Child Health and Education Study (CHES) - The Five-year Follow-up, 1975
The purpose of the 5-year survey was to review and evaluate preschool health, care and education services throughout Britain as experienced by the cohort of children born in a particular week in April 1970.
The main findings from the survey are published in two reports covering the social (Osborn, Butler and Morris, 1984) and health (Butler and Golding with Howlett, 1986) aspects of the children's lives. The following notes on the survey methods are extracted from the report by Osborn, Butler and Morris (1984) from which further details can be obtained.
Tracing the children
The children in England and Wales were traced through the cooperation of the registration division of the Registrar General's Office (RGO), the National Health Service Central Register (NHSCR) and Family Practitioner Committees (FPCs). The RGO produced a computer listing of every child whose date of birth was registered as occurring during that particular week in April 1970. Using the details provided by the RGO the NHSCR was then able to identify the Area Health Authority (AHA) in which each child was last registered with a NHS general practitioner. The NHSCR could also say if a child had gone abroad, was registered with a Service Medical Officer, i.e. if the father was a member of the HM Forces, or had died, in which case a copy of the death certificate was supplied. Using the NHS number provided by the NHSCR, Family Practitioner Committees could pinpoint the home address of children who had not emigrated or died. This information was passed to the community nursing service of the AHA in which the FPC was located, and the local health visitors personally approached the families in question, inviting them to participate in the study. This procedure ensured that information confidential to the health authorities such as the home address of the study child was not disclosed to the research team unless the parents had agreed to take part.
A similar procedure was adopted in Scotland, but because the administrative structure was slightly different, the Scottish NHSCR was provided with the names of the children who had taken part in the British Births (1970) study. They added the child's NHS number and located the Health Board in which the child was last registered with a NHS general practitioner. Administrators of Primary Care in each Health Board used this information to identify each child's home address which was communicated to the health visitors who then invited the parents' participation.
The whereabouts of children of servicemen who were registered with a Service Medical Officer were determined with the help of the Service Children's Education Authority. This group included 64 children of service families who were interviewed by nurses of the Soldiers, Sailors and Airmen Families Association (SSAFA) in West Germany, Malta, Gibraltar and Singapore.These children were included in the survey because they were members of the British Births cohort and although they were overseas in 1975 were likely to return to Britain and be included in future follow-ups of the cohort.
These methods of tracing depended entirely on the availability of information obtained at the time of the children's birth. Thus children who were born outside Great Britain but were resident here in 1975 could not be traced in this way. Health Visitors located some of these children by scanning child health records for children born during the study week, but immigrant children were inevitably under-represented in the sample. The NHSCR could provide no information about children who were adopted for reasons of confidentiality. Some adopted children were located from child health records, but information obtained on these children at age five could not be linked with that obtained at birth because of the lack of necessary information concerning their origins.
The cohort size in 1975 was estimated to be 16,284 children. Of these 13,135 (80.7%) were successfully traced and interviewed on or shortly after their fifth birthdays in April 1975. This was considered to be a reasonable response rate considering the difficulty of tracing children in the preschool period.
Four research documents were used in the five-year survey. These were:
1. Maternal Self-completion Questionnaire
2. Home Interview Questionnaire
3. Test Booklet
4. Developmental History Schedule
These documents were designed to obtain different types of information.
1. The Maternal Self-completion Questionnaire (MSQ) was designed for completion by the mother herself. This was done mainly to reduce the effect of interviewer bias in completing the attitudinal data. However, this also served to reduce the duration of the home interview. More than half the mothers (56.4%) completed the MSQ out of the presence of the health visitor and a further 28.1% completed it unaided during the health visitor's visit. The health visitor's help was needed by only 15.5% of the mothers, some of whom had difficulty in reading or required an interpreter. The MSQ contained questions concerning the child's behaviour at home and maternal depression. These were based on the Rutter A Scale of behavioural, deviance and the Malaise Inventory (Rutter et al., 1970). There were also forty-three attitudinal items designed to elicit attitudes towards child rearing, maternal employment, television viewing and hospital visiting.
2. The Home Interview Questionnaire (HIQ) was administered by health visitors who carried out the interviews in the children's own homes. Usually the interviewee was the mother (92.3%). Relatively few fathers were present at interviews (7%). Fewer than 1% of the interviews were carried out with persons other than the child's parents. Many uestions in the HIQ had precoded response categories where a finite number of responses could be anticipated. Other questions were of the open-ended type which required responses to be written down. This approach was used where the potential range of responses was unknown in advance. Replies to open-ended questions were coded according to schemes devised by scrutinising a thousand randomly selected questionnaires.
