and Sunol's classification, in particular their distinction between ideal and predicted expectations. Thompson and Sunol also discussed the factors which influence expectations, and classified these into three categories: personal, social and contextual. They argued that in order to understand the relative effects of these different kinds of influence, it is necessary to conduct research in different types of health‐care and with different groups of patients.
The significance of patients' expectations for health service researchers extends beyond their relevance or otherwise for satisfaction. Patients' expectations have been shown to influence clinicians' behaviour, both subjectively and objectively. In particular, patients' expectations for prescriptions have been shown to influence doctors' prescribing decisions. In studies of inappropriate and uncomfortable prescribing decisions, doctors claimed that they were influenced by patients' expectations. 3 , 4 Other studies have shown statistical associations between doctors' prescribing decisions and patients' expectations for prescriptions. 5 , 6 , 7 Thus the value of examining patients' expectations for prescriptions is twofold. First, prescribing decisions represent a particular context for teasing out the relative effects of personal, sociodemographic and contextual influences on patients' expectations. Secondly, the exploration of patients' expectations for prescriptions, and the variables determining these expectations, is likely to have significance for understanding prescribing decisions as well as the outcomes of these decisions.
A number of studies have investigated variables associated with patients' expectations for prescriptions in primary care settings. Virji and Britten 5 found that patients' attitudes to medicines were significantly associated with patients' expectations. Webb and Lloyd 8 showed that patients' expectations for prescriptions were significantly related to previous consultation forthe same problem, the nature and duration of the problem, but not to age and gender. InAustralia, Cockburn and Pit 6 found that patients' expectations for prescriptions were related to age but none of the other variables measured. In Germany, Himmel et al. 9 found that patients' expectations for prescriptions were not related to their attitudes towards medicines, the chronicity or otherwise of their condition, but were related to age. Macfarlane et al. 10 found that patients who both thought that their symptoms were caused by an infection, and who thought that an infection was present, were more likely to think that an antibiotic would help them. These five studies all measured expectations in different ways, only one of them 10 distinguishing between ideal and predicted expectations.
Thus the evidence about the antecedents ofpatients' expectations for prescriptions is inconsistent in relation to patients' attitudes tomedicines, their demographic characteristics especially age, and the nature of their symptoms. Part of the reason for this may be that different studies are measuring different aspects of patients' expectations. This suggests that these aspects need to be explicitly compared. In addition, little attention has been paid to the context of the consultation in terms of organizational features such as timing of the appointment. In the context of British general practice, whether the patient has an appointment or not is likely to affect their perceptions and expectations. Patients without appointments are often fitted in at the end of a booked surgery and may be well aware of time constraints. It is commonly believed that appointments are less available immediately before and after weekends, suggesting that patients' expectations may be influenced by the day of the week on which they see their doctor.
Two studies examining the influence of patients' attitudes to medicines were based on the same set of questions 5 , 9 but came to different conclusions, suggesting that this aspect in particular needs further investigation. A qualitative study of lay views of medicines suggested ways in which patients' views might be related to their ideal expectations for prescriptions. 11 In this interview based study, a distinction was made between orthodox and unorthodox accounts. Orthodox accounts were defined as being medically legitimated while unorthodox accounts were defined as being self legitimated. People giving orthodox accounts talked about medicines in a taken for granted fashion, while those giving unorthodox accounts showed an aversion to medicines. The former tended to be uncritical of doctors' prescribing habits while the latter complained about overprescribing. It seems likely that orthodox accounts would be associated with expectations for prescriptions rather more than unorthodox accounts. This study did not suggest that orthodox and unorthodox views were mutually exclusive, rather that people draw on different kinds of accounts in different situations.
The present study aimed to build on the earlier interview study 11 by examining these issues quantitatively amongst a larger sample of patients consulting their general practitioner (GP). In particular, it aimed to test the findings of the earlier qualitative research using self‐completion questionnaires; to describe the demographic characteristics of those who held orthodox and unorthodox attitudes; to investigate the relationship between patients' ideal and predicted expectations for prescriptions; and todetermine the relative effects of attitudinal, demographic, organizational and illness variables on both ideal and predicted expectations for prescriptions.
Methods
The study was carried out in four general practices in South London using questionnaire data obtained from patients waiting to see their GP. A three‐stage sampling procedure was used to identify practices, surgery sessions and patients. Practices were chosen using theoretical sampling (see below), surgery sessions were chosen using time sampling, and the patient sample was a census of all patients attending these sessions. The four practices were chosen to represent high and low prescribing levels and high and low levels of deprivation, relative to other practices within the same Health Authority area. A surgery session was defined as the list of all patients seen in the practice during the periods designated as morning or evening surgery. From within the chosen practices, surgery sessions were chosen to represent different days of the week, times of day, and appointment or non‐appointment sessions. The patient sample consisted of all patients seen in the chosen surgery sessions except those who had already completed a questionnaire. All those unwilling or unable to fill in a questionnaire while waiting to see the doctor were also excluded. The questionnaire enquired about the patient's symptoms, prior self medication, prescription exemption status, whether they hoped for a prescription and whether they actually expected to get one. Prescription exemption status identifies those people whose medicines are free at the point of use. Patients were also asked to respond to 20attitudinal items about drugs and medicines, in a Likert format. The pre‐coded responseswere ‘agree’, ‘neither agree nor disagree’, and ‘disagree’. These items were based on the earlier qualitative research 11 and represented separate aspects of orthodox and unorthodox accounts as well as common aspects. Lastly, patients were asked for their demographic details, including gender, age, ethnic group, marital status, housing tenure, educational attainment and employment situation. It was emphasized that neither practice staff nor GPs would see the completed questionnaires, and patients were asked to return them in the sealed envelopes. The patient's appointment status and the day of the week were also recorded.
