- Email: [email protected]
Socioeconomic gradients in cardiorespiratory disease and diabetes in the 1960s: Baseline findings from the GPO study
ARTICLE IN PRESS Public Health (2006) 120, 685–695
www.elsevierhealth.com/journals/pubh
ORIGINAL RESEARCH
Socioeconomic gradients in cardiorespiratory disease and diabetes in the 1960s: Baseline findings from the GPO study J.E. Ferriea,, M.J. Shipleya, E. Breezea, G. Davey Smithb a
Department of Epidemiology and Public Health, University College London, 1-19 Torrington Place, London WC1E 6BT, UK b Department of Social Medicine, University of Bristol, Bristol, UK Received 9 September 2005; received in revised form 1 February 2006; accepted 15 February 2006 Available online 7 July 2006
KEYWORDS Cardiorespiratory disease; Diabetes; Risk factors; Socioeconomic distribution; 1960s
Summary Objectives: To describe the socioeconomic distribution of risk factors for cardiorespiratory disease and diabetes in employed women and men in the late 1960s. Study design and methods: Cross-sectional data were collected from 3345 General Post Office (GPO) employees in London, via a questionnaire and clinical examination, between October 1966 and April 1967. Results: Our occupational grade classification conformed to expected patterns of greater car ownership and gardening among higher-grade women and men, and greater height in higher-grade men (highest–lowest grade 175.0–170.7 cm, Po0.001). A strong inverse grade gradient in bronchitis (2.1–9.4%, Po0.001) and a strong positive gradient in FEV1 (3.10–2.58 l, Po0.001) were observed in men, although smoking was less consistently associated with grade. There was no consistent inverse association between grade and any cardiovascular risk factor in either sex, but strong inverse gradients in prevalence of impaired glucose tolerance (IGT) (5.1–18.2%, Po0.001) and 2-h glucose (4.14–4.25 mmol/l, Po0.001) in nondiabetic men. Using car ownership as an alternative measure of socioeconomic position, findings in men were replicated for respiratory measures, IGT and 2-h glucose prevalence. Inverse gradients were additionally observed for blood pressure, cholesterol and electrocardiogram abnormalities. Conclusions: The GPO study confirms existing evidence of socioeconomic gradients in respiratory risk factors and provides new evidence of gradients in risk factors for diabetes in men. Although there was no conclusive evidence of an occupational gradient in any cardiovascular risk factor, car ownership was a good indicator of lower risk in men. No socioeconomic gradients in cardiorespiratory or diabetic risk factors were observed in women. & 2006 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved.
Corresponding author. Tel.: +44 207 679 5643; fax: +44 207 813 0288.
E-mail address: [email protected] (J.E. Ferrie). 0033-3506/$ - see front matter & 2006 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.puhe.2006.02.010
ARTICLE IN PRESS 686
Introduction In 1968 Antonovsky1 reviewed 35 studies of cardiovascular mortality and 21 studies of morbidity. He concluded that for diseases of the circulatory system, disease of the heart, arteriosclerotic and degenerative heart disease, and coronary heart disease (CHD) the evidence did not substantiate the widely held view that during the first half of the 20th century men in higher socioeconomic groups were more affected by cardiovascular disease than those lower down. The number of studies that reported direct class gradients was equal to the number that reported inverse class gradients and both were outnumbered by studies showing no clear gradient. The conclusions of Antonovsky’s review were endorsed by a more recent review of well-characterized studies with standardized measures of disease prevalence or incidence, carried out in the UK or the USA, with initial recruitment up to and including 1960. With one exception, the studies showed either no association or a nonsignificant inverse association between socioeconomic position and disease prevalence or incidence.2 Currently, steep inverse socioeconomic gradients (higher status, lower morbidity) are observed for most major causes of morbidity and mortality in industrialized countries and, increasingly, in less developed countries.3–7 While the emergence of the socioeconomic gradient in cardiovascular disease among men in industrialized countries has been relatively well documented, less work has examined the socioeconomic gradient in cardiovascular risk factors among women, or diabetes and impaired glucose tolerance (IGT) in either sex. Determinants of the inverse socioeconomic gradient in respiratory disease, which affected both sexes,8 have similarly been relatively well documented among men,9 but not among women. There were no women in the first Whitehall study10 or the British Regional Heart study,11 the two studies that are the source of most of what is known about the development of socioeconomic gradients in cardiovascular disease in Britain. The only early UK study of cardiorespiratory disease to include a substantial proportion of women, the Renfrew and Paisley or Midspan study, was conducted in an urban area of Scotland with an unusually high level of socioeconomic deprivation.12 From 1661, when the first British postmark was introduced, the Post Office has been a national institution. During the 1960s the General Post Office (GPO) was a government department, part of the Civil Service. It had a monopoly of all mail, telegraph and telecommunications services in the
