Section of Epidemiology & Community Medicine Table 8 Trend in days of certified incapacity for work in selected regions
Davs o /certified incapacityt (per miiale at risk) Rcqi(ll Wales North W Midlainds East Aniglia and South East All regiolns
Increase in 18 years
1953 54 20.2 16.4 11.9 10.1
1971 72 30.6 24 13.8 10.2
(per cent)
12.8
15.9
24
52 46 16 1
Table 9 Trend in selected causes of certified incapacity for work Dais of certified incapacitv (per mnale at risk)
C(ause All causes All causes except influenza Digestive Respiratory Circulatory Psychiatric Ill-defined conditions Accidents &c.
1962/63 13.88 13.17 2.21 4.03 1.65 1.14 1 1.12
1972 73 16.78 16.07 1.09 3.91 2.62 1.42 1.58 1.68
system, mental disorders and an increase over the country as a whole of more than 50 % in conditions too ill-defined to classify. There has been a very striking fall over a ten-year period in incapacity due to disorders of the digestive system. Primary Health Care In 1973 the DHSS commissioned David Hicks to carry out a comprehensive review of the information then available on all aspects of primary health care (Hicks 1976). In some 600 pages the author has quoted freely from over 400 books and papers so that the reader will rarely need to consult any of the original references. It covers in detail all the fields discussed here as well as many others, including health visiting, home nursing, child health and social work in general practice. What it cannot do is provide statistics and other information later than the time at which it was compiled; for this one must turn to the different sources to which reference has been made. Nevertheless this review fulfils its aim 'to bring together in one place the available data that may be helpful to those who aim to construct models of the balance of care from which realistic proposals for change can be proposed and tested'. This objective reminds us that collection of information should have a purpose; while curiosity is an important human attribute it is only valuable to the community when its results lead to action.
407
REFERENCES Butler J R, Bevan J M & Taylor R C (1973) Family Doctors and Public Policy. Routledge and Kegan Paul, London Cartwright A & Marshall R (1965) Medical Care 3, 69 Central Statistical Office (1975) Social Trends No. 6 Department of Health and Social Security (1973) Report of Joint Working Party on General Medical Services (1975a) Health and Personal Social Services Statistics for England (1 975b) Annual report of the chief medical officer General Medical Services Committee of the BMA (1968) Annual Report, appendix 9 Hicks D (1976) Primary Health Care - a Review. HMSO, London Irvine D & Jefferys M (1971) British Medical Journal iv, 535 Office of Population Censuses and Surveys (1973) General Household Survey 1971 (1975) General Household Survey 1972 (1976) Health Trends 8, 24
Dr D L Crombie (Royal College of General Practitioners, General Practice Research Unit, Harborne, Birmingham, B17 9DB)
Information for and from General Practice The primary function of any health care system is to help patients as individuals or groups to prevent, cure or alleviate their clinical problems. A secondary function is to help patients to maximize their social competence. The general practitioner fulfils a primary role as the assessor of previously undifferential clinical problems (Crombie 1962) and a secondary role as the main therapist and provider of continuity of care. The needs of patients generate the problems and these in turn dictate the information required for their solution. Clinical problems are more than disease processes but are composed of organic and psychosocial elements. The Individual Patient People during a lifetime bring approximately 100 episodes of illness to medical care (Crombie 1975a) of which approximately 5 can be classed as serious or significant at the time or have a high probability of having significance for the patient's future. Most of the information needed for the solution of clinical problems in general practice has to be generated de novo and could not have been previously recorded (Crombie & Pinsent 1976). The
408
Proc. roy. Soc. Med. Volume 70 May 1977
information content of the diagnostic labels used in general practice in terms of implied knowledge of pathology and etiology is poor (College of General Practitioners 1963a). Only 56% of diagnoses will be classed as firm. No diagnosis will be made in 8 % of cases, a tentative diagnosis in 30 % and an eliminative diagnosis in a further important 6 % (College of General Practitioners 1958). So far we have been concerned with information generated at the face to face consultation with the patient. For approximately 90 % of all episodes of illness no other information is needed and a basic clinical record which contains the dates of contact with the patient, a label for each problem and the action taken suffices. The principle of maximal reduction in space and time of all communication links dictates the location of the domiciliary care team in one group practice building. Here, consultation between general practitioner, district nurse, health visitor and social worker colleagues are best dealt with by telephone or face to face consultation. The primary records of the different professionals are seldom of much use to the others without some interpretative function. For the 16.40% of patients consulting who are referred for diagnostic tests (OPCS et al. 1974), a standard form suffices for request and result. The ad hoc letter is most often the appropriate form of communication for patients who are referred to hospital for specialist advice because of the enormous variation in the range of questions and answers.