3. Test Booklet (TB) was administered by the health visitor during her visit to the child at home.
4. The Developmental History Schedule (DHS) was designed to obtain information from child health records. Details of developmental screening throughout the preschool period were obtained by reference to child health clinic and health visitor records where these were available. The number of missing records, however, make these data of limited use, and they are not deposited in the ESRC archives.
To achieve a reasonable degree of consistency in the manner in which interview and tests were carried out nationally, a comprehensive set of explanatory notes was prepared for the health visitor interviewers. In addition, briefing meetings were held at regional centres throughout Britain. Criticism of the deployment of health visitors as survey interviewers (Newson, 1970, p.19) is totally misplaced in studies of this type (Douglas, 1976, pp.11-13). Nearly half the families in this study were known to the health visitor interviewers through previous professional contact. Health visitors are seen by the mothers as medical workers with a legitimate interest in all aspects of the children's health, development, social and family circumstances. Health visitors also had access to child health records which provided an important additional source of information on the early health surveillance of these children as well as the means of tracing additional children born in the study week.
The main period of data collection took place over six months in 1975, during which time information was obtained on 95% of the traced sample. Thus, inferences can be made about children aged between five and five and a half years. The same statistical considerations apply to this sample as to that of the National Child Development Study. Davie et al., 1972, p 216 and Pringle et al., 1966, p 10, concisely summarise these considerations.
There are basically two types of non-response which have been termed specific and gross non-response. Specific nonresponse occurs where respondents cooperate in the study but fail to complete every part of it and results in some questions having a proportion of 'not stated' responses. Some questions in the present study were affected more by specific non-response than others. For example, information about the parental situation was obtained on every child, but information about the father's experience of any unemployment was not available for over 30% of the children. The majority of variables, however, had a specific non-response rate of under 5%.
Gross non-response was due to cohort members being completely missing from the survey. The two main causes of gross non-response were failure to trace children born in the study week and unwillingness on the part of some families to cooperate in the study. The total gross nonresponse rate was only a fifth (19.3%) of the estimated sample size despite the difficulty of tracing the children before they had all entered infant school. Assessment of bias was tackled in two ways: firstly, comparison of birth characteristics of those followed up with those who were not, revealed no social class differences, and no sex differences. Children born to teenage mothers, those of high parity, those who were heavy smokers were somewhat less likely to be contacted. This difference was relatively small. The strongest bias, however, concerned a failure to trace the children of mothers who were single, separated, widowed or divorced at the time of birth (Butler & Golding 1986). Secondly, it was decided to trace and obtain limited information about the children who were missed in the 1975 study in order to check whether these differed in any important respects from those whn were successfully traced and interviewed. This additional survey took place just two years later, when the children were seven.
There was little regional or social class variation in trace rates. Nevertheless, factors more directly related to the problems of tracing and interviewing cohort members, such as eographical mobility and family disruption, proved to be more important considerations (Osborn, Butler & Morris, 1984). Higher proportions of children in the 1975 survey were living with their natural parents.
Children of ethnic minorities were more difficult to trace, especially if they had come into Britain since the time of their birth. Interviewing difficulties, especially where there were language differences, further increased the risk of bias in the study. The 1975 survey found 92.3% of the children were with parents who were both of UK origin, i.e. not members of an ethnic minority, but in the 1977 study of children missed in 1975 this figure was down to 84%.
Families which are geographically mobile are expected to be difficult to trace and this factor could have been a major reason why some families were not traced in time for the 1975 survey.
Even with this sizable difference between the children in the two surveys, however, the relatively large size of the sample interviewed in 1975 compared with the 1977 sample means that any bias resulting from the loss of children in mobile families will be trivial for analyses involving the whole 1975 survey sample. Analysis carried out on a sub-sample of highly mobile children (i.e.those who moved more than twice in five years), however, would be very prone to error because over a fifth (22.9%) of this group were untraced for the 1975 survey. Comparisons were made in terms of overcrowding (persons per room ratio), and availability of four basic household amenities (kitchen, bathroom, indoor lavatory and hot water supply). The observed differences were so small that statistical significance was achieved for only two of the five housing indicators crowding and availability of a kitchen. These items suggested that the 1975 sample had a slight excess of children in poor housing conditions. The differences, however, were not sufficient to create a bias in the 1975 sample.