7.1 At various places in this Report we have drawn upon data from a variety of sources, including routine birth and death statistics, statistical returns collected for administrative or management purposes (such as the Hospital In-Patient Inquiry, HIPE), annual surveys (such as the General Household Survey, and the National Food Survey), the 3 national Birth Cohort studies and, of course, specific research projects. The scope and the quality of national statistics relating to health and mortality are exceptional: we appreciate that our task would have been more difficult in most if not all other European countries.

7.2 Nevertheless, at this point we must turn to consideration of information needs which are not met, and which we feel necessitate either the initiation of further research or modification of the regular instruments of data collection. Although this distinction is obvious and necessary, the collection of routine statistics (generally for administrative purposes) should not be seen as, or become, independent of research whether once and for all or carried out at regular intervals. In the health field as in many others social research may serve to point up new ways in which the provision of services, or use of services, or unmet need, should be monitored. The categories by which statistics are presented (such as age groups, social classes and income groups) must be sensitive to the implications of research for their adequacy and validity.

7.3 There is another general point about administrative statistics which needs to be made. The form of such statistics both reflects and determines the way in which the adequacy and performance of a service to which they relate is understood. This is broadly true at all levels of administration and provision. Modification of such statistics might serve in an important way to sensitise those responsible for providing health care to the saliency of possibly neglected matters, or to transform the way in which ‘performance’ is conceived. Two examples will suffice. Comprehensive and careful recording of occupation (or social class) on a hospital in-patient’s record could have value in educating clinical and nursing staff to the importance of social factors in the aetiology of diseases, in ill-health. Second in the UK community health returns are largely restricted to the volume of services provided (eg the number of children registered at Child Health Centres) in different areas.

7.4. In Finland, by contrast, reporting is of the proportion of all pregnant women and of all children (within age groups) who register, Assessment of provisions is thus manifestly on the basis of coverage. Wynn and Wynn (1974) attribute some of Finnish success in reducing perinatal and infant mortality rates to precisely this fact.


7.5. Because direct information on the distribution of sickness in the population is scanty, we have found it necessary (as did the Resource Allocation Working Party) to make use principally of (standardised) mortality rates as a proxy for the relative ill-health of different groups. In this section we discuss the sources of information on sickness itself which are now available.

7.6. In Chapter 1 of our Report we referred to the variety of notions of health and ill-health: it is of course health (and most desirably health in its positive sense of ‘well-being’) that we are concerned to promote. Direct information on the health of the population (rather than its sickness) is still less available, as we shall also show.

1. General Household Survey

7.7. The kind of data on the prevalence of long-standing illness, limiting long standing illness, and restricted activity due to acute illness which the GHS yields has been discussed elsewhere.

7.8. It has to be borne in mind that the GHS gives self-reported illness. Results are based on respondents’ perception of their own ill-health and on their willingness and ability to explain it to lay interviewers. The answers given are not validated against any medical records, and it is known that certain conditions (notably mental illness) are under-reported. No attempt is made to measure the “medical” severity of any condition reported. On the other hand, we have no evidence that variation between social groups in reported health is explained by differences in their perceptions of health rather than differences in their health conditions.

7.9 In the case of acute sickness whereas in the early years of the survey a clear social class gradient appeared this was less obviously the case in 1975 and 1976: it may in general be valid to deduce trends only from aggregated annual samples.

7.10 For the 1977 and 1978 Surveys, the health questions were redesigned and comparison with earlier years is not now possible. The basic approach was to identify by means of questions related to lists of illnesses all those respondents who felt that they had either a long-term or a short-term health problem however trivial. At this stage a much higher proportion of the sample reported ill-health than did so in previous surveys. A further series of questions then sought to determine whether these respondents acted differently from normal or restricted their normal level of activities because of ill-health. 1977 data show a clear class gradient for chronic sickness, but no apparent class-relationship in the case of short-term ill-health. (This contrast is perhaps rendered over-sharp by the exclusion from the category of short-term ill-health of all consequences of, or acute episodes in, chronic connditions).

7.11 We do not believe that the different indicators used for the 1977 and 1978 surveys were a real improvement over those used earlier and now reintroduced for 1979. In particular the usual distinction between chronic sickness or disablement and acute sickness is confused by extension of these concepts to “chronic health problems” and “short-term health problems” respectively. Better would be to obtain information, for different population groups, about episodes acute or short-term illness during a 12 month period and about long-term chronic illness or disablement and relate the two in terms of frequency and severity so that comparisons with (eg.) annual GP consultation and hospital out-patient attendance rates become more meaningful,.

2. Sickness Benefit

7.12 The annual DHSS publication Social Security Statistics includes a number of tables relating to sickness benefit. These tables give figures of the payment sickness benefit by sex, age, region, and cause. Information is not available for occupational groupings so analysis by social class or SEG is impracticable. Even if it were, limitations of this source of information are, firstly that it applies to the employed and insured population only and therefore excludes children, the elderly and many married women; second that it omits sickness absences of three days or less for which benefit is not paid; and finally that the diagnosis recorded may not properly reflect the physician’s views as these evolve in the light of further tests. It also needs to be remembered that ability to continue work despite being ill or injured is influenced by the nature of the employment. (The current DHSS cohort study of a sample of unemployed men and their families should yield valuable information on the relations between ill-health, sickness benefit receipts and absence from work/ unemployment).

3. National Morbidity Survey

7.13 This is a sample inquiry of general medical practitioners which examines the patterns of morbidity seen in general practices. The first such study carried out in 1955-1956, related to consultations with 76 practices (120 practitioners). It was reported on by Logan and Cushion in 1960, and we have referred to its conclusions earlier. A second study (in 1970/71) was based upon 53 practices, and there is a continuing exercise involving only 20 GPs, though these have not yet reported consultations, diagnosis etc, on an occupational basis. A further survey is planned for 1981. Given the size of the sample (doctors participating are volunteers) and the conflation of need for treatment with presentation, it is doubtful whether this survey could, be a major source of distribution of morbidity. Of course this is not to deny that the records of GP consultations could be used as a most significant source of data.