The sample size calculation was based on thefollowing assumptions: that orthodox and unorthodox attitudes were mutually exclusive; 75% of the sample would hold orthodox attitudes and 25% would hold unorthodoxattitudes; and 55% of those with orthodox attitudes would expect a prescription compared with 40% of those with unorthodox attitudes. The sample size required to detect this 15% difference was 348 patients in the orthodox group and 116 in the unorthodox group, giving a total of 464 patients. It was decided to aim for 600 consultations to allow for missing data.
Symptom data and reasons for encounter were coded using the International Classification of Primary Care. 12 Ethnic group was coded according to the categories used for the 1991 decennial census.
The response rate for each practice was calculated from the surgery lists after excluding repeat attenders, those who did not attend and those known by the practice staff not to speak English. Census data on the demographic characteristics of the local population were obtained from the Office of Populations Censuses and Surveys for comparison purposes.
Data were analysed using the Stata software. 13 The chi‐squared test for association was used to analyse the relationship betweenpatients' ideal and predicted expectations. A principal components analysis was carried out on the attitudinal variables. Two principal components were extracted (and subsequently labelled as orthodox and unorthodox) and rotated using the varimax approach. The scores on the two principal components were derived so that they had a mean of zero and a standard deviation of one. By definition these components were derived such that they were uncorrelated with each other. Linear regression analyses were carried out of both principal components to identify the demographic variables which were associated with orthodox and unorthodox attitudes towards medicines. Logistic regression analyses were then carried out to determine whether attitudinal, demographic, organizational or illness variables were associated with ideal and predicted expectations. Both univariate analyses, examining the relationships of the outcome with each independent variable separately, and multivariate analyses, in which all variables were included in the regression model, were carried out. The samples for the regression analyses were restricted to those cases with non‐missing values on all independent variables for a given outcome. Sensitivity analyses for each regression were carried out in which missing data on any covariate were assigned non‐missing codes and included in the analyses. Larger coefficients in the regression analyses of attitudinal variables (Tables 4 and 5) indicate a greater tendency to have orthodox and unorthodox attitudes towards medicine, respectively, and larger odds ratios in the regression analyses of patients' expectations (Tables 7 and 8) indicate a greater tendency to hope for or expect to receive a prescription. In the tables coefficients and odds ratios for categorical variables are expressed relative to the base categories (for example ‘male’ is the base category for the variable gender).
The data from this study have a clustered, or hierarchical, structure in the sense that many patients were seen by each of the 15 GPs in one of four general practices. It is likely that responses from patients in the same practice or seeing the same practitioner would be correlated with each other. 14 , 15 This has implications for analysis, as standard statistical methods are invalid where responses are not independent of each other. Correlation between the responses of different patients seeing the same general practitioner was adjusted for using random effects models for the linear regression analyses 16 and the approach of marginal modelling using Generalized Estimating Equations (GEEs) with robustified standard errors was used to implement the logistic regression analyses. 17 An exchangeable correlation matrix was specified for the GEEs models. Ideally one would also allow for correlation between subjects' responses within the same general practice, but there were too few practices to make use of the above techniques feasible. Analyses in which the effect of general practice was included in the model as a fixed effect yielded essentially the same findings as the main analyses. The Wald test was used toobtain significance levels in the regression models.
The procedure used to implement the principal components analysis of the attitudinal items does not allow for correlation between subjects consulting the same general practitioner. Software for implementing this type of analysis hasnot yet been made widely available to the research community. In order to assess the potential effect on the results, separate principal components analyses were carried out for each general practice. This was done to gauge the extent to which the results varied across practices. The ratio of patients to attitudinal items was too small to run separate analyses for each practitioner.
Results
A total of 544 questionnaires were obtained, representing a response rate of 64.8%. A comparison of the age and gender of respondentsand non‐respondents revealed that non‐respondents were more likely to be male and over the age of 44 years. The characteristics of the patient sample were also compared with those of the local resident population from the 1991 census. The study population contained a smaller proportion of men, people over 44 years of age, single people, and those who were economically active. However, housing tenure and ethnic group of the study sample were similar to those of the local resident population. Table 1 shows the distributions of the variables, other than the attitudinal variables, used in the subsequent analyses. They represent demographic variables (gender, age, marital status, housing tenure, school leaving age, ethnic group, employment status, and prescription status), organizational variables (day of the week, appointment status) and aspects of the patient's illness or presenting problem (symptoms, reason for encounter, self medication).
Table 1.