Increasing efficiency gained by reducing complexity will then reduce the need for information flow. This can only be achieved in the clinical situation by using the principle of one problem, one problem solver, by maximizing the autonomy of each member, by clear definition of problems and by an equally clear definition of the roles of the participants as problem solvers. Shared problems are a recipe for non-solution. The Whole Practice The set of basic clinical records for each registered patient will also constitute a register of the practice but this can usefully be supplemented in all practices by an age-sex register. A disease index (College of General Practitioners 1963b) as the other standardized supplementary record system is appropriate for the minority of practices involved in undergraduate and postgraduate education or research. Repeat prescription and self-auditing record systems (Crombie 1975b) are being used with increasing frequency.
Age-sex registers: These consist of a card for each patient, carrying the name, sex, date of birth and last known address. These cards are filed, males and females separately, by year and month of birth, in the first year of life.
The register can be used for managing and monitoring any routine immunization, prophylactic, screening and surveillance programmes. For example, when the newly registered child reaches the age for the first triple immunization, the cards for children of that age in months are withdrawn and the mother is sent an invitation to bring her child for the appropriate procedure. The health visitor can then visit those mothers who do not respond to three such invitations. The completion of each stage of the immunization programme is marked by an appropriate entry on the age-sex register cards. The same procedures can be used for measles, rubella and any other routine prophylactic procedure, for monitoring cervical cytology and for any routine surveillance or screening procedures in the elderly. The register can also be used by health visitors and social workers to identify patients in particular age groups and for inviting specific groups for health education programmes (Pike 1976). Obvious examples are women at menopause and men at retirement. Such registers on separate cards can be created for routine surveillance and follow up of groups of patients specified by some common characteristics other than age. These can include patients with diabetes, hypertension, breast problems, malignant disease, and patients on long-term therapy with digoxin, psychotropic drugs or phenylbutazone. Over five million of the standard age-sex cards have so far been purchased from the Birmingham Research Unit of the Royal College of General Practitioners. Disease index: This is an indexing system for identifying patients by their recorded morbidity. A disease index in its simplest form is a Twinlock or similar binder containing sheets, one sheet for each disease or clinical problem in which the recorder is interested. In its standard form there is one sheet for each of the 586 disease or problem categories in the College of General Practitioners' classification of disease (College of General Practitioners 1963b) or from the International Classification of Health Problems of Primary Care (RCGP 1976) which has been developed from it. When a patient presents for the first time with any of these diseases, the name, date of first consultation for the condition and date of birth are entered on the appropriate sheets. These names
Section oJ Epidemiology & Community Medicine
accumulate as a total cohort of patients with that disease or problem who have presented during the period of recording. Such indexes are essential for good undergraduate and postgraduate teaching and for all retrospective studies.
the diagnosis of malignant hypertension and thyroid disorders and the adverse effects of immunization and vaccination procedures. The range of variability of general practitioners' consultation rates and referral to hospital has also been studied from data from the first year of the survey (Table 1). Inpatient and outpatient usage is primarily determined by referrals from general practice and these data show that the largest source of variability in hospital usage, the most expensive single element in the health service, must be determined by variation in the rates of initial referral by individual general practitioners. The second National Morbidity Survey data presents unique opportunities for examining the possible sources of this enormous range of variability. The basis will be a linked file of all morbidity experienced by each patient for all years at risk with selected socioeconomic data for each patient from the census file for 1970 and his consultation and hospital referral experience. In addition there will be an 'environmental profile' of certain geographic, social, meteorological and other physical characteristics of the practice area. A 'practitioner profile' will also be prepared of various academic and other characteristics of the participating practitioners and their practice organization. There will also be a profile of each practitioner's prescribing costs and patterns of prescribing.