7.14 The limitations of these sources are all the more apparent when the problems inherent in the conceptualization of health are called once more to mind. As we explained elsewhere our view of ‘health’ is broader than the ‘absence of sickness’. It is thus our view that the monitoring of ill-health (itself still so imperfect) should evolve into a system also of monitoring health in relation to social and environmental conditions. This of course represents a major research task.

7.15 Two areas where progress in this direction could certainly be made, are i. in relation to the development of children and ii. in relation to disability. This brings us to the fourth existing source of information.

4. Community Health Statistics

7.16 In the context of our particular focus upon child health it is of interest to see what information could be made available on a regular basis. We know from the various birth cohort studies upon which we have drawn earlier in our report that the prevalence of defective vision and hearing, tooth decay, respiratory disorders and other indicators of ill-health and inhibited development are class-related among children. However because of the relatively small number of cases of certain handicapping conditions it is not always possible to be confident of these apparent variations. Data from Scotland based on medical examinations given to school children at the age of 5 also indicate variation in tooth decay and refractive error (Scottish Health Service Common Services Agency).

7.17 There is also some evidence for the interaction of social class and environment such that children born into social class V and also into (eg.) a heavily polluted area may be at particular risk (Colley and Reid, 1970), complementing evidence (eg in the RG’s Decennial Supplement) that infant mortality within a given social class varies between standard regions. It is unlikely that relatively small research studies could yield conclusive evidence on what could be an important priority for public policy, and hence it is worth seeing what may be obtained from national statistics. AHAs currently report on their school health surveillance procedures, on the proportions of pupils referred for further medical examination and on the presence of skin disease. The Working Group on Community Health Statistics has recommended that they also report on their policy of screening for hearing and vision, and on the numbers referred as a result of these tests. (Implementation of this recommendation has been deferred from 1979 to some later date). Scottish experience suggests that it might be possible to note ‘parents occupation’ when these tests are carried out since such information may well be available within the school. Unfortunately present practice with regard to regularity of surveillance/screening differs greatly between one AHA and another: perhaps inevitably given differing manpower situations. Moreover,the results of tests are presented only in terms of proportions suffering from visual or hearing impairments

7.18 There is a real need for continuing assessment of the development of children from birth at least through primary school: this is ‘health’ in one of its most crucial senses. Such assessment needs to be carried out in relation to social class, and must include surveillance of the nutrition of children. Studies of the kind we have in mind were in fact initiated in 1971 (when the availability of school milk was reduced and the price of school meals raised) under the auspices of the Committee on Medical Aspects of Food Policy (COMA), and a further study is being planned. Some of the findings of this work are discussed in Chapter 9. where we present our recommendations for policy relating to school meals and milk. The feasibility of surveillance of that kind has now been demonstrated, and the assessment of children’s growth in relation to class and nutrition should become a matter of routine.

7.19 We recommend that school health statistics should routinely provide the results of tests of hearing, vision, and measures of height and weight, in relation to occupational class. Authorities might also be requested to report separately upon schools in inner city areas, or to differentiate between a wide range of urban-rural locations. As a first step we recommend that health authorities, in consultation with education authorities, select a representative sample of schools in which assessments on a routine basis be initiated. This should be done in consultation with the National Survey of Health and Growth at St Thomas’. In our view it is information of this kind, more than current workloads, which should form the basis of planning provision in this area of community health. We therefore suggest that some of the additional resources which we recommend be put into the school health service be used in improvement of school health statistics, along these lines.

7.20 In the course of discussions with Health Authorities, substantial difficulties have been seen in the collecting and annual reporting of data in this form, and it might be that prior study of feasible means of overcoming these, difficulties would be desirable.

7.21 We should also like to see progress towards routine collection and reporting of accidents to children. Such reporting should ultimately distinguish not only between occupational classes and age groups, but also locations of accidents (road, home, school, other) and (as appropriate) the articles or building design features involved in accidents. At present the only national source of information is the HIPE. However not only does HIPE fail to record social (occupational) class, but it also omits the vast bulk of accidents not requiring in-patient admission. More promising is the Home Accident Surveillance System established by the then Department of Prices and Consumer Protection towards the end of 1976. This involves the collaboration of 20 hospitals distributed throughout England and Wales, and is particularly concerned with the formulation of safety standards for consumer goods used in the home. One report on domestic accidents to children has recently appeared (Department of Prices and Consumer Protection, 1979). It is clear in principle that an extension of this approach could have important implications for establishment of safety standards relating not only to domestic consumer goods (important enough) but to building and to road safety provision. It is worth recalling that fully a 1/3 of child deaths between ages 1 and 14 are due to accidents. We recognise that there are difficulties in working towards a national system of child accident reporting, which may require the co-operation not only of hospital A and E departments, of the police (in the case of traffic accidents), but conceivably of health visitors also. In view of the importance of the topic, we recommend that representatives of appropriate government departments (Health and Social Security, Education and Science, Environment, Trade, Transport, and the Home Office) as well as of the NHS and of the police, should consider how progress might rapidly be made on improving the information on accidents to children. The Child Accident Prevention Committee, if suitably constituted and supported, might provide a suitable forum for such discussions, to be followed by appropriate action by government departments.