Characteristics of patients and consultations
| Variable | Category | Number | Percentage |
|---|---|---|---|
| Gender | Male | 164 | 30.1 |
| Female | 380 | 69.9 | |
| Age | Up to 24 years | 89 | 16.5 |
| 25–34 | 182 | 33.8 | |
| 35–44 | 105 | 19.5 | |
| 45–54 | 53 | 9.9 | |
| 55–64 | 46 | 8.6 | |
| 65 years and over | 63 | 11.7 | |
| Marital status | Married or cohabiting | 282 | 53.6 |
| Others | 244 | 46.4 | |
| Housing tenure | Tenant (council or other) | 300 | 58.5 |
| Owner occupier | 186 | 36.3 | |
| Others | 27 | 5.3 | |
| Age left school | 16 (or earlier) | 291 | 57.1 |
| After 16 | 219 | 42.9 | |
| Ethnic group | White | 366 | 70.4 |
| Black Caribbean | 45 | 8.7 | |
| Black African | 52 | 10.0 | |
| Others | 57 | 11.0 | |
| Employment status | Working (full or part time) | 231 | 43.8 |
| Others | 296 | 56.2 | |
| Prescription exemption | Not exempt | 203 | 37.9 |
| status | Partially exempt | 48 | 9.0 |
| Totally exempt | 285 | 53.2 | |
| Day of week | Monday | 189 | 34.7 |
| Tuesday | 170 | 31.3 | |
| Thursday | 128 | 23.5 | |
| Friday | 57 | 10.5 | |
| Appointment status | Had appointment | 333 | 61.2 |
| Non‐appointment or extra | 211 | 38.8 | |
| Symptoms | General | 42 | 7.7 |
| Digestive | 45 | 8.3 | |
| Circulatory | 37 | 6.8 | |
| Musculoskeletal | 65 | 11.9 | |
| Psychological and social | 35 | 6.4 | |
| Respiratory | 117 | 21.5 | |
| Skin | 50 | 9.2 | |
| Pregnancy and child‐bearing | 38 | 7.0 | |
| Others | 115 | 21.1 | |
| Reason for encounter | Symptoms and complaints | 292 | 53.7 |
| Diagnoses and disease | 176 | 32.4 | |
| Others | 76 | 14.0 | |
| Self medication | Yes | 279 | 52.9 |
| No | 248 | 47.1 |
Note: Totals vary because of variable non‐response to individual questions.
Attitudes towards medicines
The distributions of the 20 attitudinal variables are shown in Table 2. The percentage agreement with each item varied from 86% for the statement ‘I prefer not to take any medicine if I can avoid it’ to 12% for the statement ‘If the doctor does not give me a prescription I feel I have wasted his (or her) time’.
Table 2.
Summary of patients' responses to the attitudinal statements on medicines
| Statement | Agree | Neither agree nor disagree | Disagree | n |
|---|---|---|---|---|
| Medicines help your body recover from illness | 395 (74.4%) | 119 (22.4%) | 17 (3.2%) | 531 |
| I would be happy to take a medicine over a long period of time | 128 (24.2%) | 71 (13.4%) | 329 (62.3%) | 528 |
| All medicines have side‐effects | 153 (29.1%) | 151 (28.8%) | 221 (42.1%) | 525 |
| If you have a cold or a cough it is best to get an antibiotic to get rid of it | 138 (26.2%) | 95 (18.1%) | 293 (55.7%) | 526 |
| I am worried about getting addicted to medicines | 218 (41.4%) | 93 (17.6%) | 216 (41.0%) | 527 |
| If you take too many antibiotics they won't work when you really need them | 375 (70.1%) | 98 (18.3%) | 62 (11.6%) | 535 |
| If I'm feeling ill I like to take medicine | 140 (26.2%) | 107 (20.0%) | 287 (53.7%) | 534 |
| I prefer to take a natural remedy and not a medicine | 210 (40.9%) | 177 (34.4%) | 127 (24.7%) | 514 |
| Modern medicines have improved people's health | 368 (69.4%) | 127 (24.0%) | 35 (6.6%) | 530 |
| Doctors give too many prescriptions when they are not really necessary | 107 (20.0%) | 188 (35.2%) | 239 (44.8%) | 534 |
| I prefer not to take any medicine if I can avoid it | 461 (86.3%) | 44 (8.2%) | 29 (5.4%) | 534 |
| You cannot get well if you do not take a medicine | 88 (16.6%) | 108 (20.3%) | 335 (63.1%) | 531 |
| Some medicines are poisonous | 285 (54.4%) | 154 (29.4%) | 85 (16.2%) | 524 |
| Antibiotics are the only cure for infections | 188 (35.6%) | 131 (24.8%) | 209 (39.6%) | 528 |
| I always take as small a dose as possible | 303 (58.0%) | 130 (24.9%) | 89 (17.0%) | 522 |
| If you leave a cold for more than a week it can turn into something nasty | 161 (30.1%) | 117 (21.9%) | 257 (48.0%) | 535 |
| Antibiotics interfere with the body's natural ability to fight infection | 170 (32.3%) | 181 (34.3%) | 176 (33.4%) | 527 |
| Most illnesses cure themselves without having to go to the doctors | 185 (34.5%) | 150 (28.0%) | 201 (37.5%) | 536 |
| If the doctor does not give me a prescription I feel I have wasted his (or her) time | 62 (11.6%) | 56 (10.5%) | 417 (77.9%) | 535 |
| When you are ill you should always take a medicine | 82 (15.3%) | 98 (18.2%) | 357 (66.5%) | 537 |
The results of the principal components analysis are shown in Table 3. Using the scree test criterion 18 only the first two principal components represented meaningful variation amongst the attitudinal items. Further, only these two components were interpretable. The attitudinal items are separated into two groups: those having higher loadings on the first component, labelled orthodox attitudes; and those having higher loadings on the second component, labelled unorthodox attitudes. Items of salience to the orthodox component were those endorsing the use of medicines, while those of salience to the unorthodox component reflected the range of negative attitudes to drugs, criticism of doctors' prescribing habits and modern medicine as well as preference for natural treatments. This range of attitudes corresponds to the unorthodox accounts identified in the qualitative study. 11 The percentage of the total variation explained by these two components was 28.3%. The components were independent of each other, so that it was possible for an individual to score highly on both components.
Table 3.
