 |
 |
 |
 |
||
 |
||
| Home | Command Papers | House Papers | Departmental Papers | Search | Site Map | Contact Us | Links | ||
 |
|
|
|
HSE - Methodology and Documentation
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Booklet for adults | GHQ12, social support, social capital, perception of weight, pregnancy history, contraceptive pill use |
| Booklet for young adults | Smoking, drinking, GHQ12, social report, social capital, perception of weight, pregnancy history, contraceptive pill use |
| Booklet for 13-15 year olds | Smoking, drinking, GHQ12 |
| Booklet for 8-12 year olds | Smoking, drinking, cycle helmet use |
Interviewers also measured the height and weight of all informants. (For those aged six weeks to under two years, length was measured by the nurse.)
At the nurse visit, prescribed medication was asked about. For infants, additional information was collected on immunisations. The nurse took the blood pressure of those aged 5 and over, and made waist and hip measurements for those aged 16 and over. Nurses also measured infant length (for those aged six weeks to under two years). With written agreement, a small sample of blood was taken by venepuncture from those aged 16 and over. The blood sample was analysed for total and HDL-cholesterol, fibrinogen, C-reactive protein and glycated haemoglobin. Nurses also sought agreement for the storage of a small sample of blood for possible future analysis. Samples of saliva (for the analysis of cotinine) were taken from children aged 4-15.
At one-sixth of the sampled addresses the nurse visit was longer and involved additional procedures. These involved collecting a saliva sample and a urine sample (for the analysis of dietary sodium) from those aged 16 and over. This is the first time in the HSE series that urine samples have been taken. Nurses also took a fasting blood sample (from those aged 35 and over) on a separate visit for the analysis of triglycerides, LDL-cholesterol and glucose.
Nurses administered a self-completion booklet about eating habits to those aged 16 and over.
Each sampled address was sent an advance letter which introduced the survey and stated that an interviewer would be calling to seek permission to interview.
At initial contact, the interviewer established the number of households living at an address. If there were three or fewer, all were included in the sample. If there were more than three, three were selected for inclusion in the survey, using a random selection procedure.
The interviewer then made contact with each household and attempted to interview all adults and up to two eligible children aged 0-15 (see Section 4.4).
At each co-operating household, the interviewer first completed a Household Questionnaire, information being obtained from the household reference person or their partner wherever possible. This questionnaire obtained information about all members of the household, regardless of age. If there were one or two children aged under 16, they were automatically included in the sample for further interview. If there were three or more children aged under 16, the computer program used random numbers to select two of the children for interview. Individual Questionnaires were created by the program for these two children only.
An Individual Questionnaire interview was carried out with all adults and with sampled children. In order to reduce the amount of time spent in a household, interviews could be carried out concurrently, the program allowing for up to four informants to be interviewed in a session.
Height and weight measurements were obtained towards the end of the interview.
At the end of the interview, informants were asked for their agreement to the second stage of the survey, the follow-up visit by a nurse. In the case of children aged under 13, it was the parent's permission that was sought (see Section 4.4 for details). Wherever possible an appointment was made for the nurse to visit within a few days of the interview. At this visit the nurse carried out the measurements described in Section 3.3 and obtained the saliva, blood and urine samples from those eligible to provide these samples.
Before a blood sample was taken, written consent was obtained from the informant. Nurses also asked informants for consent to store part of the blood sample for additional analyses at some future date. If the informant agreed, written consent was obtained.
In addition to an advance letter, informants were given two leaflets describing the purpose of the survey and the associated measurements. Interviewers handed out one describing the purpose of the interview, and nurses handed out one explaining the purpose of their visit. Copies of these two documents are included in Appendix A. Informants were also given a leaflet summarising some findings from previous surveys.
4.4 Interviewing and measuring children
Children aged 13-15 were interviewed directly by interviewers, permission having first been obtained from the child's parent or guardian. Interviewers were instructed to ensure that the child's parent or guardian was present in the home throughout the interview. Information about younger children was collected from a parent. Younger children were present while their parent answered questions about their health. This was partly because the interviewer had to measure their height and weight and, in the case of those aged 8 and over, to ask the child to complete a short self-completion booklet during the interview. It also ensured that the child could contribute information.
Permission for a nurse to carry out any measurements on a child aged under 16 had to be obtained from the child's parent or someone else with legal parental responsibility for that child. This person had to be present during the nurse visit. In six households there was a child for whom no-one claimed such responsibility, and no nurse visit was therefore arranged.
Each informant was given a Measurement Record Card in which the interviewer and the nurse entered the informant's height, weight, waist and hip and blood pressure measurements. Informants were also sent (if they wished) the results of their blood sample analyses. Informants who saw a nurse were asked if they would like their blood pressure readings and blood sample analyses to be sent to their GP. If they did want results to go to their GP, written consent was obtained. Written consent to send information to a child's GP was obtained from the parent.
Nurses were issued with a set of guidelines to follow when commenting on informants' blood pressure readings (see Appendix B for details). If an adult's blood pressure reading was severely raised, nurses were instructed to contact the Survey Doctor at the earliest opportunity. They were instructed not to comment on a child's reading but to leave the Survey Doctor to assess whether any action was required. Where permission had been given for results to be sent to an informant's GP, the Survey Doctor contacted the GP if any blood pressure or blood sample results were abnormal.
5 Fieldwork quality control and ethical clearance
5.1.1 Training interviewers and nurses
Interviewers were fully briefed on the administration of the survey. They were given training in measuring height and weight (including a practice session).
All nurses were professionally qualified and proficient in taking blood before joining the Health Survey team. They attended a two-day training session at which they received equipment training and were briefed on the specific requirements of the survey with respect to taking blood pressure, taking anthropometric measurements and taking saliva and blood samples.
Full sets of written instructions, covering both survey procedures and measurement protocols, were provided for both interviewers and nurses (Appendix B contains a copy of the measurement protocols).
Interviewers and nurses who had worked on the previous year's survey attended full day refresher training sessions, where the emphasis was on improving measurement skills and gaining informant co-operation.