7.22 A further area where a concept of health broader than (acute) sickness has to be employed is in relation to the distribution of impairment and disability. Harris’ survey (1971) found that there are some 3 million impaired persons over the age of 16 living in private households in Great Britain, or 7.8% of the appropriate population. This figure ranged from 0.89% of 16-29 year olds to 37.8% of 75s and over. The General Household Survey shows the prevalence of restrictive long standing illness, disability or infirmity to rise with SEG. Nevertheless, and despite the requirements of the Chronically Sick and Disabled Persons Act 1970, little is known of the provision of services to this group, their need for services (which might vary with social class as well as with the nature of the handicap), the aggravation of their conditions by bouts of acute sickness (and its implications) and so on. No such information is available on a national basis, and we would like to see local authorities reporting systematically on numbers of disabled as well as (however crudely) on assessments of severity of disablement. In the late 1970s local authorities had begun to classify handicapped people by the self-care criteria used by Harris, but that classification is not yet comprehensive, even for the physically handicapped, and needs to be extended, above all to the mentally handicapped. Social service departments and local officers of the Department of Employment should seek to introduce categorisations of severity, preferably on the basis of the limitation of activities put forwards in the WHO definition (See Chap 1).This would permit comparison of priorities between services as well as between authorities.


1. General Household Survey

7.23 Since 1971 the General Household Surveys have contained information about consultations with GPs in the previous two weeks (NHS consultations only), attendance at hospital out-patient departments and hospital in-patient spells both in the last three months, The data on GP consultations are generally regarded as accurately reported (though not subject to any independent verification) and produce higher rates than the National Morbidity Study. The situation is somewhat different with regard to hospital statistics since administrative data are also available (see below). However so far as outpatients are concerned overall rates reported in the GHS are not compatible with those reported on SH3 returns to the DHSS. This is partly because reporting is on a different basis. The GHS gives the rates per 1000 for people who during a three month period attended “as a patient the casualty or outpatient department of a hospital (apart from straightforward ante- or post-natal visits)”. Secondly the average number of attendances per outpatient, within the three month period are given. (Overall for England and Wales, the attendance rates (1975) are 97 for males, 88 for females, each attending on average 2.2 times). Administrative returns give ‘new out-patients per 1000 population’, ‘total out-patient attendances per 1000 population’ on an annual basis. Partly as a check on survey data it would seem useful if a GHS tabulation could be given of total out-patient attendances per 1000 population, converted to an annual basis. Moreover, unlike GP consultation rates, hospital out-patient utilization rates have not, since 1972, been published broken down by SEG, and we should like to see the tabulation proposed above given by SEGs. The situation is different again for utilization of hospital services on an in-patient basis, where once more administrative statistics are available, Again, however, they are available on rather different bases. SH3 and HIPE returns (see below), apart from being on an annual basis, include long stay (eg geriatric) hospital patients [who will be under represented in the GHS] and report rates of discharge and death together, though the two can be separated. Thus, rates of ‘discharge and death’ given in tables based on these Administrative returns are very much higher per 1000 population than rates reported in the GHS, The two cannot be reconciled because of differences of definition. Moreover, the GHS rates (“Number of medical and surgical in-patient spells per 1000 persons in a 3-month reference period”) are very low indeed (about 2%) and so difference between SEG rates, at least for a single year, would not be statistically significant.

7.24 In the early years of the GHS, the attempt was made to inquire into the use made of community health services in a single month (including Health Visitor, District Nurse, Chiropody, and Child Health (Clinic/Child Welfare Centre) services). The question was later dropped because not only was utilization at a very low level, but also respondents had difficulties in distinguishing between the various services with which (in some cases) they had been in contact.

2. NHS Hospital Activity Statistics

7.25 The basic source of information on hospital activity is the SH3 return, which shows the level of activity for each hospital classified by specialty. Data collected include the average number of available and occupied beds, the number of discharges during the year, the number on the waiting list at the end of the year, the average length of stay of patients discharged during the year, the number of discharges and deaths per available bed and the number of day case attendances. Information on out-patient activities includes the annual number of clinic sessions held, the number of new out-patients seen during the year and the total attendances due in the year (both new and old patients).

7.26 A particular point needs to be made in respect of waiting lists statistics which have been suggested as an indicator of differential need. Even if occupation data were available, the use of waiting list information in this way presents problems, as the existence of a long waiting list may itself serve to suppress potential demand. There are believed to be variations in clinical practice both in respect of the propensity of general practitioners to refer patients to hospital and also the policy adopted by consultants in the management of their waiting lists. Marked differences in both these operating characteristics make it difficult to draw conclusions based on comparison between waiting list statistics.

7.27 The SH3 return really provides only a set of measures of activity and workload and says nothing as such about the patients themselves. For this type of information we need to turn to HAA/HIPE.

7.28 It is reasonable to consider the Hospital-In-Patient Enquiry, (HIPE) and Hospital Activity Analysis (HAA) together. Although their historic origins differ, HIPE data is now generally produced as a sample extracted from HAA records. The underlying concept of hospital activity analysis is that the basis document, should form the front sheet of the patient’s case notes. The information collected on admission includes (as well as medical data) Name and Address, Age and Date of Birth, Country of Birth, Occupation of Patient or of Husband, Source of referral, Next of Kin, Source of Admission.

7.29 Although hospital activity analysis provides considerable information in terms of individual episodes of patients care, it in of limited value from our present point of view. The level of health care activity depends on both the need for services and their availability. The considerable geographical variation in services provided per head of population (which would almost certainly be reflected in similar data for social classes) cannot be separated into the element due to differences in need and those due to differences in provision. A further weakness is that the information on occupation (on which any social classification must depend) is not held centrally. The practice by individual RHAs is known to vary: one region is known to code up the occupation data collected on HIPE while another has deleted the item from the form altogether.

7.30 Although reservations must be held about the quality of occupation information which is collected on admission to hospital we feel this merits further study as outlined below. It should also be noted that published results of the equivalent Scottish survey (SHIPS) do include a class breakdown.