Principal components analysis of statements measuring attitudes towards medicines – rotated principal component loadings
| Loadings on principal components of rotated matrix | Component 1 Orthodox | Component 2 Unorthodox |
|---|---|---|
| When you are ill you should always take a medicine | 0.77 | −0.13 |
| You cannot get well if you do not take a medicine | 0.73 | −0.06 |
| If you leave a cold for more than a week it can turn into something nasty | 0.66 | 0.09 |
| If I'm feeling ill I like to take medicine | 0.61 | −0.26 |
| Antibiotics are the only cure for infections | 0.57 | −0.17 |
| If you have a cold or a cough it is best to get an antibiotic to get rid of it | 0.57 | −0.20 |
| If the doctor does not give me a prescription I feel I have wasted his (or her) time | 0.56 | −0.12 |
| I always take as small a dose as possible | 0.24 | 0.17 |
| Antibiotics interfere with the body's natural ability to fight infection | 0.07 | 0.56 |
| If you take too many antibiotics they won't work when you really need them | −0.03 | 0.50 |
| I prefer to take a natural remedy and not a medicine | −0.23 | 0.50 |
| I am worried about getting addicted to medicines | 0.29 | 0.44 |
| I would be happy to take a medicine over a long period of time | 0.33 | −0.44 |
| Medicines help your body recover from illness | 0.20 | −0.42 |
| Modern medicines have improved people's health | 0.03 | −0.41 |
| Doctors give too many prescriptions when they are not really necessary | −0.17 | 0.40 |
| All medicines have side‐effects | 0.26 | 0.38 |
| Some medicines are poisonous | −0.02 | 0.35 |
| I prefer not to take any medicine if I can avoid it | −0.14 | 0.32 |
| Most illnesses cure themselves without having to go to the doctors | −0.12 | 0.30 |
| n = 429 | ||
| Principal component | Percentage of variation explained | Cumulative percentage |
| 1 | 16.4 | 16.4 |
| 2 | 11.9 | 28.3 |
The principal components analyses carried out separately for each practice provided broadly similar results to the main analysis. Again, using the scree test, a two factor solution was optimal in three practices and a one factor solution was optimal in the fourth. Generally the attitudinal items with high loadings on the two components were the same as those with high loadings in the main analysis although there was slightly less consistency in the loadings for the most important items onthe unorthodox factor. The two extracted factors in each of the within‐practice analyses explained approximately the same amount of variation as in the main analysis. These findings give some confidence that the results of the principal components analysis are not seriouslyaffected by the lack of adjustment for clustering.
Tables 4 and 5 show the results of the univariate and multivariate linear regression analyses of orthodox and unorthodox attitudes against the demographic variables. Table 4 shows that in the multivariate analysis, only age, school leaving age and ethnic group were significantly related to orthodox attitudes. Those with high orthodox scores were patients over 65 years of age, early school leavers and black Africans. Table 5 shows that in the multivariate analysis of patients' unorthodox attitudes, only ethnic group remained significant. Black African patients were the most likely to have unorthodox attitudes.
Table 4.
Linear regression of orthodox attitudes to medicine score on demographic characteristics
| Univariate results | Multivariate results | ||||||
|---|---|---|---|---|---|---|---|
| Variable | Category | Coefficient | (95% CI) | P‐value | Coefficient | (95% CI) | P‐value |
| Gender | Male | 0 | 0.96 | 0 | 0.73 | ||
| Female | 0.005 | −0.217 to 0.227 | 0.035 | −0.162 to 0.231 | |||
| Age | Up to 24 years | 0 | <0.001 | 0 | <0.001 | ||
| 25–34 | 0.011 | −0.264 to 0.287 | 0.042 | −0.223 to 0.308 | |||
| 35–44 | 0.010 | −0.301 to 0.321 | 0.078 | −0.227 to 0.382 | |||
| 45–54 | 0.140 | −0.237 to 0.517 | 0.318 | −0.044 to 0.679 | |||
| 55–64 | 0.817 | 0.420 to 1.213 | 0.917 | 0.541 to 1.293 | |||
| 65 years and over | 1.086 | 0.708 to 1.464 | 1.118 | 0.752 to 1.485 | |||
| Marital status | Married/cohabiting | 0 | 0.10 | 0 | 0.57 | ||
| Others | 0.169 | −0.031 to 0.368 | 0.054 | −0.134 to 0.241 | |||
| Housing tenure | Tenant | 0 | 0.049 | 0 | 0.57 | ||
| Owner occupier | −0.269 | −0.488 to –0.049 | −0.106 | −0.321 to 0.108 | |||
| Others | −0.004 | −0.443 to 0.435 | 0.051 | −0.332 to 0.434 | |||
| Age left school | 16 (or earlier) | 0 | <0.001 | 0 | 0.002 | ||
| 17 (or later) | −0.372 | −0.569 to −0.176 | −0.293 | −0.475 to −0.110 | |||
| Ethnic group | White | 0 | <0.001 | 0 | <0.001 | ||
| Black Caribbean | 0.055 | −0.316 to 0.426 | 0.080 | −0.252 to 0.413 | |||
| Black African | 0.866 | 0.486 to 1.246 | 1.052 | 0.701 to 1.402 | |||
| Others | 0.567 | 0.265 to 0.870 | 0.763 | 0.488 to 1.038 | |||
| Employment status | Working (FT/PT) | 0 | <0.001 | 0 | 0.87 | ||
| Others | 0.332 | 0.136 to 0.527 | 0.018 | −0.200 to 0.237 | |||
| Prescription | Not exempt | 0 | <0.001 | 0 | 0.58 | ||
| Exemption status | Partially exempt | 0.100 | −0.251 to 0.451 | 0.121 | −0.202 to 0.443 | ||
| Totally exempt | 0.452 | 0.249 to 0.655 | 0.116 | −0.122 to 0.353 |
n = 381.
Table 5.