J.E. Ferrie et al. UK and was one of its largest employers.13 Over a period spanning the end of 1966 to the beginning of 1967 a cohort of women and men employed by the GPO and aged 15–73 underwent a clinical examination. This cohort was recruited and surveyed under the aegis of Harry Keen, Donald Reid and Geoffrey Rose, who went on, shortly after (1968–1970), to recruit and survey the 19 000 men included in the first Whitehall study of white-collar civil servants. The GPO study was designed as a pilot for Whitehall I. However, possibly due to the untimely death of Patrick Hamilton, one of the investigators, the proximity of the studies, or the sufficiency of the data furnished by Whitehall I, data from the GPO study have never been published. In this paper we report socioeconomic distributions of risk factors for cardiorespiratory disease and diabetes amongst women and men in the GPO study. Mortality follow-up is available for a 38-year period and will be reported in another paper.
Methods Study sample The target population for the GPO study was women and men of working age employed by the telecommunications arm of the GPO in central London in late 1966. Of the 4230 invited, 3345 women and men completed a short questionnaire and participated in a clinical examination, a response rate of 79%. Letters of invitation delivered to potential participants at their place of work contained a copy of the questionnaire which participants were asked to complete and bring with them to their clinical examination. Clinical examinations took place during working hours between October 1966 and April 1967 and covered both day and night shifts. Participants were flagged for mortality at the National Health Service Central Register.
Questionnaire measures Questionnaires were checked for completeness during the clinic visit by a trained interviewer who sought missing information. Personal details: age was determined from the participant’s date of birth and date of examination. Marital status was pre-coded as married, single, widowed, or divorced. Measures of socioeconomic position: occupation was divided into four main categories: manual, clerk/telephonist, supervisor, and engineer (men only). Alternative socioeconomic
ARTICLE IN PRESS Socioeconomic gradients in cardiorespiratory disease indicators were measured by questions on car ownership and gardening (In an average week in the summer do you work as much as 2 hours in the garden?). Health-related behaviour: questions covered current tobacco use, smoking history and physical activity at work. Respiratory symptoms: chronic bronchitis was assessed using the Medical Research Council’s chronic bronchitis questionnaire (short form).14 Cardiovascular symptoms: angina of effort (Rose angina), possible myocardial infarction and intermittent claudication were determined using questions developed at the London School of Hygiene and Tropical Medicine and approved by the World Health Organization for use in field studies.15 Possible myocardial infarction (MI) was defined as a history of severe pain across the front of the chest lasting half an hour or more.16 The remainder of the questionnaire covered: hospital admissions, treatment for heart disease or blood pressure, medication, diabetes, symptoms of diabetes and family history of diabetes. For married women and formerly married women only, the questionnaire elicited details of pregnancies.
Clinical measures Participants were requested to fast overnight or for at least fours hours before their examination. Height was measured to the nearest 1/4 inch (6.3 mm) below, with the participant’s shoes removed. Weight was recorded using a standard lever balance to the nearest 1/2 pound (227 g) below, without shoes and with the participant wearing blouse/shirt, skirt/trousers, and underwear.16 A single reading of blood pressure was measured on the left arm with the participant seated. The diastolic pressure was recorded at both muffling (phase 4) and disappearance of the Korotkoff sound (phase 5). Phase 4 readings were used in the analyses. To reduce observer bias and variation, observers were specially trained17 and used the London School of Hygiene sphygmomanometer.18 Lung function was assessed in participants over 35 years of age as forced expiratory volume in 1 second (FEV1) using the Garthur ‘Vitalograph’.10 Following standard instructions the process was repeated until three technically satisfactory tracings were obtained and the mean used for analysis. FEV1 is presented in litres and as a percentage of that predicted. Predicted values of FEV1 were obtained from linear regressions on age and height among the 159 women and 134 men who had never smoked and who responded ‘no’ to questions about wheeze, breathing, and asthma.19
687 Predicted FEV1 for women ¼ 1:733
ð0:0303 ageðyearsÞÞ þ ð0:0123 height ðcmÞÞ; Predicted FEV1 for men
¼ 2:861 ð0:0312 ageðyearsÞÞ þ ð0:0443 height ðcmÞÞ.