7.31 Similar procedures apply to the Mental Health Enquiry which is similar to HAA/ HIPE but is conducted in respect of patients in mental illness or mental handicap hospitals and units. Again occupation data is not held centrally and it is not known how extensively it is entered on the original form.

3. Community Health Service Statistics

7.32 Area Health Authorities report also on the range of community health services provided in their areas, notably:

  • registered Nursing Homes;
  • ante and post natal clinics, mothercraft and relaxation classes;
  • child health clinics;
  • health visiting, home nursing and midwifery services;
  • ambulance services;
  • immunisation;
  • chiropody;
  • tuberculin tests and BCG Vaccination;
  • school health services;
  • child guidance clinics;
  • family planning services;
  • dental services.

7.33 Much of this information is reported in annua1 volumes of Health and Personal social services statistics. In no case, however, do the data collected include Occupational (or any proxy) which would permit evaluation of use by social classes. The Working Group on Community Health Service Statistics, in making proposals for rationalizing and improving these statistics, made no reference to social class.

7.34 It is apparent that at present, systematic knowledge of the use made of the various health services by different social groups is extremely scanty. The principal source, the General Household Survey, presently reports only on (NHS) GP consultations on this basis. Results obtained on use of hospital services are not so given, and the sample size and rates of utilization would not permit a meaningful breakdown of inpatient hospital usage by SEG.

7. 35 Collection and central reporting of occupational data within the context of the various administrative returns obviously poses major problems of feasibility and of accuracy. However, because we feel the arguments in favor of there being some measures of differential use available for planning are powerful, we set them out below. We would like to draw these arguments to the attention of those responsible for the major review of all health service statistics which we understand has been initiated.

7.36 The first argument relates to the basis for establishing priorities in public expenditure. In so far as estimates of the redistributive effects of different elements of public expenditure (or proposed new spending) are, or might become an important element of public policy making, it becomes crucial to know how services are used.. It is obviously of particular importance to have this information in respect of the hospital services, which represent over 60% of all health expenditure. From the redistributional point of view class may be at least as important as age (on which basis data are available). The procedure currently used which allocates the benefits from NHS expenditure among income groups soley on the basis of the average age/sex composition of households in each income group plus knowledge of average age/sex differences in utilization, is not adequate. It remains to be seen whether GHS data will suffice here although in view of the limitations on analysis of hospital in-patient rates, this is doubtful.

7.37 The second argument concerns the extent to which some social groups may be making inadequate use of services in relation to their needs, and obviously has relevance for any attempt at directing services more specifically to areas of need. Although there is little direct evidence on ‘need’ for health care we do have indicative information. Some is provided by the General Household Survey from which the conclusion has been drawn in the Reports that those in SEGs IV and V are not, in fact less likely to consult a GP when they are ill. (We have discussed this conclusion, which conflicts with the “use-need” rations often calculated from SEG rates of illness and consultation in Chapter 4). In the area of preventive medicine however the situation in different. Here there is substantial research, much of which we have drawn on elsewhere in our report, which indicates declining use of child health, ante-natal, dental, and other services with declining occupational class. However These findings are often based on relatively small samples, are out of date (eg data on 7 year olds deriving from the National Child Development Study relate to 1965), and cannot be related to the differential extent of provision of the individual services at the area level.

7.38 Especially in child health success must in measured as much in terms of the extent to which services reach those most at risk, as by extent of overall coverage and both of these are more important indicators than a simple measure of scale of service provided or workload. We referred earlier to the importance of local accountability on the basis of coverage in the Finnish ante-natal and child health services. There would seen to be a strong case for something similar being attempted here. But the ideal measure and one which developments in statistical reporting must move towards, is assessment of the extent to which children from social classes IV and V and other high risk groups are reached by appropriate health services. Inclusion of ‘parents’ occupation’ in school and other child health records and reporting on this basis, is an invaluable first step.

7.39 The third argument refers to the ‘substitution’ of services, touched on earlier in this chapter. If it is true that many working class people use hospital outpatients where middle class ones would consult their GP, or that health visitors pay more attention to the working class children that the clinics do not see, then there are important implications for the utilisation of resources at the local area. Although there is need for more research here (since if such substitutions systematically occur it in important to know which and why) it will be through statistics devised in the light of research that the situation can be kept in view.

7.40 Research seems particularly necessary in relation to the differential utilization of community health services, since as we have shown such data are available neither through the GHS nor from administrative returns. Moreover, earlier experience with the GHS underlined the difficulties of investigating this particular issue. We consider that a small number of experimental studios should be mounted, possibly making (confidential) use of service records, and designed both to give preliminary indication of differential use by occupational class under different scales of provision of services, and to explore the problem of routine recording of occupational data in administrative returns. It may also be that such studies could be extended in the attempt to develop questions suitable for inclusion in the GHS, and which would ultimately permit assessment of the extent to which those within different social groups have access to these services.

7.41 We would like, at this point, also to draw attention to the importance of the National Food Survey for which the Ministry of Agriculture Fisheries and Food has responsibility. This annual survey is the principal source of information on the food purchase/consumption (and hence diet) of the population, and hence of great importance from the perspective of this report. There are, however, problems relating to the low response rate of the survey. We feel that much could be done, through greater recourse to epidemiological expertise (perhaps particularly the expertise of the OPCS) to transform the survey into a more effective instrument of nutritional surveillance in relation to health.

7.42 We recommend that consideration be given (drawing upon epidemiological expertise within the OPCS and elsewhere) to development of the National Food Survey into a more effective instrument of nutritional surveillance in relation to health, through which various ‘at risk’ groups could also be identified and studied. We fully recognise that such a development may raise questions about proper responsibility for the survey.