Linear regression of unorthodox attitudes to medicine score on demographic characteristics
| Univariate results | Multivariate results | ||||||
|---|---|---|---|---|---|---|---|
| Variable | Category | Coefficient | (95% CI) | P‐value | Coefficient | (95% CI) | P‐value |
| Gender | Male | 0 | 0.41 | 0 | 0.34 | ||
| Female | 0.095 | −0.129 to 0.319 | 0.112 | −0.117 to 0.341 | |||
| Age | Up to 24 years | 0 | 0.26 | 0 | 0.71 | ||
| 25–34 | 0.192 | −0.110 to 0.495 | 0.153 | −0.158 to 0.465 | |||
| 35–44 | 0.223 | −0.116 to 0.562 | 0.199 | −0.157 to 0.556 | |||
| 45–54 | 0.258 | −0.157 to 0.672 | 0.290 | −0.135 to 0.714 | |||
| 55–64 | −0.112 | −0.547 to 0.323 | −0.016 | −0.458 to 0.426 | |||
| 65 years and over | −0.104 | −0.515 to 0.307 | 0.075 | −0.351 to 0.502 | |||
| Marital status | Married/cohabiting | 0 | 0.42 | 0 | 0.28 | ||
| Others | 0.084 | −0.121 to 0.289 | 0.122 | −0.098 to 0.342 | |||
| Housing tenure | Tenant | 0 | 0.31 | 0 | 0.27 | ||
| Owner occupier | 0.047 | −0.166 to 0.260 | 0.036 | −0.205 to 0.278 | |||
| Others | 0.345 | −0.100 to 0.789 | 0.369 | −0.075 to 0.814 | |||
| Age left school | 16 (or earlier) | 0 | 0.13 | 0 | 0.64 | ||
| 17 (or later) | 0.160 | −0.045 to 0.365 | 0.051 | −0.162 to 0.265 | |||
| Ethnic group | White | 0 | 0.02 | 0 | 0.03 | ||
| Black Caribbean | 0.026 | −0.363 to 0.416 | −0.019 | −0.407 to 0.370 | |||
| Black African | 0.498 | 0.102 to 0.894 | 0.481 | 0.072 to 0.890 | |||
| Others | 0.346 | 0.028 to 0.665 | 0.339 | 0.018 to 0.660 | |||
| Employment status | Working (FT/PT) | 0 | 0.03 | 0 | 0.31 | ||
| Others | −0.222 | −0.425 to −0.020 | −0.132 | −0.389 to 0.125 | |||
| Prescription | Not exempt | 0 | 0.18 | 0 | 0.68 | ||
| Exemption status | Partially exempt | 0.003 | −0.367 to 0.373 | −0.022 | −0.401 to 0.358 | ||
| Totally exempt | −0.191 | −0.405 to 0.023 | −0.123 | −0.403 to 0.156 |
n = 381.
Patients' expectations for prescriptions
Responses to the questions on patients' ideal and predicted expectations revealed that 67.3% (354/526) hoped to receive a prescription and 65.1% (328/504) expected to receive a prescription. These figures are higher than those obtained from recent comparable studies. The proportion of patients expecting prescriptions was 47% in Cockburn and Pit's 6 study and 51% in Webb and Lloyd's 8 . Higher estimates (65 and 72%) have been obtained in studies of patients consulting with respiratory symptoms. 10 , 19
There was a significant association between patients' ideal and predicted expectations (P < 0.001, Table 6). Of the patients who hoped to receive a prescription 91% expected to receive one. In contrast only 12% of those who did not hope to receive a prescription actually expected to receive one. Only 10% (49/495) of the sample expected an outcome that they did not hope for, whether it was to receive a prescription or not.
Table 6.
Patients' ideal and predicted expectations
| Patient expects to receive a prescription | |||
|---|---|---|---|
| Yes | No | Total | |
| Patient hopes to receive prescription | |||
| Yes | 303 ( 91.3%) | 29 (8.7%) | 332 ( 100%) |
| No | 20 (12.3%) | 143 ( 87.7%) | 163 ( 100%) |
Chi‐squared statistic = 300 . 9 , d.f. = 1 , P < 0.001 .
Table 7 shows the attitudinal, demographic, organizational and illness variables which were related to ideal expectations in the univariate and multivariate analyses. In the multivariate analysis both attitudinal variables, age, ethnic group, prescription exemption status, day of theweek, appointment status and symptoms remained significant. Patients with high orthodox scores were more likely to hope for prescriptions while those with high unorthodox scores were less likely to hope for prescriptions. Apart from the attitudinal variables, patients most likely to hope for prescriptions were those aged 45–54 years, Black Africans, those totally exempt from prescription charges, those consulting on Fridays, those without appointments, and those with digestive symptoms.
Table 7.