Electrocardiograms (ECG) were restricted to participants over 35 years of age and taken using limb leads only with a Mingograph 31B. Five technically adequate complexes were recorded for each of the six limb leads.10 Each ECG was interpreted independently by two trained coders, and an arbitrated ruling was given in cases of disagreement.16 The Minnesota code system was used throughout.20 ECGs with Q/QS items (codes 1.1–3), ST/T items (codes 4.1–4 or 5.1–3) or left bundle branch block (code 7.1) present were combined to form the category ECG abnormality. On entry to the clinic each participant was given a glucose preparation equivalent to an oral load of 50 g anhydrous dextrose (235 ml of ‘Lucozade’). Two hours after the glucose load, two samples of capillary blood were taken from the participant’s ear lobe using a new sterile needle. One sample of 0.1 ml was drawn into a clean dry pipette and added to 0.9 ml of a 1% solution of sodium fluoride. Blood glucose was estimated by the ferricyanide reduction micromethod on an autoanalyser (Technicon method N-9a). Impaired glucose tolerance (IGT) was defined as a blood glucose reading of 96–199 mg/dl.21 The diabetic group was a combination of known diabetics plus those with a blood glucose level of X200 mg/dl. A second sample of approximately 0.5 ml of capillary blood was collected into a heparinized capillary tube for estimation of plasma cholesterol concentration using the standard Technicon method N24a.10
Statistical analysis All analyses were age standardized by 5-year age groups. When comparing women and men, the total study population was used as the standard, whereas the total sex-specific study population was used as the standard for analyses of risk factors by occupational category within the two sex groups. Tests of differences and of trend were carried out by assessing the significance of sex and occupational category in linear regression models with age as a covariate for continuous outcomes and by Cochran–Mantel–Haenszel tests of association, adjusted for age group, for dichotomous outcomes. All data were analysed using SAS version 6.12 for Windows (SAS Institute Inc., 1990).
ARTICLE IN PRESS 688
J.E. Ferrie et al.
Results The distribution of sociodemographic factors among the 1251 women and 2094 men in the GPO study, and the distribution of cardiorespiratory symptoms and risk factors, are shown by sex in Tables 1 and 2, and by occupational grade in Table 3. The bimodal distribution of women by age reflects their absence from work during the childbearing and rearing years, and in later working life. Women were much less likely to be married than men and occupational grade was inversely associated with marriage in women although there was a strong direct association in men. The occupational range among women was very narrow. With the exception of 13 women classified as manual workers, women were either clerk/ telephonists or supervisors. Car ownership and gardening were much less common in women
Table 1
Distribution of sociodemographic factors and health-related behaviours Women (N ¼ 1251) n
Age at screening 15–19 20–29 30–39 40–49 50–59 60+ Marital status Married (%) Widowed (%) Single (%) Other (%) Occupational grade Engineer (highest) (%) Supervisor (%) Telephonist/clerk (%) Manual (lowest) (%) Owns a car (%) 2 h gardening in summer (%) Smoking Current cigarette (%) Pipe/cigar (only) (%) Pipe/cigar (ex-cigarette) (%) Ex-cigarette (%) Never-smoker (%) Work activity Mainly sitting (%) Mainly standing (%) Mainly moving (%) a
(Po0.001), but direct associations between grade and car ownership and gardening were observed in both sexes. Among women there was little evidence of an association between occupational grade and bronchitis, FEV1 or smoking. Although smoking rates are non-linear by occupational grade in men, there are strong inverse grade gradients in bronchitis and FEV1. The only significant differences between grades for cardiovascular indicators among women were higher levels of cholesterol and possible MI among the supervisors. In men grade was inversely associated with height and inversely associated with blood pressure, but not with any other cardiovascular risk factor. There was no firm evidence of differences in risk factors for diabetes by occupational grade in women but in men there were strong inverse gradients in IGT and 2-h glucose among non-diabetics.