7.43 In this chapter we have made various recommendations which we feel will improve and extend the quality of class-related health and health services utilisation data on a regular basis and enhance knowledge of their interrelationship.

7.44 There are arguments in favour of at least some of the administrative data collection processes (undertaken by AHAs) including the recording of occupation. This is particularly the case in relation to child health services and the screening and surveillance of children (at least when reporting of this becomes routine), and in relation to the use of hospital in-patient facilities (where the GHS sample size is necessarily inadequate). We recognise that there are counter-arguments, based on considerations of feasibility and doubts as to the quality of occupation-related data, and we hope that in any major review of health service statistics both sets of arguments will be considered. In the meantime we feel that a pilot inquiry into data quality and implications of HIPE returns in a small number of AHAs which currently code occupation would be worthwhile. As a second stage of such inquiry, and if the collaboration of hospital staff could be secured the attempt might be made to check the reliability of occupational information recorded on admission (perhaps through next-of-kin).

7.45 The importance of the problem of inequalities in health and their causes as an area for further research needs to be emphatically stated. We recommend that it be adopted as a research priority by the DHSS and steps taken to enlist the expertise of the Medical Research Council (MRC), as well as the Social Science Research Council in the initiation of a programme of research. Such research represents a particularly appropriate area for Departmental commissioning of research from the MRC.

A strategy for Advance

7.46 In our view, the 5 areas in which further research (leading, it should be said, in some instances to improved or augmented administrative statistics) is essential are:

a. surveillance of the development of children, especially in relation to nutrition and to accidents);
b. better understanding of health effects of such aspects of (what can be regarded as) individual behaviour as smoking, diet, alcohol consumption, exercise;
c. the development of area social condition and health indicators (for use in resource allocation);
d. health hazards in relation to occupational conditions and work.

The importance of each of these derives from its relevance for our overall strategy for the development of policy.

e. study of the interaction of the social factors implicated in ill-health: over time, and within small areas.

7.47 The first has been discussed briefly above. It is discussed further, in the context of the implications for policy which flow from work already carried out, in Chapter 9. So far as the implications of behaviour for health are concerned, there is little doubt that among “the most difficult diseases to prevent are those on which depend on individual behaviour” (Prevention and Health P 38). The problem is widely recognised, and whilst we consider further research particularly on the health effects of diet and exercise/leisure behaviour to be important, enough is already known for us to make a number of recommendations in Chapter 8. The further development of social condition/ health indicators requires some discussion here.

Social factors in resource allocation and planning

7.48 In Chapter 8 we argue in relation to resource allocation for (at least partial) substitution of social condition indicators for standardised mortality indicators in calculation of revenue targets, particularly at sub-regional levels. Whilst we believe our argument to be well-founded we are conscious of certain weaknesses in the social condition data currently available. Most significantly, data are typically out-of-date (deriving only from the census), and do not reflect health needs as well as they do, say, housing need. We should like to see research designed to see how easily these difficulties can be overcome initiated. This would on the one hand be concerned with the development of health indicators: drawing not only upon the established (Census) indicators of social disadvantage but also upon information which could be gleaned from records of school health, sickness absence, concentration of handicapped people, etc. On the other hand, it would attempt to establish the feasibility (including cost) of regular sampling of health district/ local authority populations in order to provide necessary information on a sufficiently up-to date basis.

7.49 We also argue in chapter 8 for a shift of resources into community and preventative health somewhat more rapid than in current planning guidelines. Fraser in a study from which we quoted in chapter 5, found some evidence for a negative correlation between the proportion of a country’s health expenditure which is on preventive health and that country’s infant mortality rate. He felt this relationship, which he investigated using only limited data, merited further study. We agree. The same author, again using very limited data, found an inverse correlation between infant mortality rate and extent of inequality in income distribution between countries. This too is worthy of further study, and especially if cross national time series data were utilized could be of profound importance. We should like to see support given to a study of this kind.

The ‘Social Class’ Variable

7.50 One line of criticism of the notion of social class used in mortality and other statistics which we have already described (Chapter 1) in of the inadequacy with which ‘occupation’ expresses the life-style or command over resources of many substantial minorities. It was suggested that this may be the case, for example with the elderly, or the long-term unemployed, for whom ‘occupation when last employed’ may not be an entirely adequate indicator. It may that a composite indicator is to be preferred.

7.51 Indicators of this kind have been widely used in the characterization of social conditions: that is in relation to small areas (eg Gardner, Crawford and Morris 1961), As Reid says, however, they have been very little used, in Britain, for the characterisation of individuals (Reid 1977, P 15-16).

7.52 One attempt at ranking social conditions has been made in the context of the Bristol follow-up of the 1970 Birth Cohort Study (Child Health and Education in the Seventies). Their Social Index compounds occupation with level of education, housing (tenure, crowding, possession of amenities) and an ‘assessment’ of the area of residence. Each of these was scored and weighted to produce a one-dimensional index.

7.53 Preliminary indications are that Social Index score groupings are better able to discriminate on such variables as child developmental indicators and use of child health services than is RGs social class alone (Osborn and Morris 1979). Further work along these lines would be well worthwhile (part of which ought to be to examine the sensitivity of the results to the weighting method used), not least because it enables the effects of social class and area-based forms of deprivation to be studied together. Another possibility is to attempt to create composite indicators of family class, by combining husband’s occupation, wife’s occupation (if any), and the occupations of their fathers into a single index. Such an index (for which data exist at least within the national birth cohort studies) is likely to have much greater explanatory power with regard to child health than father’s occupation alone. (Though the value of such an index is likely to be far wider – eg in explanation of wealth or educational inequalities.)