All interviewers and nurses new to HSE were accompanied by an interviewer or a nurse supervisor during the early stages of their work to ensure that interviews and protocols were being correctly administered. Routine supervision of the work of both interviewers and nurses was carried out thereafter.
5.1.2 Checking interviewer and measurement quality
A large number of quality control measures were built into the survey at both data collection and subsequent stages to check on the quality of interviewer and nurse performance.
Recalls to check on the work of both interviewers and nurses were carried out at 10% of productive households.
The computer program used by interviewers had in-built soft checks (which can be suppressed) and hard checks (which cannot be suppressed) which included messages querying uncommon or unlikely answers. For example, if someone aged 16 or over had a height entered in excess of 1.93 metres, a message asked the interviewer to confirm that this was a correct entry (a soft check). For children, the checks were age-specific.
At the end of each survey month, the measurements made by each interviewer and nurse were inspected. Any problems (such as higher than average proportions of measurements not obtained, insufficient samples and so on) were discussed with the relevant nurse or interviewer by their supervisor.
Ethical approval for the 2003 survey was obtained from the London Multi-centre Research Ethics Committee (MREC). All relevant Local Research Ethics Committees were informed about the survey.
6.1 Introduction to response analysis
The sample design, described in Section 2, requires all adults and up to two children per household to be interviewed. Non-informants to the survey fall into two groups, those living in households where no-one co-operated with the survey and those living in households where at least one person was interviewed.
This section looks first at the response of sampled households, then at that of eligible individuals within these households, and then at variation in response by region and dwelling type.
As informants were asked to co-operate in a sequence of operations, beginning with a face-to-face interview, progressing to a nurse visit and ending with requests for saliva and blood samples, individual non-response accumulated through the survey stages. Individual response for adults and children is looked at in two ways: overall response by all eligible individuals and response by individuals within co-operating households.
Not every measurement obtained by an interviewer or a nurse was subsequently considered valid for analysis purposes. Full details of the numbers of measurements used for analysis, the number of exclusions and the reasons for them are given at the start of each relevant chapter.
Table 1 shows household response by calendar quarter.
The row labelled 'Total eligible households' shows the number of private residential households found at the selected addresses (after selection of three households, if more than three were found).
Households described as 'co-operating' are those where at least one eligible person was interviewed at Stage 1. Households described as 'all interviewed' are those where all eligible persons were interviewed, and 'fully co-operating' are those where all eligible persons were interviewed, had height and weight measured and agreed to a nurse visit. Households where an informant was ineligible for a height or weight measurement because of a functional impairment or pregnancy are not counted as fully co-operating for this response analysis.
73% of eligible households (8,867) took part in the 2003 survey. At 61% of households all eligible adults and children were interviewed. Table 1
6.3 Individual response - adults
6.3.1 Overall response
There were 14,836 individual interviews with adults, 11,408 saw a nurse and 8,552 gave a blood sample.
The numerator of the response rate - the number of productive outcomes (though as already noted, there are a variety of different outcomes) - is known. However, the denominator - the total number of adults in the sampled households - is not known and must be estimated. There are three groups of households to consider: co-operating households (16,399 adults in 8,867 households, average 1.85), non co-operating households where information on the number of adults is known (4,162 adults in 2,346 households, average 1.77) and non co-operating households about which nothing is known (965 households). The most reasonable assumption is to impute to the last group the same average number of adults (1.83) as for all households where the number is known (the sum of the first two groups). This assumption gives an estimated total of 22,330 adults (the 'set' sample). Evidence suggests that unproductive households tend to be smaller on average than productive households, so this estimate of the total number of eligible adults is likely to be too large, and response rates based on it will be underestimates.
A further assumption is needed to provide separate 'set' samples for men and women. In non co-operating households where the number of adults was known, the numbers of men and women were not usually obtained. However, it can be assumed that the proportion of men and women in the estimated total sample is the same as for the adults in the 8,867 co-operating households. The proportions are 46.9% men and 53.1% women. Applying these proportions to the estimated total of adults gives 'set' samples of 10,475 men and 11,855 women.
Using the estimated total number of adults in sampled households (the adult 'set' sample) as a denominator, minimum response rates at various stages were:
66% were interviewed
62% had their height measured
60% had their weight measured
51% saw a nurse
49% had their waist and hip circumferences measured
50% had their blood pressure measured
40% agreed to give a blood sample
38% gave a blood sample
5% of those seeing a nurse did not give a blood sample, either because it was not possible to obtain a sample from them, for example because of collapsed veins (2%), or because they were ineligible to give blood (3%). Informants were ineligible because of pregnancy, drug treatment or a specified health condition.
Response to the interview was 69% among women and 63% among men.
Although a smaller proportion of men than women responded to the interview, among those interviewed, higher proportions of men than women agreed to see a nurse and agreed to have their blood pressure measured. Table 4
6.3.2 Adult response in co-operating households
As adults' ages and other personal characteristics are not known in non co-operating households, indications of response differences by these characteristics are confined to co-operating households. Tables 6 and 7 show the proportion of men and women, by age, in co-operating households who participated in the key survey stages. These are summarised below:
| |
|||||||||||
| Men | Women | All adults | |||||||||
| % | % | % | |||||||||
| |
|||||||||||
| Interviewed | 86 | 94 | 90 | ||||||||
| Height measured | 80 | 87 | 84 | ||||||||
| Weight measured | 79 | 84 | 82 | ||||||||
| Saw a nurse | 66 | 73 | 70 | ||||||||
| Waist and hip measured | 65 | 69 | 67 | ||||||||
| Blood pressure measured | 65 | 70 | 68 | ||||||||
| Agreed to give non-fasting blood | 53 | 57 | 55 | ||||||||
| Blood sample obtained | 51 | 53 | 52 | ||||||||
| Saliva sample givena | 65 | 66 | 65 | ||||||||
| Urine sample givena | 58 | 61 | 60 | ||||||||
| Agreed to give fasting blooda | 53 | 53 | 53 | ||||||||
| Fasting blood obtaineda | 44 | 45 | 44 | ||||||||
a These procedures were obtained only from the one-sixth subsample.