Logistic regression of patients' ideal expectations
| Univariate results | Multivariate results | ||||||
|---|---|---|---|---|---|---|---|
| Variable | Category | Odds ratio | (95% CI) | P‐value | Odds ratio | (95% CI) | P‐value |
| Orthodox attitudes | Score on principal | 1.76 | 1.40, 2.22 | <0.001 | 1.81 | 1.20, 2.75 | 0.005 |
| component | |||||||
| Unorthodox attitudes | Score on principal | 0.73 | 0.57, 0.93 | 0.01 | 0.70 | 0.54, 0.92 | 0.01 |
| component | |||||||
| Gender | Male | 1 | 0.25 | 1 | 0.06 | ||
| Female | 1.22 | 0.87, 1.71 | 1.47 | 0.99, 2.17 | |||
| Age | Up to 24 years | 1 | <0.001 | 1 | <0.001 | ||
| 25–34 | 1.13 | 0.63, 2.03 | 2.39 | 1.46, 3.92 | |||
| 35–44 | 0.74 | 0.46, 1.18 | 1.39 | 0.77, 2.50 | |||
| 45–54 | 1.76 | 0.90, 3.44 | 3.01 | 1.03, 8.80 | |||
| 55–64 | 1.11 | 0.48, 2.59 | 1.08 | 0.36, 3.26 | |||
| 65 years and over | 2.36 | 0.99, 5.66 | 1.18 | 0.33, 4.25 | |||
| Marital status | Married/cohabiting | 1 | 0.051 | 1 | 0.30 | ||
| Others | 1.53 | 1.00, 2.35 | 1.40 | 0.74, 2.63 | |||
| Housing tenure | Tenant | 1 | 0.10 | 1 | 0.21 | ||
| Owner occupier | 0.60 | 0.37, 0.96 | 0.83 | 0.41, 1.66 | |||
| Others | 1.13 | 0.40, 3.21 | 2.48 | 0.89, 6.89 | |||
| Age left school | 16 (or earlier) | 1 | 0.03 | 1 | 0.60 | ||
| 17 (or later) | 0.56 | 0.33, 0.93 | 0.83 | 0.41, 1.68 | |||
| Ethnic group | White | 1 | 0.26 | 1 | 0.02 | ||
| Black Caribbean | 0.61 | 0.31, 1.19 | 0.36 | 0.15, 0.91 | |||
| Black African | 1.44 | 0.71, 2.89 | 1.29 | 0.62, 2.69 | |||
| Others | 0.84 | 0.42, 1.72 | 0.69 | 0.29, 1.65 | |||
| Employment status | Working (FT/PT) | 1 | 0.41 | 1 | 0.27 | ||
| Others | 1.30 | 0.69, 2.44 | 0.70 | 0.38, 1.31 | |||
| Prescription | Not exempt | 1 | 0.006 | 1 | <0.001 | ||
| Exemption status | Partially exempt | 0.57 | 0.27, 1.20 | 0.29 | 0.13, 0.64 | ||
| Totally exempt | 2.09 | 1.15, 3.80 | 2.02 | 1.03, 3.97 | |||
| Day of week | Monday | 1 | 0.002 | 1 | 0.002 | ||
| Tuesday | 0.54 | 0.34, 0.85 | 0.76 | 0.46, 1.27 | |||
| Thursday | 0.65 | 0.33, 1.26 | 0.89 | 0.48, 1.67 | |||
| Friday | 2.61 | 0.78, 8.71 | 2.84 | 0.87, 9.33 | |||
| Appointment status | Had appointment | 1 | 0.002 | 1 | 0.02 | ||
| Non‐appointment/extra | 1.57 | 1.17, 2.11 | 1.60 | 1.09, 2.35 | |||
| Symptoms | General | 1 | <0.001 | 1 | <0.001 | ||
| Digestive | 3.38 | 0.97, 11.76 | 6.14 | 2.16, 17.43 | |||
| Circulatory | 1.36 | 0.62, 3.02 | 2.61 | 0.75, 9.01 | |||
| Musculoskeletal | 1.10 | 0.45, 2.65 | 0.84 | 0.35, 2.03 | |||
| Psychological/social | 1.63 | 0.59, 4.47 | 1.46 | 0.47, 4.57 | |||
| Respiratory | 2.95 | 1.41, 6.18 | 3.65 | 1.71, 7.76 | |||
| Skin | 3.01 | 1.02, 8.94 | 3.10 | 1.09, 8.83 | |||
| Pregnancy and | 0.68 | 0.25, 1.88 | 1.01 | 0.30, 3.46 | |||
| child‐bearing | |||||||
| Others | 1.57 | 0.79, 3.11 | 2.50 | 1.10, 5.69 | |||
| Reason for encounter | Symptoms/complaints | 1 | 0.08 | 1 | 0.21 | ||
| Diagnoses/disease | 0.85 | 0.56, 1.30 | 0.82 | 0.48, 1.40 | |||
| Others | 0.52 | 0.29, 0.95 | 0.56 | 0.27, 1.15 | |||
| Self medication | Yes | 1 | 0.005 | 1 | 0.12 | ||
| No | 0.55 | 0.36, 0.84 | 0.67 | 0.41, 1.10 |
n = 358.
Table 8 shows the results of the same regression analysis carried out for the variable representing patients' predicted expectations. The variables significant in the multivariate analyses were orthodox attitudes, gender, marital status, day of the week, symptoms, reason for encounter and self medication. Thus the patients most likely to expect to receive a prescription were those with high scores on the orthodox scale, women, those not married or cohabiting, those consulting on Fridays, those with digestive symptoms, those consulting for symptomsrather than other reasons for encounter, and those who had previously self medicated for the index (or consulting) symptom. The variables remaining significant in analyses of both patients' ideal and predicted expectations were orthodox attitudes, day of the week and symptoms.
Table 8.
Logistic regression of patients' predicted expectations
| Univariate results | Multivariate results | ||||||
|---|---|---|---|---|---|---|---|
| Variable | Category | Odds ratio | (95% CI) | P‐value | Odds ratio | (95% CI) | P‐value |
| Orthodox attitudes | Score on principal component | 1.52 | 1.12, 2.06 | 0.007 | 1.43 | 1.06, 1.95 | 0.02 |
| Unorthodox attitudes | Score on principal component | 0.81 | 0.63, 1.04 | 0.10 | 0.80 | 0.63, 1.01 | 0.06 |
| Gender | Male | 1 | 0.11 | 1 | 0.002 | ||
| Female | 1.39 | 0.93, 2.10 | 1.64 | 1.20, 2.25 | |||
| Age | Up to 24 years | 1 | 0.10 | 1 | 0.73 | ||
| 25–34 | 0.77 | 0.45, 1.31 | 1.10 | 0.68, 1.80 | |||
| 35–44 | 0.57 | 0.27, 1.18 | 0.88 | 0.40, 1.94 | |||
| 45–54 | 1.21 | 0.48, 3.03 | 1.64 | 0.46, 5.84 | |||
| 55–64 | 1.29 | 0.33, 4.97 | 1.46 | 0.36, 5.99 | |||
| 65 years and over | 1.53 | 0.74, 3.15 | 0.99 | 0.38, 2.60 | |||
| Marital status | Married/cohabiting | 1 | 0.03 | 1 | 0.04 | ||
| Others | 1.60 | 1.04, 2.46 | 1.69 | 1.03, 2.79 | |||
| Housing tenure | Tenant | 1 | 0.51 | 1 | 0.99 | ||
| Owner occupier | 0.76 | 0.48, 1.21 | 0.96 | 0.51, 1.80 | |||
| Others | 0.92 | 0.32, 2.67 | 0.97 | 0.31, 2.98 | |||
| Age left school | 16 (or earlier) | 1 | 0.01 | 1 | 0.30 | ||
| 17 (or later) | 0.55 | 0.34, 0.89 | 0.69 | 0.35, 1.39 | |||
| Ethnic group | White | 1 | 0.97 | 1 | 0.78 | ||
| Black Caribbean | 1.11 | 0.55, 2.23 | 1.02 | 0.52, 2.03 | |||
| Black African | 1.20 | 0.54, 2.68 | 1.35 | 0.53, 3.40 | |||
| Others | 1.07 | 0.52, 2.22 | 1.45 | 0.69, 3.07 | |||
| Employment status | Working (FT/PT) | 1 | 0.59 | 1 | 0.64 | ||
| Others | 1.20 | 0.62, 2.35 | 0.82 | 0.36, 1.88 | |||
| Prescription | Not exempt | 1 | 0.21 | 1 | 0.26 | ||
| Exemption status | Partially exempt | 0.93 | 0.42, 2.04 | 0.66 | 0.27, 1.59 | ||
| Totally exempt | 1.60 | 0.91, 2.82 | 1.50 | 0.71, 3.16 | |||
| Day of week | Monday | 1 | <0.001 | 1 | 0.002 | ||
| Tuesday | 0.37 | 0.24, 0.59 | 0.47 | 0.29, 0.78 | |||
| Thursday | 0.72 | 0.44, 1.19 | 1.01 | 0.49, 2.08 | |||
| Friday | 2.55 | 0.85, 7.68 | 2.86 | 0.73, 11.16 | |||
| Appointment status | Had appointment | 1 | 0.04 | 1 | 0.17 | ||
| Non‐appointment/extra | 1.64 | 1.04, 2.59 | 1.38 | 0.87, 2.17 | |||
| Symptoms | General | 1 | <0.001 | 1 | <0.001 | ||
| Digestive | 12.35 | 3.14, 48.58 | 17.55 | 5.35, 57.59 | |||
| Circulatory | 1.89 | 0.69, 5.22 | 2.21 | 0.61, 8.03 | |||
| Musculoskeletal | 1.12 | 0.52, 2.42 | 0.89 | 0.40, 1.96 | |||
| Psychological/social | 2.78 | 0.97, 7.98 | 2.55 | 0.77, 8.40 | |||
| Respiratory | 3.35 | 1.51, 7.42 | 3.15 | 1.42, 6.95 | |||
| Skin | 2.77 | 0.82, 9.39 | 2.16 | 0.66, 7.03 | |||
| Pregnancy and | 0.88 | 0.30, 2.52 | 0.96 | 0.26, 3.59 | |||
| child‐bearing | |||||||
| Others | 2.41 | 1.12, 5.18 | 2.94 | 1.15, 7.49 | |||
| Reason for encounter | Symptoms/complaints | 1 | 0.08 | 1 | 0.03 | ||
| Diagnoses/disease | 0.94 | 0.65, 1.37 | 0.87 | 0.53, 1.41 | |||
| Others | 0.50 | 0.27, 0.92 | 0.42 | 0.22, 0.81 | |||
| Self medication | Yes | 1 | <0.001 | 1 | 0.003 | ||
| No | 0.46 | 0.31, 0.69 | 0.52 | 0.34, 0.81 |
n = 347.
Sensitivity analyses, in which missing values were recoded as non‐missing, increased the sample size for the regressions of the attitudinal variables by 12% and the regressions of expectations by 15%. The results of these analyses were essentially the same as those given in 4, 5, 7, 8, suggesting that our treatment of missing values did not alter the results.
Conclusions
The results from this study suggest that it is possible to measure orthodox and unorthodox attitudes towards medicines quantitatively, and that they are independently associated with expectations for prescriptions. Ethnicity was the only demographic variable significantly associated with both orthodox and unorthodox attitudes. Black African patients were the most likely to have both high orthodox and unorthodox scores. Despite the fact that those with high orthodox scores were more likely to hope for a prescription, and those with high unorthodox scores were less likely to hope for a prescription, the two sets of scores were independent. These results are consistent with other studies demonstrating the ambivalent nature of people's views of medicines 20 which could usefully be explored further.
Patients' expectations for prescriptions were higher in this study than in other previous studies. 6 , 8 As the study was carried out in an inner city area characterized by high levels of deprivation and low levels of prescribing, the results may not be generalizable to other areas. Given the need to research patients' expectations in a range of different health care settings and contexts 2 this study can make a contribution to such an endeavour.
This study has compared patients' ideal and predicted expectations for prescriptions. While the two were closely related, there were differences between them, particularly in their antecedents. In broad terms, both were influenced by attitudinal, demographic, organisational and illness variables, in particular orthodox attitudes, the day of the week and symptoms. The differences were in the details, with unorthodox attitudes for example having a stronger relationship with ideal than with predicted expectations. Similarly age was associated with ideal expectations and gender was associated with predicted expectations, although it should be recalled that the study sample over represented women and people under 44 years of age. Prescription exemption status had more salience for ideal expectations than for predicted expectations, while the converse was true for previous self medication. The interpretation of these differences must be made with caution, but they do suggest that there are differences in expectations which are important to patients on the basis of which they do not expect to be treated differently by doctors. For example, the results suggest that patients are less likely to expect their unorthodox attitudes to be reflected in doctors' actual prescribing decisions than their orthodox attitudes. The results also suggest that women expect to be treated differently in terms of receiving prescriptions, although ideal expectations are not gender related. Conversely, although ideal expectations were age related, patients' predicted expectations suggest that they do not expect doctors to treat them differently on the basis of age. The fact that age was associated with ideal but not predicted expectations may help explain the inconsistent results of previous studies. Prescription exemption status is dependent on age, health status and ability to pay. Its salience for patients' ideal but not predicted expectations suggests that although morbidity and deprivation may be important for patients, they do not expect these considerations to influence doctors' prescribing decisions.
This paper has demonstrated the differences between patients' ideal and predicted expectations for a prescription in one particular context and the various kinds of factors which influence them. It has shown that attitudinal, demographic, organizational and illness variables are all significant precursors of both types of expectation. In particular, organizational aspects of primary care such as the day of the week and the patient's appointment status influence patients' expectations. Future studies of patients' expectations for health services should distinguish between these different types of expectations, and the range of possible variables which may shape them. Other organizational variables not measured in this study, such as the waiting time for appointments or continuity of care, may also influence patients' expectations.
Acknowledgements
This study was partly funded by the project grant scheme of the NHSManagementExecutive South Thames. We are very grateful to the general practitioners, practice managers and reception staff in the four practices concerned for their willing co‐operation. We are also very grateful to the individuals who agreed to complete the patient questionnaires. Mr Ukoumunne is supported by an MRC Special Training Fellowship.
References
- 1. Staniszewska S, Ahmed L. The concepts of expectation and satisfaction: do they capture the way patients evaluate their care? Journal of Advanced Nursing , 1999; 29: 364–372. [DOI] [PubMed] [Google Scholar]
- 2. Thompson AGH, Sunol R. Expectations as determinants of patient satisfaction: concepts, theory and evidence. International Journal for Quality in Health Care , 1995; 7: 127–141. [DOI] [PubMed] [Google Scholar]
- 3. Schwartz RK, Soumerai SB, Avorn J. Physician motivations for non scientific drug prescribing. Social Science and Medicine , 1989; 28: 577–582. [DOI] [PubMed] [Google Scholar]
- 4. Bradley CP. Uncomfortable prescribing decisions: a critical incident study. British Medical Journal , 1992; 304: 294–296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Virji A, Britten N. A study of the relationship between patients' attitudes and doctors' prescribing. Family Practice , 1991; 8: 314–319. [DOI] [PubMed] [Google Scholar]
- 6. Cockburn J, Pit S. Prescribing behaviour in clinical practice: patients' expectations and doctors' perceptions of patients' expectations – a questionnaire study. British Medical Journal , 1997; 315: 520–523. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Britten N, Ukoumunne O. The influence of patients' hopes of receiving a prescription on doctors' perceptions and the decision to prescribe: a questionnaire survey. British Medical Journal , 1997; 315: 1506–1510. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Webb S, Lloyd M. Prescribing and referral in general practice: a study of patients' expectations and doctors' actions. British Journal of General Practice , 1994; 44: 165–169. [PMC free article] [PubMed] [Google Scholar]
- 9. Himmel W, Lippert‐Urbanke E, Kochen MM. Are patients more satisfied when they receive a prescription? Scandanavian Journal of Primary Health Care , 1997; 15: 118–122. [DOI] [PubMed] [Google Scholar]
- 10. Macfarlane J, Holmes W, Macfarlane R, Britten N. Influence of patients' expectations on antibiotic management of acute lower respiratory tract illness in general practice: questionnaire study. British Medical Journal , 1997; 315: 1211–1214. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Britten N. Lay views of drugs and medicines: orthodox and unorthodox accounts In: Willams SJ, Calnan M. (eds) Modern Medicine: Lay Perspectives and Experiences. London: UCL Press, 1996. [Google Scholar]
- 12. Lamberts H, Wood M (eds). International Classification of Primary Care. Oxford: Oxford University Press, 1987. [Google Scholar]
- 13. Stata Corporation. Stata Statistical Software, Release 5. College Station, Texas: Stata Press, 1997. [Google Scholar]
- 14. Duncan C, Jones K, Moon G. Context, composition and heterogeneity: using multilevel models in healthresearch. Social Science and Medicine , 1998; 46: 97–117. [DOI] [PubMed] [Google Scholar]
- 15. Ukoumunne OC, Gulliford MC, Chinn S, Sterne JAC, Burney PGJ, Donner A. Evaluation of health interventions at area and organisational level. British Medical Journal , 1999; 319: 376–379. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Goldstein H. Multilevel Statistical Models. London: Arnold, 1996. [Google Scholar]
- 17. Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics , 1986; 42: 121–130. [PubMed] [Google Scholar]
- 18. Armitage P, Berry G, Matthews JNS. Statistical Methods in Medical Research. Oxford: Blackwell Science, 2002. [Google Scholar]
- 19. Hamm RM, Hicks RJ, Bemben DA. Antibiotics and respiratory infections: are patients more satisfied when expectations are met? Journal of Family Practice , 1996; 43: 56–62. [PubMed] [Google Scholar]
- 20. Horne R, Weinman J, Hankins M. The Beliefs about Medicines Questionnaire: the developmentandevaluation of a new method for assessing the cognitive representation. Psychology and Health , 1999; 14: 1–24. [Google Scholar]