7.54 An alternative to the use of such composite indicators, which also seems worth pursuing, is to examine the independent effect of the various dimensions of social stratification for which social class stands: eg income, education, housing, etc. Partly because they respond differently to change over the life cycle the correlation between occupation and such indicators as these is far from perfect. Hence it is possible that much closer correlations between mortality/morbidity and other socioeconomic variables are concealed.

7.55 Examination of this possibility might be pursued through further analysis of GHS data, and in principle regression analysis could be attempted using the wide range of socioeconomic variables included in the Survey. We recognise that this does pose problems (for example the income data have not been considered to be of high quality, and there would be practical problems in producing a tape permitting analyses of health data in terms of socioeconomic variables not hitherto considered in relation to health) – but we feel that much may be gained, and we should like such a study to be attempted.

7.56 A third significant issue which has been raised (Askham 1969) is the possibility that within social class V there is actually concealed an ‘underclass’: an ‘unstable social class V’ .  If this is so, and it seems very likely, it might explain the growing discontinuity at class V in the association between class and (eg) post-neonatal death rate. Moreover, Baird, Birch, Horobin, Illsley and, Richardson (1969) have shown that within class V the incidence of mental subnormality is significantly greater among those families with a large number of children, those living in the ‘poorest’ type of area, in the ‘poorest’ type of residence, those with a disorganised family life and those families which lived in poverty. There is other evidence for a relationship between family size and childhood illness (eg bronchitis), and between childhood accidents/ violence and family size and instability and parental history of psychiatric and physical illness. Nevertheless, as Morris has recently argued, we know far too little about the composition of class V (Morris, 1979).

7.57 It may be that the studies based on random samples (such as the GHS), or birth cohort studies, which we have suggested as a means of unravelling the relationship between social class and other (eg area-based) deprivations will not permit this issue to be broached. This is essentially because only about 6% of such a sample is now class V and the numbers involved do not therefore permit special analyses by family size, instability etc. What is needed here amounts to a study specifically of occupational classes IV and V, along the lines of the GHS, but taking more account of job/family/housing change; presence of mental handicap; family history of illness; retrospective study of periods of poverty, etc. This might be done by sampling in localities selected on the basis of their occupational class composition.

7.58 An alternative (or complementary) approach might be through study of selected industries possessing a suitable mix of social classes. Such a study would not focus exclusively upon specific occupational diseases (although it would involve rather broader questions relating to working conditions and health, such as safety, hygiene, pollution, and the effects of shiftwork upon self reported health). It would include interviews with members of employees households or families, and would use ‘industry’ both as a basis for securing a suitable national sample, and for examining the effects of a wide range of work-conditions upon health.

Ethnicity and Health

7.59 We have already indicated, in chapter 2, how slight is the information available in this country on the relationship between ethnicity (or race) and sickness. Small scale studies do suggest that somewhat different patterns of sickness obtain within immigrant communities (eg Gans (1966) found a higher prevalence of diet deficiency disorders among some children of immigrants (and we understand that rickets is still very common among Bengali children in Britain); Oppé (1964) found a higher prevalence of respiratory disorder among West Indian children; and Hood et al (1970) found West Indian children to be more prone to minor physical disorders), but this requires substantiation from larger scale work.

7.60 These differences appear to be explicable principally in terms of environmental variables. There is evidence for this from the United States (for example from comparison of differential mortality rates from birth through to the post-neonatal period – see Birch and Gussow pp 20-45). There is also evidence from this country of particularly delayed presentation for ante-natal care in some immigrant communities.

7.61 In chapter 8, which includes our proposal for an experimental action programme in a limited number of areas, we suggest that the training of health visitors for work in immigrant communities might be one aspect of such a programme. But further research is also needed. This would be designed to establish (1) the special health problems and health care needs (if any) of the principal immigrant communities; (2) whether there are other ways in which social policy can serve to reduce the health-effects of environmental and social factors upon a population which may, for various reasons, be at particular risk in respect of its health. A current study of the Bangladeshi community in Tower Hamlets is of the kind needed, but not only should there be more such research, but steps must be taken to facilitate contact between researchers in this poorly-developed field.

The Interaction of Social Factors: (1) interaction over time and transmission of ill-health

7.62 If respiratory disease is commonly to be found in all or most members of affected families, (for which there is evidence) how much more generally common is this familial aggregation of ill-health? If the family itself functions as a significant agent of transmission of ill-health (whatever the broader structural context) certain forms of intervention may acquire particular importance, as indeed may certain lines of further inquiry.

7.63 Some indications should be available from the GHS. In the context of the additional analysis which we have recommended, it should also be possible to look at the effects of (i) number of siblings and (ii) parental health on the morbidity of children. We would expect significant relationships to emerge, especially in class V, in line with the results of other research. [Eg Wilson and Herbert, studying 56 deprived families, found that in 40 “all or most members of the family were reported (ie. by themselves) as having had much illness, or as suffering from defects or conditions which affected their activities (Wilson and Herbert 1978).]

7.64 In so far as this may be the case, 2 lines of further study seem to be indicated. First, ways might be found of introducing consideration of health variables into the more common socioeconomic studies of inheritance of wealth, educational attainment and occupation. Second, it becomes desirable to determine the relative importance of parental ill-health as against other social and economic disadvantages (with which of course it will often be associated) in determining the health of children. Again, a start may be made with GHS data, since ‘parental morbidity’ might be introduced into a regression analysis (together with other socioeconomic variables), taking the health of children as dependent variable.

7.65 But in order to study the dynamics of child health – the process by which ill-health and educational under-achievement (whether a consequence of handicapping conditions, absence from school, or cultural factors) develop together and so perpetuate the link between health and social class – it is necessary to turn to longitudinal studies.