The Second National Morbidity Survey Indexing systems were used for the second National Morbidity Survey where first and subsequent consultations for diseases were separately identified, and referrals to hospital were also recorded. Incidence, prevalence, patient consulting and total consultations could therefore be estimated for each disease and disease grouping (OPCS et al. 1973). The survey began in November 1970 and for the first two years involved 115 general practitioners in 53 practices caring for a population of approximately 290 000. From November 1972 until November 1976 recording was continued by 40 general practitioners in 18 practices, caring for a population of 100 000. The data from this second National Morbidity Survey (OPCS et al. 1974) forms a useful baseline for estimates of the incidence and prevalence of disease reported to general practitioners. It has been compared with the rates from the first National Morbidity Survey carried out in 1954-55 (Crombie et al. 1975). It is also the basis for various subsidiary studies including the natural history of gout, the criteria used in Table I
Practice variations: analysis of referrals in the second National Morbidity Survey 1970-1971
Referral rates per 1000 population at risk (all ages) Inpatients Outpatients Investigation TotalO
All practicesO 18.3 86.0 110.1 233.3
5th percentile 6.5 43.2 17.9 102.7
95th percentile 43.9 178.9 216.9 407.0
* Arithmetic mean for all the specific types of referral for all practices taking part in the survey * All referrals, including local authority, deaths, multiple referrals and others
Table 2 Practice variations: analysis of usage of medical facilities in Stoke 1964-1965
Hospital usage (No. oJ patients per 1000 patients on practice register (all ages)) Practice No. 1 2 3 4 5 6 7 All practices*
Hospital and GP
1118 51 32 128 113 41 63 74
Hospital, GP and local authority 14 5 3 4 7 4 7 6
* Arithmetic mean for all practices in the survey
409
Hospital only 76 64 62 71 97 82 66 72
All hospital facilities c ombined 208 120 96 203 217 127 136 152
GP only 510 582 378 722 658 715
715 608
Episode rate per 1000 patients at risk 1445 1154
715
2194 1942 1713 1751 1541
410
Proc. roy. Soc. Med. Volume 70 June 1977
The same range of variability of referrals is also evident in another study based on disease index recording in the City of Stoke for a twelve-month period in 1964-65 (Table 2). This study is unique in that information about usage of all these main areas of the health services by each patient was integrated into one file. This identifies another unexpected result: a large proportion of patients who attend hospital during a twelve-month period do not contact their general practitioner during that period. In this survey while only 8 % of patients attended their practitioner and were referred by him to hospital during the year, a further 7.30% attended hospital without consulting him. The most likely explanation of this paradox is that a large proportion of the 'new' attendances at hospital are in fact internal referrals from one department or consultant to another and that during this spell the patients are using the hospital services as their source of primary care.
Weekly Returns Service Indexing systems provide data for a 'weekly returns' service for the notification of infective, communicable and certain other new serious or significant illnesses from 84 practitioners caring for a population of approximately 200 000. The data from this system has also been used to study the natural history of mumps and the incidence of its complications and the epidemiology of influenza (Crombie 1977). A similar complementary network of 76 recorders report to the Surrey Research Unit of the Royal College of General Practitioners. Other Potential Sources ofInformation The Hospital Activity Analysis (HAA) files are potentially a source of information about the variability in referral patterns of general practitioners. Although the Hospital Medical Record Form 1 (HMRI) coding sheets from which the HAA files are constructed have entries which identify each patient's general practitioner, they have not been included in any HAA file. Conclusion Information is for action. For clinical problem solving, which is the primary purpose of the health services, a network information is appropriate. In contrast, management and administration are organized on a hierarchic basis and this is reflected in their information systems. This incongruity and incompatibility is compounded by the different context of information appropriate to these two separate functions. However, this incongruity is minimized to the extent that the rational need for maximal autonomy of function of all clinical
problem solvers improves efficiency by reducing the need for information flow. For example, the responsibility for maintaining an agreed minimal level of specified vaccination, immunization, prophylactic or other agreed surveillance procedures, which could be laid at any higher administrative level, would be to ensure that the norms were being achieved. Only when the norms were not being achieved would there be any need for further information as a basis for action from outside the group practice. This principle would only be acceptable and realistic if the autonomous group practice units undertook a standardized programme for self-evaluation and self-auditing. The preparation of such a set of standardized Practice Activity Analyses (PAA) has been initiated by the Birmingham Research Unit (RCGP 1977). The coordinated results of these individual audits would also be a source of information about the performance of general practice as a whole while the individual performance of each practitioner and group practice remains confidential to the group.