%a (SE)
170 340 206 323 188 23
13.6 27.2 16.5 25.8 15.0 1.8
433 49 667 101
36.7 (1.4) 5.3 (0.7) 49.1 (1.3) 8.9 (0.8)
Men (N ¼ 2094) n 85 509 463 493 430 114
P value—difference or heterogeneity
%a (SE) 4.1 24.3 22.1 23.5 20.5 5.4
Po0.001
1348 30 674 40
61.3 (0.9) 1.27 (0.2) 35.7 (0.9) 1.8 (0.3)
(1.0) (1.0) (0.3) (1.1) (1.4)
904 230 883 69 937 984
45.9 (1.0) 10.5 (0.6) 40.6 (1.0) 3.0 (0.4) 44.9 (1.1) 45.7 (1.0)
600 0 0 144 506
48.4 (1.4) 0.0 0.0 12.0 (1.0) 39.6 (1.4)
1181 39 113 325 436
55.9 (1.0) 1.9 (0.3) 5.2 (0.5) 15.0 (0.8) 21.9 (0.9)
Po0.001
1084 20 144
85.1 (1.0) 1.8 (0.4) 13.1 (1.0)
1218 66 805
57.0 (1.0) 3.4 (0.4) 39.6 (1.1)
Po0.001
0 170 1067 13 224 384
All percentages are age-adjusted.
0.0 15.3 83.5 1.16 18.8 32.7
Po0.001
Po0.001
Po0.001 Po0.001
ARTICLE IN PRESS Socioeconomic gradients in cardiorespiratory disease Table 2
689
Distribution of biological risk factors for cardiorespiratory disease and diabetes Women (N ¼ 1251)
MRC bronchitis (%) FEV1b(l) % Predicted FEV1 Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Cholesterolc (mm/dl) Rose angina (%) Possible myocardial infarction (%) Intermittent claudication (%) ECG abnormalityb (%) Diabetes (%) Impaired glucose tolerance (%) 2-h glucose (non-diabetics only) (mmol/l) Height (cm) Body mass index (kg/m2) Obese (X30 kg/m2) (%)
n
Mean or %a (SE)
37
3.1 (0.5) 2.14 (0.02) 94.9 (0.8) 120.0 (0.5) 75.8 (0.4) 5.18 (0.05) 4.5 (0.6) 3.6 (0.6) 3.5 (0.5) 2.6 (0.7) 0.3 (0.2) 6.5 (0.7) 4.14 (0.02) 161.2 (0.2) 24.0 (0.10) 7.9 (0.8)
51 41 44 14 3 84
Men (N ¼ 2094) n 72
69 109 47 48 17 197
P value—difference
Mean or %a (SE) 3.2 (0.4) 3.06 (0.02) 92.4 (0.6) 129.5 (0.4) 79.0 (0.3) 5.04 (0.04) 3.1 (0.4) 4.9 (0.5) 2.2 (0.3) 3.8 (0.5) 0.7 (0.2) 9.2 (0.6) 4.25 (0.02) 174.3 (0.1) 24.1 (0.07) 4.6 (0.5)
P ¼ 0.96 Po0.001 P ¼ 0.01 Po0.001 Po0.001 P ¼ 0.01 P ¼ 0.04 P ¼ 0.08 P ¼ 0.03 P ¼ 0.16 P ¼ 0.13 P ¼ 0.01 Po0.001 Po0.001 P ¼ 0.25 Po0.001
MRC, Medical Research Council; FEV1, forced expiratory volume in 1 s; ECG, electrocardiogram. a All means and percentages are age-adjusted. b FEV1 and ECGs were measured only on those subjects aged 35 years or greater (644 women, 1274 men). c Cholesterol was only measured on a sub-sample of the cohort (596 women and 941 men), mainly those aged 35 years or more.
