| This chapter gives the background to the Scottish Health Survey, outlines the design and content of the 1995 survey, and describes the structure of this report. |
1.1 The 1995 scottish health survey
The 1995 Scottish Health Survey is the first of a series of surveys designed to make a major contribution to monitoring progress towards the health targets set out in 1991 in Health Education In Scotland: A National Policy Statement,1 and towards the dietary targets announced in 1994. The establishment of the programme of surveys was set out in 1992 in Scotland’s Health: A Challenge to Us All.2 The survey was commissioned by The Scottish Office Department of Health, and is planned to be repeated every three years.
The aim in setting targets is to focus the development and delivery of health services, including health promotion services, so as to achieve gains in health and thus an increase in life expectancy and in the quality of people’s lives. Health Education in Scotland recognised that a health strategy for improving life quality involved a number of approaches designed not only to reduce the amount of ill-health - through high quality services, healthier lifestyles and improved physical and social environments - but also to alleviate its effects.
The health of the population of Scotland, as measured by standardised mortality rates, is consistently worse than in England and many other western developed countries, with deaths and ill-health from coronary heart disease being a particular problem in Scotland. Reducing the prevalence of coronary heart disease was identified as a top priority in Health Education in Scotland. To date, there has been little systematic information available on health and health-related behaviour which could help researchers understand the reasons for these differences in mortality. While there are statistics on the number and causes of death, and other statistics (such as hospital admissions) derived from people’s contacts with the National Health Service, these tend to be concerned with limited aspects of health, albeit at the more serious end of the spectrum of disease. There is no comprehensive picture of the health of the population, its biological characteristics or health-related behaviour available at national level, nor on how these characteristics may be changing over time; thus, there is little information available in order to observe trends or to aid in the explanation of the differences in mortality.
The Scottish Health Survey was therefore designed to overcome this lack of knowledge, with a number of specific aims:
1 to estimate the prevalence of particular health conditions in Scotland
2 to estimate the prevalence of certain risk factors associated with these health conditions and to document the pattern of related health behaviours
3 to look at differences between regions and between subgroups of the population in the extent of their having these particular health conditions or risk factors, and to make comparisons with other national statistics for Scotland and England
4 to monitor trends in the population’s health over time
5 to make a major contribution to monitoring progress towards the health targets set out in Health Education In Scotland and towards the dietary targets announced in 1994
The survey programme will thus fill many gaps in current knowledge and supplement the useful but limited information available from mortality statistics.
The Scottish Health Survey was closely modelled on the Health Survey for England, an annual survey which started in 1991. Similar to the survey in England, it is intended that the Scottish survey should consist of a number of core questions and measurements (such as height and weight) which would be included in each survey, plus modules of questions on selected topics that could change, as desired, from one survey to the next. It is planned that the Scottish Health Survey will be carried out every three years. The specific topic included in the 1995 survey was cardiovascular disease and its associated risk factors.
The 1995 Scottish Health Survey was carried out by the Joint Health Surveys Unit of Social and Community Planning Research (SCPR) and the Department of Epidemiology and Public Health at University College London Medical School (UCLMS) to a design determined by The Scottish Office Department of Health.
This section provides a short description of the survey methodology. A more detailed description can be found in the Technical Report.
The sample was designed to provide a nationally representative sample of the working age population of Scotland (ages 16-64) living in private households. Residents living in institutions (who tend to be older than people in private households) were excluded from the survey.
As one aim of the Health Survey was to look at regional differences within Scotland, it was important for the sample sizes to be large enough to permit analysis of the data at a regional level. For the purposes of the survey, seven ‘regions’ were defined by aggregating (mainly) contiguous Health Boards, as follows:
Highland & Islands (Orkney, Shetland, Western Isles)
Grampian & Tayside
Lothian & Fife
Borders, Dumfries & Galloway
Greater Glasgow
Lanarkshire, Ayrshire & Arran
Forth Valley, Argyll & Clyde
In order to provide sufficient sample sizes within each of these seven regions, the two smallest regions - Highland & Islands and Borders, Dumfries & Galloway - were slightly over-sampled, while the other five regions were sampled (roughly) in proportion to their population size.
Overall, a random sample of 14,358 addresses was selected within 312 postcode sectors from the Postcode Address File (PAF), using a stratified multi-stage sample design. Within the sampled postcode sectors, depending on region, either 45 or 47 addresses were systematically selected. At each address containing a private household, one person aged 16-64 was randomly selected (using the Kish Grid technique) to be included in the survey. A full description of the sample design is given in the Technical Report (Section 1.4).