3% of men and 5% of women were ineligible for the non-fasting blood sample because of a clotting or other disorder or because of pregnancy. A non-fasting blood sample was obtained from 51% of eligible men and 53% of eligible women in co-operating households.
In co-operating households, response to the interview visit was highest among those aged 65 and over (95% of men and 97% of women) and lowest among those aged 16-24 (76% of men and 85% of women). Tables 6, 7
6.4 Individual response - children (0-15)
6.4.1 Overall response among children
Interviews were carried out with 3,717 children aged 0-15. 2,835 children were seen by a nurse.
Children (aged 0-15)
To compute the response rate, the number of eligible children in sampled households (the 'set sample') is needed as the denominator. This was estimated by assuming that the households where the numbers of children were not known had the same average number of boys and girls as those where it was known (and that the proportion of boys and girls was the same). This results in a 'set' sample of 5,073 children. This is likely to be an over-estimate, since non-contacted households have fewer children, on average, than those contacted. Response rates computed for children, like those for adults, are therefore conservative. Almost all non-responding children were in households where no-one (child or adult) co-operated with the survey. The total number of children in the sampled households would be slightly greater than the set sample as some households would have contained more than two children.
Response to the interview was 73% for both boys and girls. Height measurements were limited to those aged 2 and over, infant length measurements to those aged at least 6 weeks but under 2 years, blood pressure measurements to those aged 5-15 years and saliva samples to those aged 4-15. On the assumption that the age distribution of children in the 'set sample' is the same as that of children living in interviewed households, responses to these measurements were:
| |
|||||||||||
| Boys | Girls | All children | |||||||||
| % | % | % | |||||||||
| |
|||||||||||
| Interviewed | 73 | 73 | 73 | ||||||||
| Height measured (aged 2 and over) | 66 | 65 | 65 | ||||||||
| Weight measured | 65 | 63 | 64 | ||||||||
| Saw a nurse | 55 | 57 | 56 | ||||||||
| Infant length measured (aged 6 weeks to under 2 years) | 43 | 45 | 44 | ||||||||
| Blood pressure measured (aged 5 and over) | 52 | 54 | 53 | ||||||||
| Saliva sample given (aged 4 and over) | 49 | 50 | 50 | ||||||||
6.4.2 Response in co-operating households
Child response rates, like adult response rates, have also been calculated on a co-operating household base. The proportion of children aged 0-15 in co-operating households who were interviewed was high overall, at 96% of eligible boys and 97% of eligible girls (up to two children at each household). The proportion interviewed was lower among children aged 11-15 (92% of boys and 94% of girls) than among those aged under 11, where a parent acted as proxy informant (98% for boys and 99% for girls).
Tables 8 and 9 show the proportion of boys and girls, by age, in co-operating households who participated in the key survey stages. These are summarised below:
| |
|||||||||||
| Boys | Girls | All children |
|||||||||
| % | % | % | |||||||||
| |
|||||||||||
| Interviewed | 96 | 97 | 96 | ||||||||
| Height measured (aged 2 and over) | 86 | 86 | 86 | ||||||||
| Weight measured | 86 | 83 | 84 | ||||||||
| Saw a nurse | 72 | 75 | 74 | ||||||||
| Infant length measured (aged 6 weeks to under 2 years) | 57 | 59 | 58 | ||||||||
| Blood pressure measured (aged 5 and over) | 67 | 71 | 69 | ||||||||
| Saliva sample given (aged 4 and over) | 64 | 67 | 65 | ||||||||
The majority of children in all age groups co-operated with the measurements and provided a saliva sample. Tables 8, 9
6.5 Variations in survey response
6.5.1 Regional variations in response
As in previous years, response varied by region (Government Office Region). Household response was highest in the North East and East Midlands regions and was lowest in the London region. Table 2
6.5.2 Response by type of dwelling
Table 3 shows household response by the type of building housing the address (as classified by interviewers).
Response was highest among households living in detached (76%) or semi-detached houses (74%) and lowest among households living in other types of accommodation (57%) or flats, especially those living on higher floors (68%) or in a converted house (66%). Table 3
6.6 Age and sex profile of the sample
Tables 10 and 11 compare the age and sex profile of responding adults and children at the two survey stages (interview and nurse visit) with the mid-2003 population estimates. Adjusted estimates which exclude the institutional population aged 65 and over were used since the HSE sample covers only the private residential population.
According to the 2003 population estimates, men form 49% of all adults (aged 16 and over), while in HSE 2003 they form 44% of all interviewed adults. In HSE 2003 men aged under 35 are slightly under-represented at both interview and nurse visit relative to their proportions in the census population. Men aged 55 and over are slightly over-represented. Women aged under 25 are under-represented at both stages, while women aged 55 and over are over-represented. Table 10
As Table 11 shows, among children aged 0-15, both the sex and age profiles of the achieved HSE sample are very close to the population estimates for this age group. Table 11
The Health Survey for England has generally presented a good match to the population. Nevertheless, there are consistent discrepancies in some areas, such as the persistent over-representation of women. This has caused some concerns about non-response bias in the survey. In keeping with recent changes on many large-scale government sponsored surveys, and after a review of the options, a decision was taken to introduce non-response weighting to the HSE data, starting with the 2003 data, with the aim of reducing the possible biases.
Non-response weights have been calculated for both adults and children, but given that the focus of this report is adults, only the description of the derivation of the weights for adults is given here.
7.2 Overview of the weighting method
Four sets of non-response weights have been generated in total. Firstly a household weight was calculated to adjust for non-contact and for refusals of entire households. In addition, three sets of weights have been calculated to adjust (a) non-response among individuals in responding households (b) non-response to the nurse visit stage and (c) refusal to give a blood sample. The aim of each set of weights is that each of the main datasets (households, individuals, individuals who see a nurse, and individuals who give blood) can be treated as broadly representative of the general household population.