Alternative measures of socioeconomic position Car ownership and gardening were also examined as measures of socioeconomic position (Tables 4 and 5). Among women there were no significant inverse associations between these alternative measures of socioeconomic position and any cardiorespiratory or diabetes risk factor, although blood pressure and IGT prevalence were higher in women with no car access. In men there were strong associations between car ownership and FEV1, smoking, blood pressure, cholesterol, ECG abnormalities, IGT, 2-h glucose in non-diabetics, height and body mass index (BMI), and the indication of an association with bronchitis. Gardening was associated with a lower prevalence of smoking and possible MI, and with higher FEV1 and height. A measure that combined car ownership and gardening performed no better than car ownership alone.
Discussion Although intuitive, our occupational grade classification conforms to the expected pattern of higher car ownership and garden access among the highergrade occupations for both sexes. Height, a marker
of growth and socioeconomic position in childhood, as well as a marker of socioeconomic position in adulthood, also showed a steep occupational gradient in men. A strong inverse grade gradient was observed for bronchitis and a strong positive gradient in FEV1 in men, although smoking was less consistently associated with grade. There was no consistent inverse association between occupational grade and any cardiovascular risk factor in either sex and an inconsistent positive association between occupational grade and diabetes in men, although there were strong inverse associations between grade and IGT prevalence and 2-h glucose levels among non-diabetic men. Analyses using car ownership as an alternative measure of socioeconomic status, replicated the findings for respiratory and diabetes risk factors and demonstrated strong inverse associations with blood pressure, cholesterol and ECG abnormalities in men. The differences in marital status and car ownership between women and men supervisors are a reminder of labour market conditions in the mid1960s. Just over 20% of women supervisors were married compared to 73% of men supervisors, despite the rise of married women in the workforce during the 1960s from 30% to 42%.22 Until 1946 women employed by the Post Office and Civil Service had to resign on marriage, a factor likely
33.1 36.5 4.1 50.8 8.6 17.5 29.9 48.8 n/a n/a 11.6 39.6 95.3 0.2 4.5 3.2 95.7 118.2 74.2 5.1 4.4 2.6 3.3 2.2 0.3 7.0 4.1 161.5 23.7 6.9
47.9 21.1 1.8 71.2 5.8 21.8 37.9 42.5 n/a n/a 11.5 46.0 40.7 10.7 48.6 1.6 93.9 120.7 76.2 5.4 2.6 7.0 5.2 2.5 0.0 4.8 4.2 161.1 23.7 6.9 P ¼ 0.37 P ¼ 0.32 P ¼ 0.10 P ¼ 0.10 P ¼ 0.004 P ¼ 0.41 P ¼ 0.02 P ¼ 0.22 P ¼ 0.87 P ¼ 0.32 P ¼ 0.21 P ¼ 0.40 P ¼ 0.56 P ¼ 0.90 P ¼ 0.98
Po0.001
P ¼ 0.31
P ¼ 0.26 P ¼ 0.07
Po0.001
Po0.001
29.2 3.7 67.1 2.1 93.2 128.7 76.8 5.0 3.1 4.0 1.6 2.6 1.2 5.1 4.1 175.0 24.3 4.7
53.1 2.7 4.9 15.6 23.8
70.0 1.6 27.5 1.0 61.1 57.3
33.7
Mean or %b
46.1 6.1 47.8 1.7 94.1 130.6 80.6 5.0 6.0 10.3 2.1 5.1 1.8 10.3 4.3 174.4 24.2 2.0
48.9 1.8 7.6 20.8 21.0
73.1 0.6 25.8 0.5 50.4 57.9
47.5
Supervisor (n ¼ 230)
95.0 0.9 4.1 4.9 92.1 131.8 82.7 5.1 3.1 5.0 3.1 4.2 0.1 13.0 4.3 173.7 24.1 5.0
62.0 1.2 5.7 13.7 17.4
56.3 1.6 38.6 3.5 28.7 35.7
42.4
Clerk/ telephonist (n ¼ 883)
12.7 13.5 73.8 9.4 80.7 127.9 77.1 4.9 0.0 7.7 1.2 7.0 0.8 18.2 4.2 170.7 24.1 9.5
51.9 0.2 0.0 21.0 26.9
63.7 0.0 36.8 0.0 19.4 19.9
45.6
Manual (n ¼ 69)
Po0.001 Po0.001 P ¼ 0.007 Po0.001 P ¼ 0.34 P ¼ 0.46 P ¼ 0.38 P ¼ 0.10 P ¼ 0.17 P ¼ 0.02 Po0.001 Po0.001 Po0.001 P ¼ 0.26 P ¼ 0.27
Po0.001
P ¼ 0.001
Po0.001 Po0.001
P o0.001
Po0.001
P-value for trenda
690
MRC, Medical Research Council; FEV1, forced expiratory volume in 1 s; ECG, electrocardiogram. a All means and percentages are age-adjusted. b FEV1, cholesterol and ECGs were measured in a sub-sample, see footnote Table 2.; n/a: The 13 manual grade women were not included in the analyses.
