eSUPPLEMENT I: Survey methods and response to testing for C-reactive protein and fibrinogen in the Health Survey for England 1994, 1998 and 2003
Survey methods in the Health Survey for England
In each survey year of the Health Survey for England, adults were recruited using a multistage sampling strategy.1-3 The sampling frame was the Postcode Address File, which lists all postal addresses in the UK, including full postcodes, of which there are over a million in England, each containing between 1 and 80 addresses. British postcodes have two parts separated by a space, the first containing up to four letters or numbers denoting the postcode district and the second a number followed by two letters. In the Health Survey for England, the sampling unit was based on postcode sectors, which are denoted by the first part of the full postcode followed by the number that starts the second part. There are about 7600 postcode sectors in England, each containing around 2500 addresses, with areas ranging from less than 0.002 square kilometres (km2) in the centre of big cities to about 400km2 in sparsely populated rural areas, with a median area of about 4km2.
In 2003, postcode sectors with fewer than 500 addresses were combined with neighbouring sectors. In 1998, postcode sectors with fewer than 1000 addresses were combined with neighbouring sectors. Postcode sectors (or, where they were combined, pairs or triads of postcode sectors) were the sampling units in the Health Survey for England.
After stratification to ensure regional and socioeconomic balance (2003: by Local Authority and percentage of households with a head of household in a non-manual occupation; 1998: by health authority, percentage of households with a head of household in a non-manual occupation; 1994: regional health authority, size of population, proportion of households without a car, proportion of economically active males unemployed, proportion of non-white adults), 720 sampling units (made up of single or multiple postcode sectors as described above) were selected with a probability proportional to the number of addresses within the sector, giving each address an equal chance of being selected. From each selected postcode sector, 19 addresses (18 in 1994) were systematically drawn giving a total sample of 13,680 addresses (12,960 in 1994). Most addresses include only one household. All households living at selected addresses were invited to take part, unless they were addresses of businesses, institutions or were vacant, up to a total of three. If more than three households had the same address, three were randomly selected to take part.
Interviewers visited each selected household and asked an adult to complete a household questionnaire which enumerated the people living there. The interviewer then invited all adults aged 16 and over, up to a maximum of 10, living at each household to take part. If they agreed, they were invited to complete a individual health questionnaire. Co-operating households were defined as a households where at least one adult completed a health questionnaire. Adults aged 16 and over (18 and over in 1998) who completed a health questionnaire were offered a nurse visit to take a non-fasting blood sample unless they were pregnant, had a history of epilepsy, had a clotting disorder or were taking anticoagulants.
Household response
Table eI.1 shows household response in the Health Survey years of interest, adapted from tables provided in documentation by the National Centre for Social Research.1-3 Overall response by household was about 75%.
Little is known about non-responding households so it is not possible to compare characteristics of responding and non-responding households. Table eI.2 shows response by household by region of England.
Individual response within household
Table eI.3 shows the response to C-reactive protein and fibrinogen testing for the three Health Survey years of interest. About 90% of adults (defined as aged 16 and over in 1994 and 2003 and 18 and over in 1998) living in co-operating households completed an individual health questionnaire. Of the white adults who were eligible to provide a blood sample, 64% had a valid fibrinogen result and 66% a valid C-reactive protein result.
Detailed information is available about white adults who completed an individual questionnaire allowing us to compare the characteristics of those who did and did not have valid fibrinogen or C-reactive protein results. Tables eI.4 and eI.5 show the age group, sex and social class of head of household of people who do did and did not have results available on fibrinogen and C-reactive protein. People who had blood test results were more likely to be aged 36-74, more likely to be male and more likely to be of non-manual social class than people who did not.
Table eI.6 shows median exposure to the four pollutants of interest by whether or not a valid fibrinogen or C-reactive protein test was available. Exposure to PM10, NO2, SO2 was marginally higher and O3 marginally lower in people without valid fibrinogen and C-reactive protein results in 1998 and 2003 but we did not see this pattern in 1994.
References
(1) National Centre for Social Research, Department of Health. Health Survey for England 1994. 1996. London, The Stationery Office.
(2) National Centre for Social Research, Department of Health. Health Survey for England 1998 Cardiovascular Disease Volume 2: Methodology & Documentation. 1999. London, The Stationery Office.
(3) National Centre for Social Research and Department of Health. Health Survey for England 2003 Volume 3 Methodology and Documentation. Proston K, Primatesta P, editors. 2004. London, The Stationery Office.
8
eTable I.1. Household response in the Health Survey for England 1994, 1998 and 2003
1994 / 1998 / 2003 / AllSelected addresses / 12,960 / 13,680 / 13,680 / 40,320
Ineligible addresses (a) / 1,445 / 1,430 / 1,644 / 4,519
Eligible addresses / 11,515 / 12,250 / 12,036 / 35,801
Addresses withdrawn because no Local Research Ethics Committee approval / 126 / 0 / 0 / 126
Addresses at which interview sought / 11,389 / 12,250 / 12,036 / 35,675
Extra households sampled at multi-household addresses / 320 / 196 / 159 / 675
Total eligible households / 11,709 / 12,446 / 12,195 / 36,350
Co-operating households (b) (% of eligible households) / 9,068 (77.4) / 9,208 (74.0) / 8,867 (72.7) / 27,143 (74.7)
(a) Addresses where no private households were found.
(b) Households where at least one adult was interviewed.
eTable I.2. Household response by region in the Health Survey for England 1994, 1998 and 2003
1994 / Co-operating/ eligible households (%) / 1,323/
1,603
(82.5) / 1,145/
1,537 (74.5) / 967/
1,181
(81.9) / 916/
1,153 (79.4) / 999/
1,275
(78.4) / 1,198/
1,684 (71.1) / 1,250/
1,705 (73.3) / 1,270/
1,571
(80.8)
1998 / Co-operating/ eligible households (%) / 1,298/
1,610
(80.6) / 1,188/
1,664
(71.4) / 1,007/
1,286
(78.3) / 957/
1,319 (72.6) / 1,012/
1,359
(74.5) / 1,232/
1,740
(70.8) / 1,214/
1,782
(68.1) / 1,300/
1,686
(77.1)
North East / North West / Yorks and Humber / East Midlands / West Midlands / East England / London / South East / South West
2003 / Co-operating/ eligible households (%) / 561/
699
(80.3) / 1240/
1,720
(72.1) / 859/
1,188
(72.3) / 833/
1,045
(79.7) / 975/
1,289
(75.6) / 994/
1,367
(72.7) / 1,141/
1,719
(66.4) / 1,388/
1,968
(70.5) / 876/
1,200
(73.0)
eTable I.3. Response to C-reactive protein and fibrinogen testing in Health Survey for England 1994, 1998 and 2003
1994 / 1998 / 2003 / AllAdults living in co-operating households / 17,227 / 17,320 / 16,504 / 51,051
Adults aged 16 and overa who completed individual questionnaire (% of adults living in co-operating households) / 15,447 (89.7) / 15,908 (91.8) / 14,836 (89.9) / 46,191 (90.5)
White adults aged 16+a who completed questionnaire / 14,887 / 14,538 / 13,626 / 43,051
White adults aged 16+a ineligible to provide blood sample / 316 / 648 / 756 / 1,720
White adults aged 16+a eligible to provide blood sample / 14,571 / 13,890 / 12,870 / 41,331
White adults aged 16+a with valid fibrinogen result
(% of those eligible and completed questionnaire) / 9,678 (66.4) / 9,433 (67.9) / 7,399 (57.5) / 26,510 (64.1)
White adults aged 16+a with valid C-reactive protein result (% of those eligible and completed questionnaire) / Not applicable / 9,830 (70.8) / 7,736 (60.1) / 17,566 (65.6)
a aged 18+ in 1998
eTable I.4. Response to fibrinogen testing
Valid result / No valid result / Valid result / No valid result / Valid result / No valid result
Age (years) / 16-34* (%) / 2,863 (30) / 1,897 (36) / 2,383 (24) / 1,581 (34) / 1,512 (20) / 1,878 (30)
35-74 (%) / 6,089 (63) / 2,781 (53) / 6,522 (66) / 2,603 (55) / 5,203 (70) / 3,601 (58)
75+ (%) / 726 (8) / 531 (10) / 925 (9) / 524 (11) / 684 (9) / 748 (12)
Sex / Male (%) / 4,664 (48) / 2,085 (40) / 4,631 (47) / 1,919 (41) / 3,411 (46) / 2,635 (42)
Female (%) / 5,014 (52) / 3,124 (60) / 5,199 (53) / 2,789 (59) / 3,988 (54) / 3,592 (58)
Social class of
head of household / Non-manual (%) / 4,516 (47) / 2,263 (43) / 4,686 (50) / 2,444 (48) / 4,223 (57) / 3,256 (52)
Manual (%) / 4,725 (49) / 2,522 (48) / 4,570 (48) / 2,494 (49) / 3,047 (41) / 2,793 (45)
Not known (%) / 437 (5) / 424 (8) / 177 (2) / 167 (3) / 129 (2) / 178 (3)
*(18-34 in 1998)
Totals do not always add up to 100 because of rounding
eTable I.5. Response to C-reactive protein testing
1998 / 2003Valid result / No valid result / Valid result / No valid result
Age (years) / 16-34* (%) / 2,383 (24) / 1,581 (34) / 1,592 (21) / 1,798 (31)
35-74 (%) / 6,522 (66) / 2,603 (55) / 5,420 (70) / 3,384 (57)
75+ (%) / 925 (9) / 524 (11) / 724 (9) / 708 (12)
Sex / Male (%) / 4,631 (47) / 1,919 (41) / 3,550 (46) / 2,496 (42)
Female (%) / 5,199 (53) / 2,789 (59) / 4,186 (54) / 3,394 (58)
Social class of
head of household / Non-manual (%) / 4,885 (50) / 2,245 (48) / 4,393 (57) / 3,086 (52)
Manual (%) / 4,754 (48) / 2,310 (49) / 3,207 (41) / 2,633 (45)
Not known (%) / 191 (2) / 153 (3) / 136 (2) / 171 (3)
*(18-34 in 1998)
Totals do not always add up to 100 because of rounding
eTable I.6. Average exposure to pollutants by response to fibrinogen and C-reactive protein testing
Valid result / No valid result / Valid result / No valid result / Valid result / No valid result
Fibrinogen
Median PM10 (μg/m3) / 19.38 / 19.36 / 17.68 / 17.96 / 15.97 / 16.21
Median NO2(μg/m3) / 26.13 / 26.01 / 34.30 / 35.47 / 21.60 / 22.49
Median SO2 (μg/m3) / 9.35 / 9.60 / 6.16 / 6.29 / 4.08 / 4.19
Median O3 (μg/m3) / 53.58 / 53.48 / 51.10 / 50.71 / 56.12 / 55.91
C-reactive protein
Median PM10 (μg/m3) / 17.69 / 17.94 / 15.97 / 16.24
Median NO2(μg/m3) / 34.54 / 35.38 / 21.61 / 22.50
Median SO2 (μg/m3) / 6.20 / 6.22 / 4.09 / 4.16
Median O3 (μg/m3) / 51.08 / 50.72 / 56.10 / 55.94
8