The weights are built up in stages. The main stages of the calculation are described in the following section.
7.3 Calculation of the weights
7.3.1 Household selection weight
In a selected address, a maximum of three households were selected for the sample. An inverse probability weight was calculated to adjust for addresses with more than three households per address.
7.3.2 Household non-response weight
To reduce household non-response bias, household level weights were calculated using a method known as 'calibration weighting'. These weights were applied separately within Government Office Region to bring the age and sex distribution of adults and children within responding households into line with each region's population age-sex distribution, but with the constraint that adults and children from the same household are all given the same weight. The rationale behind calibration weighting is that it attaches an estimated probability of response to each household that 'explains' any discrepancy between the survey age-sex distribution and the population age-sex distribution.
The population control totals used for this exercise were the ONS projected population estimates for 2003, but with a small adjustment to exclude (our best estimate of) the population aged 65 and over living in communal establishments. The tables below show the control totals used.
The final household weight (hhld_wt) is the product of household selection weight and the calibration weight.
2003 Projected household population, by age and sex
| |
|||||||||||
| Men | Women | ||||||||||
| N | N | ||||||||||
| |
|||||||||||
| 0-15 | 5,003,384 | 4,766,688 | |||||||||
| 16-24 | 2,840,079 | 2,767,721 | |||||||||
| 25-34 | 3,432,369 | 3,432,266 | |||||||||
| 35-44 | 3,774,967 | 3,827,094 | |||||||||
| 45-54 | 3,143,039 | 3,193,390 | |||||||||
| 55-64 | 2,775,546 | 2,856,902 | |||||||||
| 65-74 | 1,944,770 | 2,170,833 | |||||||||
| 75 and over | 1,347,799 | 2,089,106 | |||||||||
| Total | 2,426,1954 | 25,104,000 | |||||||||
These figures are adjusted to exclude those aged 65 and over living in institutions.
2003 Projected household population, by Government Office Region
| |
|||||||||||
| N | |||||||||||
| |
|||||||||||
| North East | 2,507,253 | ||||||||||
| North West | 6,733,110 | ||||||||||
| Yorkshire and the Humber | 4,970,957 | ||||||||||
| East Midlands | 4,199,464 | ||||||||||
| West Midlands | 5,300,263 | ||||||||||
| East England | 5,408,171 | ||||||||||
| London | 7,368,881 | ||||||||||
| South East | 7,963,033 | ||||||||||
| South West | 4,914,821 | ||||||||||
| Total | 49,365,953 | ||||||||||
These figures are adjusted to exclude those aged 65 and over living in institutions.
7.3.3 Individual level non-response weights
Three additional sets of weights were calculated for individual respondents to the survey to adjust for non-response at the interview stage, non-response to the nurse visit and refusal to give blood.
The calculation of these weights makes use of the fact that non-response at each of these stages is hierarchical: responding individuals are a sub-set of individuals from responding households; individuals who agree to a nurse visit are a sub-set of those individuals who respond to the survey; and individuals giving blood are a sub-set of those who agree to a nurse visit. So:
a) The probability of being a responding individual is estimated from the base of 'all individuals within responding households'. The individual weight is then calculated as the inverse of this probability multiplied by the household weight.
b) The probability of consenting to a nurse visit is estimated from the base of 'all responding individuals'. The nurse visit weight is then calculated as the inverse of this probability multiplied by the individual weight.
c) The probability of giving blood is estimated from the base of 'all individuals consenting to a nurse visit'. The 'blood weight' is then calculated as the inverse of this probability multiplied by the nurse visit weight.
The probability of response at each stage was estimated using a logistic regression model. The dependent variable in each case was whether the individual responded or not. The independent variables used in each model were age group, sex, Government Office Region, household type and the social class of the household reference person.9 The interaction between age and sex was also included. In addition, self-assessed general health (a health outcome) and cigarette smoking status (a health behaviour variable) were included in the models for non-response to the nurse visit and the blood sample.
This procedure generates three sets of weights, the interview weight (int_wt), the nurse weight (nurse_wt) and the blood weight (blood_wt). The table below shows the age and sex distribution after applying the interview weights (int_wt).This distribution matches the population distribution very closely.
Sample percentages after weighting responding adults, by age and sex
Men
Women
%
%
16-24
14.5
13.5
25-34
17.7
16.8
35-44
19.7
18.9
45-54
16.5
15.7
55-64
14.5
14.1
65-74
10.1
10.7
75 and over
7.0
10.3
An added complication in 2003 is that there were two types of nurse visit, a short and long nurse visit. Although everyone who took part in the interview was offered a nurse visit, informants in 120 randomly selected postcode sectors (one-sixth of the total) received an extended nurse visit with three extra procedures. When calculating the nurse weight for 2003, an additional step was taken to post-stratify the weights generated for the standard and longer nurse visit samples separately, so that both samples match the population age-sex distribution and Government Office Region distribution.
The table below shows the sample size, the range of the weights and their variance. The sample for the household weight includes all responding adults and children; the sample for the interview weight includes those aged 16 and over who responded to the interview. The samples for the nurse and blood weights also only include adults who responded to those stages.
Range of weights and variance
| |
|||||||||||
| N | Range | Variance | |||||||||
| |
|||||||||||
| Household weight | 18553 | 0.65-1.48 | 0.023 | ||||||||
| Interview weight | 14836 | 0.59-1.96 | 0.050 | ||||||||
| Nurse weight | 11408 | 0.37-3.40 | 0.137 | ||||||||
| Blood weight | 8552 | 0.33-3.97 | 0.181 | ||||||||
7.3.4 Trimming and scaling of the weights
Finally, the very smallest and largest of each set of weights were 'trimmed'. The 'small' weights were those below the first percentile and the 'large' weights were those above the 99th percentile. This trimming avoids the situation where some individuals have a very disproportionate influence on the survey estimates (either disproportionately large or disproportionately small). As a last step, each set of weights has been scaled by a constant factor so that the weighted sample size is the same as the unweighted sample size (so that the mean of the weights is equal to 1).
7.4 Effects of weighting on the HSE
7.4.1 General effects
The weighting should reduce non-response bias for many key HSE variables and this is the primary reason for their introduction. Other reasons include the fact that the post-stratification (calibration) to population age and sex totals should slightly reduce the standard errors of many estimates.
Nevertheless we are aware that applying weights from 2003 onwards will introduce a discontinuity in the HSE time series and will inhibit comparisons with published data from previous years. For this reason we have included unweighted figures in the 2003 report. The data sets from previous years will be retrospectively weighted using the same methodology and the revised data sets deposited with the data archive (during 2005). The trend tables on the Department of Health website will also be updated using weighting.
7.4.2 Impacts of weighting on key survey estimates
The tables below compare the unweighted estimates for some key survey variables with the weighted estimates using 2003 data. They show that there are small differences between the weighted and unweighted results. In general the differences for men are larger than the differences for women. For example, for men the percentage with a limiting longstanding illness is 25% unweighted and 23% weighted. Among women the corresponding unweighted figure is 27% and the weighted figure is 26%.
Looking at data on blood pressure, collected during the nurse visit, mean systolic blood pressure for men is 135.1 mmHg unweighted and 134.2 mmHg weighted. For women the figures are 130.1 mmHg unweighted and 129.4 mmHg weighted. For diastolic blood pressure the mean figures for men are 73.2 mmHg unweighted and 72.6 mmHg weighted and for women are 72.0 mmHg unweighted and 71.6 mmHg weighted.10
When looking at individual age groups the effects of weighting are small.
| |
|||||||||||
| Comparison of weighted and unweighted estimates (interview data) | |||||||||||
| Unweighted | Weighted | ||||||||||
| % | % | ||||||||||
| |
|||||||||||
| Men | |||||||||||
| Limiting longstanding illness | 25 | 23 | |||||||||
| Current cigarette smokers | 25 | 27 | |||||||||
| Percentage obese | 22.9 | 22.2 | |||||||||
| Drinking over recommended amount | 46 | 47 | |||||||||
| Diabetes prevalence | 4.8 | 4.3 | |||||||||
| IHD/Stroke in last year | 9.1 | 7.9 | |||||||||
| Women | |||||||||||
| Limiting longstanding illness | 27 | 26 | |||||||||
| Current cigarette smokers | 24 | 24 | |||||||||
| Percentage obese | 23.4 | 23.0 | |||||||||
| Drinking over recommended amount | 30 | 30 | |||||||||
| Diabetes prevalence | 3.6 | 3.4 | |||||||||
| IHD/Stroke in last year | 6.3 | 5.8 | |||||||||
| |
|||||||||||
| Unweighted Mean | SE | Weighted Mean | SE | ||||||||
| |
|||||||||||
| Men | |||||||||||
| Systolic blood pressure (mmHg) | 135.1 | .23 | 134.2 | .26 | |||||||
| Diastolic blood pressure (mmHg) | 73.2 | .16 | 72.6 | .20 | |||||||
| Total cholesterol (mmol/l) | 5.6 | .02 | 5.5 | .02 | |||||||
| Women | |||||||||||
| Systolic blood pressure (mmHg) | 130.1 | .25 | 129.4 | .29 | |||||||
| Diastolic blood pressure (mmHg) | 72.0 | .14 | 71.6 | .17 | |||||||
| Total cholesterol (mmol/l) | 5.7 | .02 | 5.6 | .02 | |||||||
HSE is a cross-sectional survey of the population. It examines associations between health states, personal characteristics and behaviour, but such associations do not necessarily imply causality. In particular, associations between current health states and current behaviour need careful interpretation, as current health may reflect past, rather than present, behaviour.
8.2 Weighted and unweighted data and bases in the report tables
Data presented for 2003 alone are weighted in this report. Where trend comparisons are being made with previous years the data for all previous years and 2003 are presented unweighted. In addition, in all trend tables weighted data for 2003 are presented in the right hand column or bottom row, adjacent to the unweighted data. There are thus two different estimates for the same variable in some of the 2003 data. Whenever tables include weighted data, both unweighted and weighted bases are shown. The unweighted bases show the number of cases involved. The weighted bases show the relative sizes of the various sample elements after weighting.
The absolute size of the weighted bases has no particular significance, since they have been scaled to the achieved sample size.
8.3 Defining age for data collection
A considerable part of the data collected in HSE 2003 is age specific, with different questions directed to different age groups. The informant's date of birth was ascertained. For data collection purposes, an informant's age was defined as their age on their last birthday before the interview. The nurse, who visited them later, treated them as being of the same age as at the interview, even if they had an intervening birthday.
8.4 Age as an analysis variable
Age is a continuous variable, and an exact age variable on the data file expresses it as such (so that, for example, someone whose 24th birthday was on January 1 2003 and was interviewed on October 1 2003 would be classified as being aged 24.75 (243/4)).
The presentation of tabular data involves classifying the sample into year bands. This can be done in two ways, age at last birthday and 'rounded age', that is, rounded to the nearest integer. In the present report all references to age are age at last birthday.
Age standardisation has been used in order to enable groups to be compared after adjusting for the effects of any differences in their age distributions. When different sub-groups are compared in respect of a variable on which age has an important influence, any differences in age distributions between these sub-groups are likely to affect the observed differences in the proportions of interest.
Age standardisation was carried out, using the direct standardisation method. The standard population to which the age distribution of sub-groups was adjusted was the mid-year 2001 census population estimates for England. The age-standardised proportion p´ was calculated as follows, where pi is the age specific proportion in age group i and Ni is the standard population size in age group i:
Therefore p´ can be viewed as a weighted mean of pi using the weights Ni. Age standardisation was carried out using the age groups: 16-24, 25-34, 35-44, 45-54, 55-64, 65-74 and 75 and over. The variance of the standardised proportion can be estimated by:
8.6 Comparisons with previous HSE reports
Topic chapters in Volume 1 and Volume 2 make comparisons with previous HSE reports. They always state in the text or table which years' data form the basis for the comparison. When making comparisons with previous years the data are unweighted (except that for 2003 the trend tables show weighted data alongside unweighted).
8.7 Design effects and true standard errors
HSE 2003 used a clustered, stratified multi-stage sample design. One of the effects of using this complex design is that standard errors for survey estimates are generally higher than the standard errors that would be derived from a simple random sample of the same size.
The ratio of the standard error of the complex sample to that of a simple random sample of the same size is known as the design factor. Put another way, the design factor (or 'deft') is the factor by which the standard error of an estimate from a simple random sample has to be multiplied to give the true standard error of the complex design.
The true standard errors and defts for HSE 2003 have been calculated using a Taylor Series expansion method. The deft values and true standard errors (which are themselves estimates subject to random sampling error) for selected survey estimates are shown in Tables 12-23. Tables 12 to 23
9.1 Blood pressure equipment change
In January 2003 a new type of blood pressure machine was introduced into HSE. The Omron HEM-907 replaced the Dinamap 8100. The new Omron machines had been selected after a feasibility study in which the following factors had been considered: ease of use, durability, ability to follow existing protocol, appropriateness for different age groups and weight. Full results of the feasibility study are presented in a paper available on the Department of Health website.11
The feasibility study was also designed to obtain comparative blood pressure readings using both the old and new equipment in order to assess the effects of the change on blood pressure results. 451 respondents had their blood pressure measured using both machines (103 aged 5-15, 146 aged 16-54 and 202 aged 55 and over). It was found that blood pressure readings (both systolic and diastolic) taken using the Omron HEM 907 were significantly higher than those taken using the Dinamap 8100.
HSE data are used for examining trends in blood pressure. It is important that results collected using the Omron can be compared with results from previous years using the Dinamap. Therefore calibration factors have been calculated allowing Omron results to be translated to the Dinamap equivalent and vice versa. These calibration factors have been calculated for adults using a linear regression model. Fuller details of the calibration are provided in a paper available on the Department of Health website.12 Separate calibration factors have also been calculated for children based on existing trends in blood pressure results rather than data from the experiment.
9.2 Inclusion of urine sampling for the first time
Spot urine samples were collected from a sub-sample of those aged 16 and over for the first time in 2003. A study was carried out to assess the comparability of mid-flow spot urine samples with 24 hour urine samples. This study was designed to test the suitability of a single spot urine sample collection for the determination of dietary sodium and potassium. The spot urine samples were compared against the 24-hour urine samples with regard to the sodium and potassium concentrations. Electrolyte excretion rates reflect the day-to-day diet of an individual. Therefore, unless the diet is very stable over time, the day-to-day variation in sodium (Na) and potassium (K) excretion within the same individual can be very large, often larger than the variation between groups in the population. Therefore a single sample is not enough to estimate accurately the true mean intake for an individual, but can be used in epidemiological studies to characterise and compare the intakes of groups of individuals.
Informants were selected from a sub-sample of those who had taken part in HSE 2003 and who consented to take part in the urine study. Overall, 360 adults (162 men, 198 women), aged 16 and over, completed the study and provided three spot urine samples and one 24-hour sample. Data collection took place between June 2003 and March 2004. After the second nurse visit, all samples were labelled and despatched to the laboratory at the Royal Victoria Infirmary (Newcastle) where the analyses were carried out. Following laboratory analysis, an aliquot of the 24-hour sample was sent to the MRC Dunn Human Nutrition laboratory (Cambridge) where the completeness of the 24-hour collection was assessed. This was achieved using the para-amino-benzoic acid (PABA) recovery method.
The results show that the correlation between 24-hour and spot urine samples for both potassium and sodium are weak, but significant and comparable with those observed in other populations. This finding supports the ability of a single spot urine sample to differentiate between subgroups of the population in the same way as the 24-hour sample. Full details of the comparison study and the results obtained are contained in a paper on the Department of Health website.13
10 Quality control of blood and saliva analytes
10.1 Introduction and key conclusions
This section describes the assay of analytes, quality control and quality assessment that were carried out during the survey period. Details of procedures used in the collection, processing and transportation of the specimens are described in Appendix B.
The overall conclusion for the data provided in this chapter is that methods and equipment used for the measurement of blood, saliva, and urine analytes produced internal quality control (IQC) and external quality assessment (EQA) results within expected limits. The results of the analyses for each of the main blood and urine analytes, and saliva cotinine levels were acceptable for the 2003 Health Survey for England.
The Royal Victoria Infirmary (RVI) in Newcastle upon Tyne was the main analysing laboratory used in the Health Survey for England 2003 for the blood and urine samples analysis. As in previous years, cotinine analysis for the 2003 Health Survey for England was conducted by the Nicotine Laboratory at New Cross Hospital.
10.3.1 Non-fasting blood samples
Following written consent from eligible informants, three non-fasting blood samples were collected for adults 16 and over (one 6 ml plain, one 4 ml EDTA and one 4.5 ml citrate tube). The order of priority for collecting samples was firstly into the 6 ml plain tube (no anticoagulant) followed by the 4 ml EDTA tube and the 4.5 ml citrate tube. After collection the tubes were despatched to the Department of Clinical Biochemistry at RVI who acted as the co-ordinating department for transport of samples to the individual departments undertaking the analysis. If written consent was given by the informant, approximately 1 ml of whole blood was stored in a freezer at -20°C ± 5°C for possible future analysis.
Samples collected in the 6 ml plain tube for serum (non-fasting)
This provided the sample for total cholesterol, HDL-cholesterol, and C-reactive protein analysis. If written consent was given by the informant, a minimum of 0.5 ml of the remaining serum was stored in a freezer at -70°C ± 5°C for possible future analysis.
Samples collected in the 4 ml EDTA (ethylene diamine tetra-acetic acid) tube
This provided the sample for the glycated haemoglobin analysis.
Samples collected in the 4.5 ml citrate tube
This provided the sample for fibrinogen analysis for informants aged 16 and over.
10.3.2 Fasting blood samples
After collection of the non-fasting blood sample, a sub-sample of adults aged 35 or over was asked to provide a fasting blood sample. Informants were asked not to consume food or drink, except water, in the eight hours previous to this sample being collected. The fasting sample was either collected at the same time as the non-fasting sample if all the criteria were met, or at a second nurse visit. Two fasting blood samples were collected, one 6 ml plain, and one 4 ml fluoride-oxalate. The order of priority for collecting samples was firstly into the 6 ml plain tube (no anticoagulant) followed by the 4 ml fluoride-oxalate tube.
Samples collected in the 6 ml plain (no anticoagulant) tube for serum (fasting sample)
This provided the sample for triglyceride measurement.
Samples collected in the 4 ml fluoride-oxalate (anticoagulant) tube for serum
This provided the sample for serum glucose measurement.
10.3.3 Saliva
A saliva sample was obtained from all informants aged 4 to 15 and a sub-sample of adults 16 and over. Saliva samples were collected for analysis of cotinine (a derivative of nicotine that shows recent exposure to tobacco smoke). A saliva collection tube was used for this purpose. Informants aged 16 and over were also offered the option to provide the saliva sample using a dental roll which was saturated with informants' saliva and was placed in the tube.
10.3.4 Urine
A mid-flow spot urine sample was obtained from a sub-sample of adults 16 and over, for analysis of sodium, potassium and creatinine. A special urine collection syringe was used for this purpose.
10.4.1 Laboratory procedures
The Department of Clinical Biochemistry at the RVI acted as the co-ordinating department for the transport of samples to a variety of individual departments within the RVI undertaking analysis for the 2003 Health Survey for England as well as New Cross in the case of saliva samples. All analyses were carried out according to Standard Operating Procedures by State Registered Medical Laboratory Scientific Officers (MLSOs) under the supervision of the Senior MLSO. All results were routinely checked by the duty Biochemist and highly abnormal results were immediately faxed to the Survey Doctor. Where prior consent was obtained, the Survey Doctor notified and advised the informant and their General Practitioner as appropriate.
A schedule of Planned Preventative Maintenance was used for each item of analytical equipment. These plans were carried out jointly by the manufacturers and the laboratories. Records were kept of when maintenance was due and carried out.
Table 24 shows reference ranges used by the various laboratories for each of the analytes measured in the 2003 Health Survey for England. Values within these reference ranges were considered to be clinically 'normal' while those outside were treated as clinically 'abnormal' (either too high or too low). Table 24
10.4.2 Non-fasting blood sample
Total cholesterol
Measurement of total cholesterol was carried out by the Biochemistry Department at RVI using the DAX Cholesterol Oxidase assay method on an Olympus 640 analyser calibrated to Centre for Disease Control (CDC) guidelines.
HDL-cholesterol
HDL-cholesterol analysis was carried out by the Biochemistry Department at RVI using a direct method (no precipitation) on an Olympus 640 analyser. In 1998 the HDL-cholesterol analysis was carried out after PTA precipitation. The 2003 data are therefore not directly comparable with those in 1998.
C-reactive Protein (CRP)
Measurement of CRP was carried out by the Biochemistry Department at RVI using the N Latex CRP mono Immunoassay on the Behring Nephelometer II Analyzer.
Glycated haemoglobin
Up to July 2003 measurement of glycated haemoglobin was carried out by the Haematology Department at RVI using the Tosoh HLC-723 (GHbV) A1c2.2 analyser. From August 2003 the analyser changed to the Tosoh G7.
Fibrinogen
Fibrinogen analysis was carried out by the Department of Haematology at RVI using the Organon Teknika MDA 180 analyser, using a modification of the Clauss thrombin clotting method.
10.4.3 Fasting blood sample
Triglycerides
Measurement of triglycerides was carried out by the Biochemistry Department at RVI using the Olympus 640 analyser using an enzymatic method involving cleavage of the molecule with lipoprotein lipase and measurement of the liberated glycerol using glycerolkinase with peroxidase/4-aminoantipyrine detection.
Glucose
Measurement of plasma glucose was carried out by the Biochemistry Department at RVI using the Hexokinase method on an Olympus 640 analyser.
LDL-cholesterol
LDL-cholesterol was calculated by the Biochemistry Department at RVI using measures of total cholesterol, HDL-cholesterol and triglycerides in the Friedewald formula. Triglyceride measurement must be below 4mmol/l for this calculation.
Friedewald formula states:
LDL-cholesterol = total cholesterol - (HDL-cholesterol + (triglycerides/2.2))
10.4.4 Urine sample
Urine sodium, potassium, and creatinine
All urine analytes were assayed on an Olympus 640 analyser. Urine sodium and potassium were analysed using the indirect ISE method. Urine Creatinine was analysed using the Jaffe method.
10.4.5 Saliva sample
Cotinine
Cotinine analysis was carried out by the Nicotine Laboratory at New Cross Hospital using a Hewlett Packard hp5890 gas chromatograph machine, with a rapid-liquid chromatography technique. This method has been used since 1993, although from 1993 to 1997 cotinine levels for adults were measured in serum. Saliva samples received at RVI are checked for correct identification, assigned a laboratory accession number and stored at 4°C. Samples are checked for details and despatched weekly in polythene bags (20 samples per bag) by courier for overnight delivery to New Cross.
10.5 Internal quality control (IQC)
10.5.1 Explanation of IQC
The purpose of IQC is to ensure reliability of an analytical run. IQC also helps to identify, and prevent the release of, any errors in an analytical run. IQC is also used to monitor trends over time.
For each analyte or group of analytes, the laboratory obtains a supply of quality control materials, usually at more than one concentration of analyte. Target (mean) value and target standard deviations (SD) are assigned for each analyte. Target assignment includes evaluation of values obtained by the laboratory from replicate measurements (over several runs) in conjunction with target values provided by manufacturers of IQC materials, if available. The standard deviation and the coefficient of variation (CV) are measures of imprecision and are presented here. Internal QC values are assessed against an acceptable range and samples are re-analysed if any of the Westgard rules14 have been violated.
10.5.2 Non-fasting blood sample
Total cholesterol
Low, Medium and High control materials are assayed at two hourly intervals. Table 25 shows the monthly internal quality control results for total cholesterol. Table 25
HDL-cholesterol
Table 26 shows the monthly internal quality control results for HDL-cholesterol. Table 26
C-reactive protein
Based on materials in use in the department, the Biochemistry department at RVI aim to achieve levels of reproducibility comparable to company literature, i.e. coefficient of variation in percent: <3%. However, realistically the imprecision at the low end of the analytical range is about 6%(CV). Table 27 shows the monthly internal quality control results for C-reactive protein. Table 27
Glycated haemoglobin
The analytical methods used for glycated haemoglobin measurement in the United Kingdom are required to be traceable to the work carried out on the Diabetes Control and Complications Trial (DCCT) part of the National Glycohemoglobin Standardisation Program (NGSP) in the USA. The Secondary Reference Laboratory (SRL) in the University of Minnesota was the main analytical laboratory for the DCCT work. The internal quality control results for glycated haemoglobin are DCCT standardised. Table 28
Fibrinogen
Control plasmas are assayed at regular intervals and instrument function tests are monitored continuously for fibrinogen with the control interval specified as every 4 hours or 100 specimens. Significant deviations from specified limits are flagged and must be acknowledged by the operator. Table 29 shows the monthly internal quality control results for two fibrinogen levels. Table 29
10.5.3 Fasting blood sample
Triglycerides
Three samples in the range low, medium and high are assayed. Table 30 shows the monthly internal quality control results for triglycerides. Table 30
Glucose
Three samples in the range low, medium and high are assayed. Table 31 shows the monthly internal quality control results for glucose. Table 31
LDL-cholesterol
The Clinical Biochemistry laboratory calculates the LDL-cholesterol values therefore no IQC is appropriate.
10.5.4 Urine sample
Sodium, potassium, creatinine
Urine samples were assayed twice a day (am and pm). The results of the monthly internal quality control results for sodium, potassium, and creatinine are shown in Tables 32, 33, and 34, respectively. Table 32, Table 33, Table 34
10.5.5 Saliva sample
Cotinine
The Medical Toxicology Unit at New Cross does participate in inter-laboratory split analyses to ensure comparable results. A summary of these monthly results for six levels of cotinine is presented in Table 35. Table 35
10.6 External quality assessment (EQA)
10.6.1 Explanation of EQA
The RVI Biochemistry and Haematology Departments and the New Cross Hospital laboratory both participate in external quality assessment (EQA) schemes.
EQA permits comparison of results between laboratories measuring the same analyte. An EQA scheme for an analyte or group of analytes distributes aliquots of the same samples to participating laboratories, which are blind to the concentration of the analytes. The usual practice is to participate in a scheme for a full year during which samples are distributed at regular frequency (monthly or bimonthly for example); the number of samples in each distribution and the frequency differ between schemes. The samples contain varying concentrations of analytes. The same samples may or may not be distributed more than once.
Samples are assayed shortly after they arrive at the laboratory. Depending on the frequency of distribution there may be weeks or months in which no EQA samples are analysed. Results are returned to the scheme organisers, who issue a laboratory specific report giving at least the following data:
- Mean values, usually for all methods and for method groups;
- A measure of the between-laboratory precision;
- The bias of the results obtained by that laboratory.
EQA is a retrospective process of assessment of performance, particularly of inaccuracy or bias with respect to mean values; unlike IQC, it does not provide control of release of results at the time of analysis.
The United Kingdom National External Quality Assessment Schemes (UKNEQAS) is a network of EQA schemes run by UK clinical laboratories. The Welsh External Quality Assessment Schemes (WEQAS), the Coulter Interlaboratory QA programme, National External Quality Assessment Scheme for Haematology, The Cambridge External Quality Assessment Schemes (EQAS) and the Central Quality Assessment Schemes (QAS) are all schemes to which the laboratories participate in on a routine basis. RIQAS (Randox International Quality Assurance Scheme) is an EQA scheme run by Randox Laboratories.
Each of the figures presented in Tables 36-46 corresponds with an individual EQA sample.
10.6.2 Non-fasting blood sample
Total cholesterol
The Clinical Biochemistry laboratory participates in UKNEQAS and WEQAS schemes. Table 36 shows the monthly external quality assessment results for total cholesterol. The target and achieved values are shown. Table 36
HDL-cholesterol
The Clinical Biochemistry laboratory participates in the WEQAS scheme. Table 37 shows the monthly external quality assessment results for HDL-cholesterol. The target and achieved values are shown. Table 37
C-reactive protein
Table 38 shows the monthly external quality assessment results for C-reactive protein. The target and achieved values are shown. Table 38
Glycated haemoglobin
Table 39 shows the monthly external quality assessment results for glycated haemoglobin. Table 39
Fibrinogen
The Haematology laboratory participates in Central QAS schemes fortnightly. Table 40 shows the external quality assessment results for fibrinogen for each quarter. Table 40
10.6.3 Fasting blood sample
Triglycerides
The Clinical Biochemistry laboratory participates in the WEQAS scheme. Table 41 shows the monthly external quality assessment results for triglycerides. The target and achieved values are shown. Table 41
Glucose
The Clinical Biochemistry laboratory participates in the WEQAS scheme. Table 42 shows the monthly external quality assessment results for glucose. The target and achieved values are shown. Table 42
LDL-cholesterol
Table 43 shows the monthly external quality assessment results for LDL-cholesterol. Table 43
10.6.4 Urine sample
The Clinical Biochemistry laboratory participates in the RIQAS scheme for the urine analytes (sodium, potassium, and creatinine). Tables 44, 45, and 46 show the monthly external quality assessment results for sodium, potassium, and creatinine. Tables 44, 45, 46
10.6.5 Saliva sample
Cotinine
There was no external quality control scheme available in 2003 to analyse cotinine but the Medical Toxicology Unit participates in inter-laboratory split analyses to ensure comparable results.