7.66 In our view 3 issues upon which research needs to be brought to bear are:

i. the interaction of processes leading to physical and mental disadvantage, handicap and ill-health;
ii. the role which services related to child health play, and might play in inhibiting the cumulation of disadvantage;
iii. the routes by which some children escape what is for most born into similar conditions an unenviable fate.

7.67 Questions such as these and others involving changes over time, such as the consequences for health of social mobility, are best examined through the longitudinal study, although the formulation of guiding hypothesis will in general derive from prior cross-sectional studies.

7.68 The OPCS longitudinal study (LA), based on a 1% sample of the census plus a 1% sample of all those subsequently added to the population by birth or immigration, is a potentially most important example of this approach. Utilizing census data as well as subsequent ‘life events’ (eg death of spouse, cancer registration, emigration, internal migration, death) routinely recorded, for each individual in the sample, the study permits a much more sophisticated examination of the relations between census indicators (such as housing conditions) and mortality than available from correlation. We have already (in Chapter 2) made use of some preliminary data from this longitudinal study, which showed the significance of housing tenure for mortality. Its value as a means of investigating the correlates of mortality is well recognised, and we need say no more than that we look forward with interest to the detailed studies being prepared (Goldblatt and Fox 1979).

7.69 The 1946 and 1958 national birth cohort studies represent a different approach in that the population is constant, and these have already yielded much of great value. Now that information on the children of the first group (who are now 39) is becoming available our understanding of intergenerational processes should be vastly enhanced. For our present purposes it is to the third such study (1970 births) that we can turn for a fundamental approach to the ‘determination’ of morbidity and development in children born into a world much more like that with which policy must now deal.

7.70 Available for analysis (but as yet largely unanalysed) are data on the 1975 follow-up of the 16,000 children in the 1970 birth cohort (1,000 of whom were also followed up at age 3). Included are interview schedules completed by the mothers of the 5 year-olds interviewed by Health Visitors and covering obstetric details (including subsequent births to the mother); breast-feeding; immunisations; visits to clinics, to GPs, to hospital and to the dentist; visits from Health Visitors; hospital admissions; operations; certain medical conditions:(including hearing or vision difficulties); details of accidents; bouts of wheezing; loss of consciousness; speech difficulties, etc. Also available are Developmental History Schedules which the health visitors completed from the records of health visitors, child health clinic and hospitals, and from ‘at risk-‘ registers (where available). This includes the results of all screening and check-ups done up to the 5th birthday.

7.71 Analysis of this data could include, as

dependent variables

  • use of child health services
  • accidents
  • specific morbidities (respiratory disorders, accidents skin disorders, etc.)
  • developmental indices (size, test performance, behaviour eyesight etc)

independent variables

  • class
  • Social Index (see above)
  • family size
  • parental and sibling health
  • mothers’ mental health
  • region
  • health service usage
  • geographic mobility

7.72 Hopefully more precise hypothesis linking such variables with morbidity will have previously been generated by the further analyses of GHS data which we have proposed. A question on the occupation of the parents’ parents at the time the child’s parents left school may permit some comment on the implications of inter-generational social mobility (as well as attempts at creating an indicator of family class).

7.73 The dynamic element enters from the linkage of these data with the original 1970 (birth) data. It would thus be possible to look for the significance of (eg) obstetric factors, birthweight, use of ante-natal services, as well as changes in family circumstances (eg social class-housing) or behaviour (eg. smoking) for the health of 5 year olds.

7.74 It should be possible to take children born into a, socially disadvantaged group (say, class V) and see which of them were now at age 5 in advance of the ‘norm’ (in terms of health, development) for class V births. To what extent is this ‘escape’ associated with above average use of health services, social or geographical mobility or other factors? Similar, analyses might be attempted for children born into very large families, or to unsupported mothers.

7.75 There is one other way in which a crucial element of change might be examined although technical questions relating to the extent to which this is feasible remain). This is through linkage between this cohort study and one of the earlier ones. Between 1946 and 1970 a great deal changed in Britain – economically, socially and in terms of the availability of health services, not least through establishment of the NHS. Although we have been able to look at trends in mortality data in relation to social class earlier in our Report, linkage of cohort studies would enable us to go much further in assessing the consequences of specific long-term shifts in the economic and social structures, and of health, educational and other social policies, upon that most crucial outcome: the health of children.

The Interaction of Social Factors: (2) interaction in small areas.

7.76 The longitudinal approach as described above is uniquely able to permit analysis of the way in which the variety of socioeconomic and environmental variables relevant for health interact over time. Because of its value as a means of investigating (class differences in) child development we set great store by such an approach.

7.77 A second approach which we feel offers scope for major advances in understanding focuses upon the way in which economic, social and environmental variables interact within small geographical areas. Such a study would be limited to a small number of such areas, selected on the basis of social condition or health data. It would involve collection of detailed economic, social, environmental and occupational data, as well as data on the health, ill-health, and mortality of the population. It would necessarily be a multi-disciplinary exercise. Such a study, we believe, would also permit far more detailed appreciation of the health effects of social and economic policies (without the need to assume the independence of such policies) than is possible from aggregate level data.


7.78 In this Chapter we make various recommendations which will improve and extend the quality of class related health and health service utilisation data on a regular basis and enhance knowledge of their inter-relationships.

7.79 It is argued, in relation to health, that the monitoring of ill-health (itself still so imperfect) should evolve into a system also of monitoring health in relation to social and environmental conditions. Two areas where progress could be made are (i) in relation to the development of children, and (ii) in relation to disability. Certain modifications to community health statistics are proposed:

We recommend that school health statistics routinely provide, in relation to occupational class, the results of tests of hearing, vision, and measures of height and weight. As a first step we recommend that health authorities, in consultation with education authorities, each select a representative sample of schools in which assessments on a routine basis be initiated.