Since addresses and individuals did not all have equal chances of selection, the data were weighted at the analysis stage.
Each sampled address was sent an advance letter and then visited by an interviewer. The interviewer randomly selected one adult aged 16-64 at the address, and sought that person’s consent for an interview. The interview was carried out using Computer Assisted Personal Interviewing (CAPI) and covered amongst other topics: general and dental health, longstanding and acute illness, cardiovascular disease, eating, smoking and drinking habits, recent accidents, pyschosocial factors, and background information. Interviewers also measured informants’ height and weight. The interviewer then sought agreement for a visit by a nurse.
The nurse collected information about use of prescribed medicines and recent experience of gastroenteritis; nurses then took blood pressure, lung function, and waist and hip measurements.
With the written agreement of the informant, a small blood sample was taken by the nurse, which was then analysed for: serum total cholesterol; HDL-cholesterol; plasma fibrinogen; haemoglobin; serum ferritin; gamma-glutamyl transpeptidase (gamma gt); and serum cotinine. In a sub-sample of sampling points, an extra tube of blood was taken, which was analysed for vitamins A, C, and E, and carotenoids.
It was made clear to informants that participation at each stage of the survey, and each measurement, was voluntary and that they could opt-out whenever they wished.
Fieldwork began in March 1995; 26 sampling points were issued each month over a twelve month period, with the vast majority of fieldwork being completed by the end of February 1996.
More detailed information on the conduct and content of the survey may be found in the Technical Report, and copies of survey documentation may be found in Appendix A to the Technical Report.
Interviews were obtained from 7,932 persons aged 16-64, which was 81% of all eligible adults.
Of those interviewed, 6,958 informants were visited by a nurse, and blood samples were taken from 6,184 informants. This represents response rates of 88% and 78%, respectively, based on adults who were interviewed. The ‘true’ response rates - that is, response based on the estimated number of eligible adults in the selected sample - for the different stages of the survey were:
|
Interviewed |
81% |
|
Height measured |
78% |
|
Weight measured |
75% |
|
Saw nurse |
71% |
|
Waist-hip measured |
69% |
|
Blood pressure measured |
70% |
|
Lung function measured |
69% |
|
Agreed to blood sample |
65% |
|
Blood sample obtained |
63% |
For a number of reasons, it was necessary to weight the data before it could be used for analysis. Details of the weighting scheme may be found in the Technical Report (Section 1.6). The weighted sample size was scaled to a total of 7,900 on the data set. All analyses contained in this report are based on weighted data.
As the Scottish Health Survey is a cross-sectional survey of the population at a particular point in time, it can examine associations between current health and personal characteristics or types of behaviour; it cannot however determine causality, because current health may reflect past, rather than present, behaviour and other conditions. Although (limited) questions about past behaviour were included in the survey, these questions are subject to memory and other types of error and therefore are limited in their usefulness.
Various analysis techniques, including age-standardisation, multiple linear and logistic regression, have been used in this report. Descriptions of the techniques used may be found in the Glossary in the Technical Report (Appendix E).
Information about the Health Survey, its objectives and design was circulated to all the Research Ethics Committees for all Area Health Boards in Scotland. Ethical approval was obtained from all fifteen committees.
The 1995 Scottish Health Survey report consists of two volumes. This volume reports the survey results (Chapters 2 to 13). Each chapter begins with a summary of the main findings, and contains a description of the data analysed. The main analysis variables in this report are age/sex, region and social class. The tables relating to each chapter are given at the end of the chapter and referenced within the text. (See Section 1.6 for a description of the conventions used within the report tables.)
The second volume is the Technical Report, which gives a detailed account of survey methods.
Given the close correspondence between the 1995 Scottish Health Survey and the 1994 Health Survey for England, and the fact that both surveys were carried out by overlapping research teams within SCPR/UCL, in order to aid comparison between the two surveys it was decided that this report should be modelled on the report of the 1994 English survey.3
In most chapters, some comparisons are made between the Scottish Health Survey results with results from the Health Survey for England and/or the Scottish Heart Health Study. These surveys are described more fully in sections 1.4 and 1.5.