REFERENCES College of General Practitioners (1958) Journal of the College of General Practitioners 1, 107 (1963a) Journal of the College of General Practitioners 6, 197-204 (1963b) Journal of the College of General Practitioners 6, 219-224 Crombie D L (1962) Lancet ii, 257-258 (1975a) In: Scientific Aids in Hospital Diagnosis. Plenum Press, New York & London; pp 251-265 (1975b) Journal of the Royal College of General Practitioners 25, 337-343 (1977) Influenza (in press) Crombie D L & Pinsent R J F H (1976) Journal of the Royal College of General Practitioners 26,502-506 Crombie D L, Pinsent R J F H, Lambert P M & Birch D (1975) Journal of the Royal College of General Practitioners 25, 874-878 Office of Population Censuses and Surveys, Royal College of General Practitioners & Department of Health and Social Security (1973) Morbidity Statistics from General Practice: Second National Morbidity Survey, 1970-1971. Preliminary Report Method. HMSO, London (1974) Morbidity Statistics from General Practice. Second National Morbidity Survey, 1970-1971. Studies on Medical and Population Subjects. No. 26. HMSO, London Pike L A
(1976) Journal of the Royal College of General Practitioners 26, 698-703 Royal College of General Practitioners (1974) Journal of the Royal College of General Practitioners 24, 545-551 (1976) Journal of the Royal College of General Practitioners, Occasional Paper 1 (1977) Journal of the Royal College ofGeneral Practitioners (in press)
Davs o /certified incapacityt (per miiale at risk) Rcqi(ll Wales North W Midlainds East Aniglia and South East All regiolns
Increase in 18 years
1953 54 20.2 16.4 11.9 10.1
1971 72 30.6 24 13.8 10.2
(per cent)
12.8
15.9
24
52 46 16 1
Table 9 Trend in selected causes of certified incapacity for work Dais of certified incapacitv (per mnale at risk)
C(ause All causes All causes except influenza Digestive Respiratory Circulatory Psychiatric Ill-defined conditions Accidents &c.
1962/63 13.88 13.17 2.21 4.03 1.65 1.14 1 1.12
1972 73 16.78 16.07 1.09 3.91 2.62 1.42 1.58 1.68
system, mental disorders and an increase over the country as a whole of more than 50 % in conditions too ill-defined to classify. There has been a very striking fall over a ten-year period in incapacity due to disorders of the digestive system. Primary Health Care In 1973 the DHSS commissioned David Hicks to carry out a comprehensive review of the information then available on all aspects of primary health care (Hicks 1976). In some 600 pages the author has quoted freely from over 400 books and papers so that the reader will rarely need to consult any of the original references. It covers in detail all the fields discussed here as well as many others, including health visiting, home nursing, child health and social work in general practice. What it cannot do is provide statistics and other information later than the time at which it was compiled; for this one must turn to the different sources to which reference has been made. Nevertheless this review fulfils its aim 'to bring together in one place the available data that may be helpful to those who aim to construct models of the balance of care from which realistic proposals for change can be proposed and tested'. This objective reminds us that collection of information should have a purpose; while curiosity is an important human attribute it is only valuable to the community when its results lead to action.