7.80 We should also like to see progress towards routine collection and reporting of accidents to children, ultimately distinguishing age and occupational class of the parents as well as location and circumstances of accidents. In relation to traffic accidents there should be better liaison between the NHS and the police, both centrally and locally. We recommend that representatives of appropriate government departments (Health and Social Security, Education and Science, Environment, Trade, Transport and the Home Office)as well as the NHS and the police, should consider how progress might rapidly be made. The Child Accident Prevention Committee if suitably constituted and supported, might be a suitable forum for such discussions, to be followed by appropriate action by government departments.

7.81 In relation to the disabled, we should like to see local authorities report systematically on numbers of disabled as well as (however crudely) on assessments of severity of disablement applying to mentally handicapped people, elderly people in residential homes and other groups of handicapped people well as the general classes of the handicapped, as at present.

7.82 In our view it is the extent to which need is unmet, rather than pressure upon existing services, which should form the basis for planning, and it is this view which has underpinned our recommendations. Turning, then, to the health services themselves, it is clear that systematic knowledge of the use made of the various services by different social groups is extremely scanty. We recognise that collection and central reporting of occupational data within the context of the various administrative returns poses problems of feasibility and accuracy. Nevertheless we feel that the desirability of such information is such that further thought should be given to how these problems might be overcome, within the context of the current review of health service statistics.

7.83 Further, we draw attention to the importance of the National Food Survey as the major source of regular information on the food purchase (and hence nutrition) of the population. We recommend however, that consideration be given (drawing upon epidemiological expertise within the OPCS and elsewhere) to development of the National Food Survey into a more effective instrument of nutritional surveillance in relation to health, through which various ‘at risk’ groups could also be identified and studied.

7.84 Beyond this, we feel that the 5 areas in which further research is needed are:

  • surveillance of the development of children, especially in relation to nutrition and to accidents;
  • better understanding of health effects of such aspects of behaviour as smoking, diet, alcohol consumption, exercise;
  • the development of area social conditions and health indicators (for use in resource allocation);
  • health hazards in relation to occupational conditions and work;
  • study of the interaction of social factors implicated in ill-health: over time, and within small areas.

7.85 Though these issues are in an obvious sense quite distinct, yet they can also be seen as aspects of an overall strategy, and it is this strategy which we particularly wish to commend. Our concern is with the interaction of variables traditionally seen as directly implicated in ill-health (such as smoking behaviour and work conditions) with social variables. It will be necessary to examine the effects upon the health of social groups of a wide range of social, and behavioural variables-. implying further work both on the development of health indicators and upon the way in which disadvantageous social and environmental conditions may give rise to or exacerbate the effects of patterns of dietary behaviour, leisure behaviour etc.

7.86 The importance of the problem of social inequalities in health and their causes, as an area for further research needs to be emphatically stated. We recommend that it be adopted as a research priority by the DHSS, and steps taken to enlist the expertise of the Medical Research Council, as well as the Social Science Research Council, in the initiation of a programme of research. Such research represents a particularly appropriate area for Departmental commissioning of research from the MRC.

7.87 Whilst we ourselves have not attempted to develop a research strategy in detail within this report, it is our view that 2 types of study are needed. First, a study of the interaction of social and environmental variables over time, and their relationship to the (healthy) development of children. The longitudinal approach, as in the existing cohort studies, it appropriate here. Second, a study carried out in a small number of carefully selected places. Such a study would concern itself with the whole range of social conditions relevant for health, as well as patterns of behaviour which may in some senses be damaging to health. Crucial for further progress in the elimination of health inequalities is greater understanding of the interactions of this complex set of variables: social and individual. Such interactions will necessarily have both diachronic and locational dimensions, and the studies we have in mind will be sensitive to, and permit elucidation of, both.


Askham, J “Delineation of the lowest social class” J Biosocial Science 1 (1969)327

Birch, H G, S A Richardson, D Baird, G Horobin, and R Illsley Mental Subnormality in the Community (Baltimore: Williams and Wilkins, 1970)

Birch H G and J D Gussow Disadvantaged Children; Health, Nutrition and School Failure (N Y: Harcourt, Brace and World,1970)

Colley, J R T and D D Reid   “Urban and social class origins of childhood bronchitis in England and Wales” BMJ 2 (1970)213

Department of Prices and Consumer Protection: The Home Accident Surveillance System; Analysis of Domestic Accidents to Children 1979

Cans, B “Health problems and the immigrant  child” in C1BA Foundation Immigration – Medical and Social Aspects (London:  Churchill,   1966)

Gardner M J, M D Crawford and J N Morris “Patterns of mortality in middle and early old age in the county boroughs of England and  Wales”  Brit J Prev Soc Med  23/133  (1969)

Goldblatt,  P & J Fox “Household mortality from the OPCS longitudinal study” Population Trends Winter 1978, 20

Harris,  A Handicapped and Impaired in Great Britain (London HMSO 1971)

Hood, C, T E Oppe, I B Pless and E Apte Children of West Indian Immigrants – a study of one year olds in Paddington (London; Institute of Race Relation, 1970;

Morris, J N “Social inequalities undiminished” Lancet 13 January 1979, 87

Oppe, T E “The Health of West Indian Children” Proc Roy Soc Med 57 (1964)321

Osborn, A F and T C Morris “The rationale for a composite index of social class and its evaluation”  Brit J_Sociology March 1979

Reid, I Social Class Differences in Britain (London: Open Books, 1977)

Wilson, H and Herbert Parents and Children in the Inner City (London: Routledge, 1978)

Wynne M and Wynne A The Protection of Maternity and Infancy (London: Council for Children’s Welfare, 1974)

What do you think?

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 402 other subscribers

Follow us on Twitter

%d bloggers like this: