Family Practice Vol. 16, No. 1, 84-89
© Oxford University Press 1999
Do postal questionnaires change GPs' workload and referral patterns?
Backwell and Nailsea Medical Group, 15 West Town Road, Backwell, Bristol BS48 3HA, UK.
Watkins CJ. Do postal questionnaires change GPs' workload and referral patterns? Family Practice 1999; 16: 84–89.
Received 3 March 1998; Revised 16 July 1998; Accepted 7 October 1998.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Objective. We aimed to determine changes in workload in general practice associated with the postal administration of a health needs questionnaire.
Method. We carried out controlled before-and-after intervention study of the effects of delivering a postal questionnaire to assess needs for care for patients with arthropathies of the hip and knee, groin hernia and varicose veins, and to assess health service utilization, general health status and risk factors for cardiovascular disease. The setting was a seven-partner, fundholding, group practice in Avon. The subjects were patients registered with an NHS group practice situated in Backwell and Nailsea, Avon. The outcome measures were the frequency of consultation, home visits and night visits, reasons for consultation, referral to specialist agencies and patterns of prescribing.
Results. There was no significant difference between the study and control group in the year before and the year after the postal administration of the questionnaire with respect to changes in overall frequency of consultation, frequency of referral (including type of referral) and frequency of prescribing of non-steroidal anti-inflammatory drugs. In the study group there was a significant (P < 0.05) reduction in the number of daytime home visits and prescriptions written for analgesics. Analysis of the records of those who had received a medical examination, in addition to a postal questionnaire, showed that there was no significant difference between the study and control group with respect to frequency of consultation, referral to outside agencies or items prescribed.
Conclusion. Administration of a health needs questionnaire to patients registered with this general practice was not associated with an increase in consultation frequency or referral, or a change in prescribing patterns. No plausible explanation could be identified for the significant reduction in the number of home visits and prescriptions written for analgesics. It was concluded that these results were a statistical artefact. On the basis of the evidence from this study, GPs can be reassured that the administration of health needs questionnaires of the type used in this study will not result in any increase in workload or costs of care incurred by increased referrals to outside agencies or increased prescribing.
Keywords. General practice workload, health needs assessment, postal questionnaires, referral patterns..
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Postal questionaires are being used increasingly to determine the prevalence of symptoms of ill health and disability in populations, to enable planning of future provision of care. Many such surveys are conducted in general practice, either by purchasing authorities or academic organizations. They are undertaken either with the involvement of GPs, using the practice age/sex register as the sampling frame, or without the direct involvement of the GP, using the Family Health Service Authority (FHSA) register as the sampling frame. A concern expressed recently by a working party of the Royal College of General Practitioners was that: ‘A substantial increase in the number of contacts that people seek from primary care services could rapidly overwhelm the services. It is therefore of concern that the consequences of raising expectations through the media or unsolicited needs assessments [my italics] are largely unevaluated’.1 The purpose of this study was to determine whether a postal health needs questionnaire resulted in an increased workload for GPs and a consequent increase in referrals to outside agencies and changes in prescribing patterns.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
The study was undertaken in a seven-partner National Health Service (NHS) general practice providing care for 13 300 patients resident in the semi-rural communities of Nailsea and Backwell, 7 miles south-west of Bristol. The practice participated in a pilot study of a postal questionnaire to be used in the Somerset and Avon Survey of Health.2 This involved mailing a questionnaire to determine population requirements for elective surgery (hernias, varicose veins, arthropathies of the hip and knee). In addition, the opportunity was taken to include in the same envelope a questionnaire identifying risk factors for cardiovascular disease. The questionnaires were mailed in late June 1993 to an age/sex stratified random sample of 702 patients, aged over 35, derived from the practice age/sex register held by the Avon FHSA. The details of the sampling method used is described elsewhere.2 Of those patients who were sent a questionnaire, 100 individuals were invited for an examination, to validate the responses to the questionnaire.
The practice is fully computerized and, at the time of the study, supplied data to the GP Research Database (formerly the VAMP research database). The reason for every consultation is recorded, together with the date, place and time of day of consultation. In addition, the outcome of the consultation is recorded, in terms of referral to outside agencies and prescribed medication. A distinction is made between first and subsequent consultations in an episode of illness. Regular checks are made to ensure the completeness and accuracy of the data. In particular, a check is made to make sure that a medical history is entered to match every treatment prescribed and that a referral to hospital is matched with a hospital discharge report or a letter from a secondary care provider. The age/sex register in the practice matches that of the FHSA, as the two are electronically linked.
A set of all data provided by the practice to the GP Research Database was supplied by the Office of Population, Censuses and Surveys (OPCS), now the Office of National Statistics (ONS). This covered a period from 1 year before until 1 year after the mailing of the health needs questionnaire. Full dates of birth and postal codes derived direct from the practice register were amalgamated with the data from the OPCS.
As well as age and sex and season of the year, the frequency and the reasons for patients consulting with GPs vary with social class.3 For this reason, it was important to determine the social class characteristics of the sample. However, no occupational information was recorded on the patient register. A proxy for social class can be derived from the postal code, using the ‘Super-profile’ system of classification.4 This is a multi-dimensional small-area classification in which areas sharing similar distinctive patterns of socio-economic, demographic, housing, and environmental and behavioural influences are grouped together. The classification allocates virtually every enumeration district in Britain to one of 150 ‘Clusters’, each of which shares a typical social and demographic composition. These clusters can be amalgamated into larger aggregations known as ‘Groups’ and ‘Lifestyles’. Although the system was originally produced for commercial use, it has been shown to be of value in predicting variations in child health measures such as birthweight, immunization rates, childhood health screening uptake and mean height of schoolchildren.5
Table 1
shows the distribution of the practice population between the different Super-profile Clusters. Cluster groups containing 300 or fewer individuals in the total population are not included in this table, for simplicity of presentation. Ninety-two per cent of the practice population are high-income earners living in mortgaged properties. Only 8% are within the Super-profile cluster groups described as ‘hard-pressed families’ and ‘have-nots’.3
|
A control group was created by matching each individual in the study group with an individual in the remaining data set, for age (+ or –2 years), sex and Super-profile cluster. Controls could not be found for 21 patients in the study group either because a control was not available within the 2-year age band or because the Super-profile classification was unavailable. These individuals were omitted from further analysis, leaving 681 patients in the study group and the control group for further analysis. Omission of these 21 individuals did not significantly change the composition of the study group with respect to age/sex distribution, the proportion of individuals who had received a questionnaire alone or an invitation to attend a medical examination.
In order to test the hypothesis that no significant differences between study and control group could have been due to the fact that those in need of care had already have sought care from their GPs, the responses to the health needs questionnaire of the individuals in the study group were examined in relation to their reasons for consultation.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Figure 1
shows the characteristics of the study and control groups and the degree of accuracy of the matching with respect to age. The study group was originally selected as a random sample, stratified separately for age and sex in each decade from patients aged over 35, registered with six of the seven partners. Matching was perfect on Super-profile cluster and sex, and had a correlation of 0.9996 for the age differences between the two groups.
|
Table 2
shows the mean number of surgery consultations per patient, in the study and control group, in the year before and the year after the administration of the health needs questionnaire. It also shows the mean number of first or new episodes of illness, daytime home visits, night visits, consultations for varicose veins, groin hernias, hip or knee joint problems, cardiovascular system complaints and out-patient referrals. In addition, this table shows the mean number of prescriptions for analgesics and non-steroidal anti-inflammatory drugs written in the year before and the year after the administration of the questionnaire. All samples were checked for an underlying normal distribution and a paired Student's t-test was used to test the null hypothesis (of a mean of zero for the paired study/control patient differences for the change in workload). There was a significant reduction in the study group, compared with the control group, in the number of daytime home visits (P = 0.02) and number of prescriptions written for analgesics (P = 0.01). There was no significant difference in the change in frequency of occurrence of any of the other events in the year before and the year after the administration of the health needs questionnaire.
|
The number of all consultations (i.e. surgery attendances, telephone consultations, out-of-hours consultations, home visits and night visits) and prescriptions for analgesics and non-steroidal anti-inflammatory drugs was examined over shorter periods of time (6 weeks, 3 months and 6 months) before and after the administration of the health needs questionnaire. There was no difference between the groups in the change in frequency of occurrence of any of these events before and after the administration of the health needs questionnaire.
The study group was divided into four categories, according to whether or not patients returned the questionnaire and whether or not they were invited to take part in a medical examination. This was done in order to verify the questionnaire responses. The first category contained those who were sent and returned a questionnaire but were not otherwise involved in the survey (472); the second, those who were sent a questionnaire and did not respond (114); the third, those who returned the questionnaire and were invited and attended for a medical examination (68); and the final group those who were invited but for various reasons did not attend a medical examination (27). The analysis was performed down to the subgroup level. However, as the number of events was small in each category, the underlying normal distribution could not be guaranteed. The frequency of consultation, referral patterns and prescriptions offered were examined in the same ways as for the study group altogether. There were no significant differences between the study and control group for the changes in these variables in the year before and the year after the administration of the health needs questionnaire.