407
REFERENCES Butler J R, Bevan J M & Taylor R C (1973) Family Doctors and Public Policy. Routledge and Kegan Paul, London Cartwright A & Marshall R (1965) Medical Care 3, 69 Central Statistical Office (1975) Social Trends No. 6 Department of Health and Social Security (1973) Report of Joint Working Party on General Medical Services (1975a) Health and Personal Social Services Statistics for England (1 975b) Annual report of the chief medical officer General Medical Services Committee of the BMA (1968) Annual Report, appendix 9 Hicks D (1976) Primary Health Care - a Review. HMSO, London Irvine D & Jefferys M (1971) British Medical Journal iv, 535 Office of Population Censuses and Surveys (1973) General Household Survey 1971 (1975) General Household Survey 1972 (1976) Health Trends 8, 24
Dr D L Crombie (Royal College of General Practitioners, General Practice Research Unit, Harborne, Birmingham, B17 9DB)
Information for and from General Practice The primary function of any health care system is to help patients as individuals or groups to prevent, cure or alleviate their clinical problems. A secondary function is to help patients to maximize their social competence. The general practitioner fulfils a primary role as the assessor of previously undifferential clinical problems (Crombie 1962) and a secondary role as the main therapist and provider of continuity of care. The needs of patients generate the problems and these in turn dictate the information required for their solution. Clinical problems are more than disease processes but are composed of organic and psychosocial elements. The Individual Patient People during a lifetime bring approximately 100 episodes of illness to medical care (Crombie 1975a) of which approximately 5 can be classed as serious or significant at the time or have a high probability of having significance for the patient's future. Most of the information needed for the solution of clinical problems in general practice has to be generated de novo and could not have been previously recorded (Crombie & Pinsent 1976). The
408
Proc. roy. Soc. Med. Volume 70 May 1977
information content of the diagnostic labels used in general practice in terms of implied knowledge of pathology and etiology is poor (College of General Practitioners 1963a). Only 56% of diagnoses will be classed as firm. No diagnosis will be made in 8 % of cases, a tentative diagnosis in 30 % and an eliminative diagnosis in a further important 6 % (College of General Practitioners 1958). So far we have been concerned with information generated at the face to face consultation with the patient. For approximately 90 % of all episodes of illness no other information is needed and a basic clinical record which contains the dates of contact with the patient, a label for each problem and the action taken suffices. The principle of maximal reduction in space and time of all communication links dictates the location of the domiciliary care team in one group practice building. Here, consultation between general practitioner, district nurse, health visitor and social worker colleagues are best dealt with by telephone or face to face consultation. The primary records of the different professionals are seldom of much use to the others without some interpretative function. For the 16.40% of patients consulting who are referred for diagnostic tests (OPCS et al. 1974), a standard form suffices for request and result. The ad hoc letter is most often the appropriate form of communication for patients who are referred to hospital for specialist advice because of the enormous variation in the range of questions and answers.
Increasing efficiency gained by reducing complexity will then reduce the need for information flow. This can only be achieved in the clinical situation by using the principle of one problem, one problem solver, by maximizing the autonomy of each member, by clear definition of problems and by an equally clear definition of the roles of the participants as problem solvers. Shared problems are a recipe for non-solution. The Whole Practice The set of basic clinical records for each registered patient will also constitute a register of the practice but this can usefully be supplemented in all practices by an age-sex register. A disease index (College of General Practitioners 1963b) as the other standardized supplementary record system is appropriate for the minority of practices involved in undergraduate and postgraduate education or research. Repeat prescription and self-auditing record systems (Crombie 1975b) are being used with increasing frequency.
Age-sex registers: These consist of a card for each patient, carrying the name, sex, date of birth and last known address. These cards are filed, males and females separately, by year and month of birth, in the first year of life.
The register can be used for managing and monitoring any routine immunization, prophylactic, screening and surveillance programmes. For example, when the newly registered child reaches the age for the first triple immunization, the cards for children of that age in months are withdrawn and the mother is sent an invitation to bring her child for the appropriate procedure. The health visitor can then visit those mothers who do not respond to three such invitations. The completion of each stage of the immunization programme is marked by an appropriate entry on the age-sex register cards. The same procedures can be used for measles, rubella and any other routine prophylactic procedure, for monitoring cervical cytology and for any routine surveillance or screening procedures in the elderly. The register can also be used by health visitors and social workers to identify patients in particular age groups and for inviting specific groups for health education programmes (Pike 1976). Obvious examples are women at menopause and men at retirement. Such registers on separate cards can be created for routine surveillance and follow up of groups of patients specified by some common characteristics other than age. These can include patients with diabetes, hypertension, breast problems, malignant disease, and patients on long-term therapy with digoxin, psychotropic drugs or phenylbutazone. Over five million of the standard age-sex cards have so far been purchased from the Birmingham Research Unit of the Royal College of General Practitioners. Disease index: This is an indexing system for identifying patients by their recorded morbidity. A disease index in its simplest form is a Twinlock or similar binder containing sheets, one sheet for each disease or clinical problem in which the recorder is interested. In its standard form there is one sheet for each of the 586 disease or problem categories in the College of General Practitioners' classification of disease (College of General Practitioners 1963b) or from the International Classification of Health Problems of Primary Care (RCGP 1976) which has been developed from it. When a patient presents for the first time with any of these diseases, the name, date of first consultation for the condition and date of birth are entered on the appropriate sheets. These names
Section oJ Epidemiology & Community Medicine
accumulate as a total cohort of patients with that disease or problem who have presented during the period of recording. Such indexes are essential for good undergraduate and postgraduate teaching and for all retrospective studies.
the diagnosis of malignant hypertension and thyroid disorders and the adverse effects of immunization and vaccination procedures. The range of variability of general practitioners' consultation rates and referral to hospital has also been studied from data from the first year of the survey (Table 1). Inpatient and outpatient usage is primarily determined by referrals from general practice and these data show that the largest source of variability in hospital usage, the most expensive single element in the health service, must be determined by variation in the rates of initial referral by individual general practitioners. The second National Morbidity Survey data presents unique opportunities for examining the possible sources of this enormous range of variability. The basis will be a linked file of all morbidity experienced by each patient for all years at risk with selected socioeconomic data for each patient from the census file for 1970 and his consultation and hospital referral experience. In addition there will be an 'environmental profile' of certain geographic, social, meteorological and other physical characteristics of the practice area. A 'practitioner profile' will also be prepared of various academic and other characteristics of the participating practitioners and their practice organization. There will also be a profile of each practitioner's prescribing costs and patterns of prescribing.
The Second National Morbidity Survey Indexing systems were used for the second National Morbidity Survey where first and subsequent consultations for diseases were separately identified, and referrals to hospital were also recorded. Incidence, prevalence, patient consulting and total consultations could therefore be estimated for each disease and disease grouping (OPCS et al. 1973). The survey began in November 1970 and for the first two years involved 115 general practitioners in 53 practices caring for a population of approximately 290 000. From November 1972 until November 1976 recording was continued by 40 general practitioners in 18 practices, caring for a population of 100 000. The data from this second National Morbidity Survey (OPCS et al. 1974) forms a useful baseline for estimates of the incidence and prevalence of disease reported to general practitioners. It has been compared with the rates from the first National Morbidity Survey carried out in 1954-55 (Crombie et al. 1975). It is also the basis for various subsidiary studies including the natural history of gout, the criteria used in Table I
Practice variations: analysis of referrals in the second National Morbidity Survey 1970-1971
Referral rates per 1000 population at risk (all ages) Inpatients Outpatients Investigation TotalO
All practicesO 18.3 86.0 110.1 233.3
5th percentile 6.5 43.2 17.9 102.7
95th percentile 43.9 178.9 216.9 407.0
* Arithmetic mean for all the specific types of referral for all practices taking part in the survey * All referrals, including local authority, deaths, multiple referrals and others
Table 2 Practice variations: analysis of usage of medical facilities in Stoke 1964-1965
Hospital usage (No. oJ patients per 1000 patients on practice register (all ages)) Practice No. 1 2 3 4 5 6 7 All practices*
Hospital and GP
1118 51 32 128 113 41 63 74
Hospital, GP and local authority 14 5 3 4 7 4 7 6
* Arithmetic mean for all practices in the survey
409
Hospital only 76 64 62 71 97 82 66 72
All hospital facilities c ombined 208 120 96 203 217 127 136 152
GP only 510 582 378 722 658 715
715 608
Episode rate per 1000 patients at risk 1445 1154
715
2194 1942 1713 1751 1541
410
Proc. roy. Soc. Med. Volume 70 June 1977
The same range of variability of referrals is also evident in another study based on disease index recording in the City of Stoke for a twelve-month period in 1964-65 (Table 2). This study is unique in that information about usage of all these main areas of the health services by each patient was integrated into one file. This identifies another unexpected result: a large proportion of patients who attend hospital during a twelve-month period do not contact their general practitioner during that period. In this survey while only 8 % of patients attended their practitioner and were referred by him to hospital during the year, a further 7.30% attended hospital without consulting him. The most likely explanation of this paradox is that a large proportion of the 'new' attendances at hospital are in fact internal referrals from one department or consultant to another and that during this spell the patients are using the hospital services as their source of primary care.
Weekly Returns Service Indexing systems provide data for a 'weekly returns' service for the notification of infective, communicable and certain other new serious or significant illnesses from 84 practitioners caring for a population of approximately 200 000. The data from this system has also been used to study the natural history of mumps and the incidence of its complications and the epidemiology of influenza (Crombie 1977). A similar complementary network of 76 recorders report to the Surrey Research Unit of the Royal College of General Practitioners. Other Potential Sources ofInformation The Hospital Activity Analysis (HAA) files are potentially a source of information about the variability in referral patterns of general practitioners. Although the Hospital Medical Record Form 1 (HMRI) coding sheets from which the HAA files are constructed have entries which identify each patient's general practitioner, they have not been included in any HAA file. Conclusion Information is for action. For clinical problem solving, which is the primary purpose of the health services, a network information is appropriate. In contrast, management and administration are organized on a hierarchic basis and this is reflected in their information systems. This incongruity and incompatibility is compounded by the different context of information appropriate to these two separate functions. However, this incongruity is minimized to the extent that the rational need for maximal autonomy of function of all clinical
problem solvers improves efficiency by reducing the need for information flow. For example, the responsibility for maintaining an agreed minimal level of specified vaccination, immunization, prophylactic or other agreed surveillance procedures, which could be laid at any higher administrative level, would be to ensure that the norms were being achieved. Only when the norms were not being achieved would there be any need for further information as a basis for action from outside the group practice. This principle would only be acceptable and realistic if the autonomous group practice units undertook a standardized programme for self-evaluation and self-auditing. The preparation of such a set of standardized Practice Activity Analyses (PAA) has been initiated by the Birmingham Research Unit (RCGP 1977). The coordinated results of these individual audits would also be a source of information about the performance of general practice as a whole while the individual performance of each practitioner and group practice remains confidential to the group.
REFERENCES College of General Practitioners (1958) Journal of the College of General Practitioners 1, 107 (1963a) Journal of the College of General Practitioners 6, 197-204 (1963b) Journal of the College of General Practitioners 6, 219-224 Crombie D L (1962) Lancet ii, 257-258 (1975a) In: Scientific Aids in Hospital Diagnosis. Plenum Press, New York & London; pp 251-265 (1975b) Journal of the Royal College of General Practitioners 25, 337-343 (1977) Influenza (in press) Crombie D L & Pinsent R J F H (1976) Journal of the Royal College of General Practitioners 26,502-506 Crombie D L, Pinsent R J F H, Lambert P M & Birch D (1975) Journal of the Royal College of General Practitioners 25, 874-878 Office of Population Censuses and Surveys, Royal College of General Practitioners & Department of Health and Social Security (1973) Morbidity Statistics from General Practice: Second National Morbidity Survey, 1970-1971. Preliminary Report Method. HMSO, London (1974) Morbidity Statistics from General Practice. Second National Morbidity Survey, 1970-1971. Studies on Medical and Population Subjects. No. 26. HMSO, London Pike L A
(1976) Journal of the Royal College of General Practitioners 26, 698-703 Royal College of General Practitioners (1974) Journal of the Royal College of General Practitioners 24, 545-551 (1976) Journal of the Royal College of General Practitioners, Occasional Paper 1 (1977) Journal of the Royal College ofGeneral Practitioners (in press)