Age at screening (years) Marital status Married (%) Widowed (%) Single (%) Other (%) Owns a car (%) 2 h gardening in summer (%) Smoking Current cigarette (%) Pipe/cigar only (%) Pipe/cigar (ex-cigarette) (%) Ex-cigarette (%) Never-smoker (%) Work activity Mainly sitting (%) Mainly standing (%) Mainly moving (%) MRC bronchitis (%) % Predicted FEVc1 Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Cholesterolc (mm/dl) Rose angina (%) Possible myocardial infarction (%) Intermittent claudication (%) ECG abnormalityc (%) Diabetes (%) Impaired glucose tolerance (%) 2-h glucose (non-diabetics) (mmol/l) Height (cm) Body mass index (kg/m2) Obese (X30 kg/m2) (%)
Mean or %b
Engineer (n ¼ 904)
P-value for difference
Supervisor (n ¼ 170)
Clerk/ telephonist (n ¼ 1067)
Men
Women
Distribution of sociodemographic factors and risk factors by occupational grade in women and men
Occupational grade
Table 3
ARTICLE IN PRESS J.E. Ferrie et al.
ARTICLE IN PRESS Socioeconomic gradients in cardiorespiratory disease Table 4
691
Distribution of sociodemographic factors and risk factors by car ownership and gardening in women. P-value for Gardening No Car No car P-value for (N ¼ 224) (N ¼ 1020) difference or (N ¼ 384) gardening difference or (N ¼ 863) heterogeneity heterogeneity Mean or %a
Age at screening (years) Marital status Married (%) Widowed (%) Single (%)
Mean or %a
39.1
34.4
Po0.001
40.4
32.9
Po0.001
64.1 1.8 29.8
28.0 4.4 58.7
Po0.001
38.4 4.2 49.8
32.9 3.8 55.1
P ¼ 0.24
4.3
8.9
7.6
8.2
— 42.5
— 28.1
25.2 —
14.9 —
Po0.001
46.4 11.6
48.3 11.6
49.3 9.7
47.4 12.4
P ¼ 0.42
Never-smoker (%) 42.0 Work activity Mainly sitting (%) 91.9 Mainly standing (%) 1.9 Mainly on the move (%) 6.2 MRC bronchitis (%) 3.8 % Predicted FEV1b 95.3 Systolic blood pressure (mmHg) 116.8 Diastolic blood pressure (mmHg) 73.5 Cholesterolb (mm/dl) 5.1 Rose angina (%) 3.0 Possible myocardial infarction (%) 5.1 Intermittent claudication (%) 4.3 ECG abnormalityb (%) 1.4 Diabetes (%) 0.0 Impaired glucose tolerance (%) 4.2 2-h glucose (non-diabetics only) (mmol/l) 4.1 Height (cm) 160.8 Body mass index (kg/m2) 24.3 Obese (X30 kg/m2) (%) 8.1
40.1
41.0
40.3
85.4 1.1 13.6 3.2 96.0 119.4 75.1 5.1 5.4 3.8 3.2 2.6 0.0 5.5 4.1 162.4 24.2 8.0
87.5 1.8 10.7 2.9 94.6 118.4 74.4 5.9 3.5 3.0 3.7 2.0 0.4 7.4 4.1 161.0 23.6 6.6
Other (%) Owns a car (%) Does 2 h gardening in summer (%) Smoking Current cigarette (%) Ex-cigarette (%)
85.8 1.5 12.7 2.8 95.2 119.2 74.8 5.2 4.3 2.9 3.3 2.5 0.3 7.4 4.1 161.5 23.6 6.9
Po0.001 P ¼ 0.87
P ¼ 0.04
P ¼ 0.50 P ¼ 0.95 P ¼ 0.06 P ¼ 0.18 P ¼ 0.41 P ¼ 0.39 P ¼ 0.11 P ¼ 0.44 P ¼ 0.43 P ¼ 0.28 P ¼ 0.09 P ¼ 0.57 P ¼ 0.12 P ¼ 0.02 P ¼ 0.58
P ¼ 0.30
P ¼ 0.80 P ¼ 0.41 P ¼ 0.37 P ¼ 0.41 P ¼ 0.65 P ¼ 0.15 P ¼ 0.50 P ¼ 0.73 P ¼ 0.62 P ¼ 0.13 P ¼ 0.23 P ¼ 0.90 Po0.001 P ¼ 0.006 P ¼ 0.41
MRC, Medical Research Council; FEV1, forced expiratory volume in 1 s; ECG, electrocardiogram. a All means and percentages are age-adjusted. b FEV1, cholesterol and ECGs were measured in a sub-sample, see footnote Table 2.