Of the 613 individuals in the sample of 681 who returned the health needs questionnaire, 137 (22%) reported having experienced pain in or around the knees, for 7 or more days in the previous year. Seventy-three (53%) of these patients said that they had not seen a doctor about knee problems. Seventy-seven (13%) of the 613 who returned questionnaires reported that they had pain in or around the hip joint, and, of these, 44 (57%) had not consulted a doctor about this symptom. Three hundred and twenty-eight patients (54%) reported varicose veins present, of whom 265 (81%) had not consulted a doctor about this symptom. Ten patients reported that they had currently a lump in the lower abdomen or groin and of those, six had not consulted a GP about this problem.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
We have shown in this study that patients sent a health needs questionnaire subsequently consult no more often than those who do not receive such a questionnaire. To what extent does the finding of this study depend on the size of the study sample, the social attributes of the population and the content of the health needs questionnaire?
Based on a sample size of 681 pairs, representing 5% of the population of 13 300 individuals served by the practice, it was only possible to demonstrate a minimum difference of 10% in the proportion of individuals (from an anticipated usual level of 70%) who consulted with their GPs over a period of a year, with a power of greater than 95% and a significance level of 5%. For this reason, the results taken from the subgroup analysis need to be interpreted with caution. To have answered questions about differences in the subgroups would have required a total population size of up to 100 000 individuals.
The proportion of individuals in this sample apparently in need of care, in particular for knee and hip problems and varicose veins, was considerable. Thus, the hypothesis that those in need of care for these surgically remediable conditions had already sought care from their GPs was not supported.
The questionnaire was designed to identify those in need of elective surgery for common surgical complaints. It also contained questions about the responder's general health, use of health services and factors influencing cardiovascular risk status. The content of the questionnaire was thus not perceived as threatening by the GPs involved in this study, and the purpose of its administration could be readily understood by those who were sent it. Had the subject matter of the questionnaire been more intrusive, then it could reasonably have been expected to generate more anxiety in patients and increase demands on GPs. This means that the results of this study need to be applied with caution to questionnaires with an intrusive content. Nevertheless, patients are remarkably tolerant of GPs who distribute questionnaires to their patients. A good example is a survey undertaken by Hooper6 of the prevalence of sexual abuse in women registered with his practice. This survey had a 65% response rate, which was considered to be good in view of the sensitive nature of the information being sought.
The statistically significant difference in the changes in the study and control group in the number of home visits and prescribing of analgesics provoked considerable discussion and thought. The difference was accounted for by an increase in the control group, between years one and two, in the number of home visits and prescriptions for analgesics, compared with a decrease in the study group. One hypothesis which would explain this was that the control group contained a higher proportion of patients with chronic or malignant disease, whose condition deteriorated, requiring more home visits and prescriptions for analgesics in the second year than the first. This was not supported by examination of the data. No plausible explanation could be advanced for these findings. The conclusion reached was that this was a statistical artefact resulting by chance, rather than an effect of the questionnaire in reducing patients' demands for home visits and analgesics. The data were subjected to over 500 t-tests and approximately 5% of these were statistically significant, at less than the 5% level, which was what would have been expected by chance alone.
Occupational information is poorly recorded in general practice. Even when it is recorded, it may often be out of date. The solution in this study was to use a multidimensional system of classification based on classification of census enumeration districts.3 Although this system has been shown to identify differences in patterns of uptake of child health services, further work needs to be done to see how well systems such as ‘Super-profiles’ relate to occupational information, other deprivation scores, and number and content of consultations.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
GPs can be reassured that health needs questionnaires mailed to up to 5% of the practice population do not increase their workload, referral or prescribing rates. This finding will not necessarily be applicable to practices in deprived areas or to questionnaires with a highly intrusive content.
|
Acknowledgments |
|---|
I am grateful to the help which Mark Williams gave to the development of this study, and the interpretation of the SASH data. Thanks are also due to Pete Shiarly for the major contribution which he made to the computer analysis of these data, and to his comments on the manuscript. We are also grateful to Tim Peters for his statistical advice, the OPCS for providing the data from the General Practice Database, Helen Watkins for her editorial advice, Robin Lambert for providing the original idea for the study and for the partners of the Backwell and Nailsea Medical group for allowing this study to take place. The study was funded by a grant from the South and West Region Directorate of Research and Development.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
1 Report from General Practice 27. The Nature of General Practice. London: The Royal College of General Practitioners, 1996.
2
Eachus J, Williams M, Chan P et al. Deprivation and cause specific morbidity: evidence from the Somerset and Avon survey of health. Br Med J 1996; 312: 287–292.
3 Office of Population Censuses and Surveys. Morbidity statistics from General Practice. Fourth National Study 1991–1992. London: HMSO, 1995.
4 Brown PJB, Batey PWT. A national classification of 1981 census enumeration districts: the derivation of super profile area types. Area classification note 1. Liverpool: Department of Civic Design, University of Liverpool, 1987.
5
Reading R, Openshaw S, Jarvis S. Are multidimensional social classifications of areas useful in UK health service research? J Epidemiol Commun Health 1994; 48: 192–200.
6 Hooper PD. Psychological sequelae of sexual abuse in childhood. Br J Gen Pract 1990; 40: 29–31.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||