Chapter 2: Physical activity
Three types of physical activity were covered by the questionnaire: occupation, home, and sports and exercise. Two measures of physical activity were used: a frequency-intensity activity level and a maximum intensity level. The chapter examines the contribution of each of the three components to overall activity levels, and looks at attitudes towards taking more exercise.
Chapter 3: Eating habits
Information on self-reported eating habits was collected for a wide range of food types. This chapter describes the self-reported eating habits for each food type and summarises differences by region and social class.
Chapter 4: Smoking
This chapter examines self-reported current and past smoking behaviour, attitudes to stopping smoking, and exposure to other people’s tobacco smoke. Levels of serum cotinine are used to validate self-reports of smoking behaviour.
Chapter 5: Drinking
Self-reported levels of weekly alcohol consumption are described and estimates are made for daily consumption levels. This chapter also examines attitudes to cutting down drinking, problem drinking among 16-17 year olds, and the relationship between the blood analyte gamma gt and reported alcohol consumption.
Chapter 6: Blood pressure
Blood pressure levels for the survey population are described. Differences in blood pressure by region and social class are examined, as is the relationship between blood pressure and a number of behavioural characteristics.
Chapter 7: Obesity
This chapter presents the distributions of height, weight, body mass index (BMI), and waist-hip ratio. Differences in BMI, obesity and waist-hip ratio are examined by region and social class, and the relationship between them and a number of behavioural characteristics are examined.
Chapter 8: Respiratory symptoms and lung function tests
Distributions for three common respiratory symptoms - phlegm production, breathlessness and wheezing - are presented, along with variations by region, social class and smoking status. Results from lung function tests for FEV1, FVC and PEF are also presented.
Chapter 9: Blood analytes
Distributions are shown for total and HDL-cholesterol, fibrinogen, haemoglobin and serum ferritin. Regional and social class variations are examined. The distributions of vitamins A, C and E, and carotenoids are also presented.
Chapter 10: Cardiovascular disease and its risk factors
The prevalence of self-reported cardiovascular conditions and intermittent claudication are presented, and variations by region and social class are described. The prevalence of the main risk factors - obesity, smoking, drinking, raised total cholesterol, high blood pressure - among all informants and those with cardiovascular disorder and with ischaemic heart disease or stroke is also examined.
Chapter 11: General health, use of health services, prescribed medicines and dental health
The first part of this chapter looks at self-reports of general health, longstanding illness or disability, and acute sickness. The second part looks at the prevalence of gastroenteritis in the population. The third part examines the use of a number of health services - GP consultations, inpatient stays and outpatient visits, blood pressure and cholesterol monitoring. The fourth part describes informants’ reports of the prescribed medicines they take. The final section looks at prevalence of false teeth, and dental practices.
Chapter 12: Psychosocial well-being
This chapter looks at emotional well-being using the GHQ12, and its relationship with a number of socio-economic and behavioural characteristics.
Chapter 13: Accidents
The number and causes of accidents are examined, along with their location and the types of injuries incurred. Accidents in work are separately examined.
1.3.2 Volume II: Technical report and documentation
Chapter 1: Survey methodology and response
A complete account of the survey design is provided, including sample design, response rates, and weighting. Also included are sampling errors associated with the estimates given in this report.
Chapter 2: Blood analytes, quality control and quality assessment
This chapter provides a description of the biochemical assay methods used, along with external and internal quality control results.
Appendices
Appendix A contains copies of the main fieldwork documents and the CAPI questionnaire. Appendix B includes the protocols used for taking the blood sample and for all measurements: height, weight, waist and hip circumferences, blood pressure, and lung function. A map showing the Health Board regions in Scotland is found in Appendix C. Appendix D is a brief description of the results from two experiments designed to give estimates of the reliability of some of the physical measurements taken by nurses and interviewers. Appendix E is a glossary which contains descriptions of the main analysis techniques and survey terms frequently used throughout the report.
1.4 comparison with the health survey for england
As mentioned above, mortality in Scotland tends to be higher than that in England for various causes, including cancer, coronary heart disease and strokes, and there is therefore particular interest in understanding the factors which may underlie these differences. For this reason, it was decided that the Scottish Health Survey should be similar in design to the series of annual health surveys which began in England in 1991.
For four years, from its inception through 1994, the focus of the Health Survey for England was on cardiovascular disease. The main focus of the 1995 Scottish Health Survey was therefore on cardiovascular disease. (The 1995 English survey focused on asthma, accidents and disability.)
Given the aim of comparison between Scotland and England, the interviewer and nurse questionnaires used in Scotland in 1995 were almost identical to those used in England in 1994. The questions on CVD were in fact exactly the same, as were the questions for most of the main risk behaviours (with the exceptions of physical activity and eating habits) and the measurement protocols for height, weight, blood pressure and lung function (with only very small differences for the waist and hip measurements). The main blood sample analyses for the Scottish and English surveys were carried out using the same laboratory.
Given the similarities between them, it is possible to make direct comparisons between the results of the Scottish and English surveys and, in most chapters, a few such comparisons have been made. In general, results have been compared with the most recent data available in England: that is, for core topics, Scottish results were compared with those from the 1995 Health Survey for England; however, for CVD, blood pressure, obesity and blood analytes, comparisons were made with results from the 1994 English survey.
More generally, in order to aid comparison of the Scottish and English health surveys, the format and content of this report (including the layout of tables) is very closely based upon the 1994 Health Survey for England report. However, one word of caution is in order for readers who may wish to use the Scottish and English reports in order to compare results from the two countries: it should be remembered that, since the age range of adults included in the two surveys differed (16-64 in Scotland; 16+ in England), the percentages (and other figures) shown in the ‘Total’ column of most tables in the two reports will relate to differently defined adult populations and therefore are not directly comparable. In this report, all comparisons with English survey results have been made for the equivalent population of 16-64 year olds in England. (In 1995, the English sample included interviews with 5895 men and 6824 women aged 16-64.)
1.5 comparison with the scottish heart health study
In a few chapters, results from the Health Survey are also compared with data from the Scottish Heart Health Study (SHHS). This study was carried out in 1984-1986 among adults aged 40-59 years living in 22 districts in Scotland. Results from the SHHS may be found in a number of publications.4
The main focus of the study was similar to that of the 1995 Scottish Health Survey - that is, it looked at lifestyle and coronary heart disease risk factors. Thus, an important reason for comparing Health Survey data with that from the SHHS is to look for trends over time. However, differences in results between these surveys must be interpreted with caution because there were a number of important differences in methodological design between the SHHS and the Health Survey. Firstly, the SHHS questionnaire was self-administered (rather than carried out by personal interview). Secondly, informants visited a clinic staffed by nurses where measurements were made and a blood sample taken (rather than nurses visiting informants in their own homes). Thirdly, the study was based in 22 (of the then 56) mainland local authority districts in Scotland (whereas all areas in Scotland, except for the smaller islands, were eligible for inclusion in the Health Survey). Fourthly, the SHHS sample involved selecting patients from GPs’ lists (whereas the Health Survey sample was obtained by selecting addresses from PAF).
In the following chapters, whenever comparisons are made with the SHHS, the Health Survey results have been limited to informants aged 40-59 years to ensure comparability between the two surveys.
1 The following conventions have been used within tables:
* category not applicable
- no observations (zero value)
0 non-zero values of less than 0.5% and thus rounded to zero
[ ] used to warn of small sample bases, that is unweighted bases of less than 50
2 The row or column percentages may add to 99% or 101% because of rounding. (When more than one answer could be given to a question, percentages may add to considerably more than 100%.)
3 If a percentage is quoted in the text for a single category that aggregates two or more of the percentages shown in a table, the (more precise) percentage in the text has been recalculated and may differ by one percentage point from the sum of the percentages in the table.
4 Values for means, medians, percentiles and standard errors (SE) are shown to an appropriate number of decimal places.
5 ‘Missing values’ occur for several reasons, including: refusal or inability to respond to a particular question; refusal to co-operate in an entire section of the survey (such as the nurse visit or self-completion questionnaire); and questions which were not applicable to informants. In general, missing values have been omitted from all tables and analyses.
Most tables in the report show a ‘dependent’ variable (comprising the rows of the table) cross-tabulated by an ‘independent’ variable (comprising the columns). The percentage base for the values of the dependent variable is normally found at the bottom of each column. It excludes missing values. Informants with a missing value on the independent variable are not shown as a separate column, but are included in the total column (unless they also have a missing value on the dependent variable).
6 The tables show both the weighted and unweighted bases. The weighted bases are shown in order to reflect the true relative sizes of the groups analysed. The unweighted bases are shown because the accuracy of the results depends on their size.