to have had a greater effect in this older group of women. Only 20% of women supervisors were car owners compared to 50% of men supervisors, although by 1967 about 40% of households had access to a car. During the 1960s average incomes increased, the retail prices index remained static, and unemployment was very low.22,23 However, the equal pay act did not come into force until 1970,24 and there was extensive gender segregation in the labour market. As clerical work became increasingly feminised—by 1966 70% of all clerical employees were women—clerical salaries fell relative to most other sectors.25 Thus, in this cohort car
ownership—a good indicator of spending power— appears to be a better marker of socioeconomic status than occupational grade. Contrary to numerous previous observations of steep, inverse social gradients in chronic bronchitis mortality for both sexes,8 no significant differences in bronchitis, lung function or smoking by socioeconomic position were seen in women. Few studies of the time report on differences in smoking among women; the Midspan study reported a steep inverse gradient in smoking but among German factory workers smoking was more prevalent in the highest occupational group.26 Steep inverse
ARTICLE IN PRESS 692 Table 5
J.E. Ferrie et al. Distribution of sociodemographic factors and risk factors by car ownership and gardening in men. P-value for Gardening No Car No car P-value for (N ¼ 937) (N ¼ 1157) difference or (N ¼ 984) gardening difference or (N ¼ 1104) heterogeneity heterogeneity Mean or %a
Age at screening (years) Marital status Married (%) Widowed (%) Single (%) Other (%) Owns a car (%) Does 2 h gardening in summer (%) Smoking Current cigarette (%) Pipe/cigar (%) Ex-cigarette (%)
Mean or %a
38.1
40.3
Po0.001
43.0
36.0
Po0.001
71.4 1.4 26.2 1.0 — 58.7
58.8 1.4 37.1 2.6 — 37.8
Po0.001
75.4 1.5 22.6 0.4 56.5 —
55.6 1.4 40.8 3.3 34.2 —
Po0.001
51.1 9.1 17.0
60.7 5.8 14.3
52.5 9.8 17.1
59.9 5.1 13.9
20.6
21.1
54.0 3.1 42.9 3.0 93.5 130.7 79.2 5.01 3.1 3.8 2.1 3.7 0.6 8.9 4.26 174.6 24.30 4.8
62.0 3.2 34.8 3.9 90.6 129.8 80.3 5.10 3.5 6.6 2.4 4.4 1.0 10.0 4.24 173.8 24.13 4.7
Never-smoker (%) 22.8 Work activity Mainly sitting (%) 52.1 Mainly standing (%) 3.2 Mainly on the move (%) 44.7 MRC bronchitis (%) 2.6 % Predicted FEV1b 94.5 Systolic blood pressure (mmHg) 129.1 Diastolic blood pressure (mmHg) 78.1 Cholesterolb (mm/dl) 5.13 Rose angina (%) 2.9 Possible myocardial infarction (%) 4.7 Intermittent claudication (%) 1.8 ECG abnormalityb (%) 2.5 Diabetes (%) 1.0 Impaired glucose tolerance (%) 7.9 2-hour glucose (non-diabetics only) (mmol/l) 4.20 Height (cm) 174.9 Body mass index (kg/m2) 24.39 Obese (X30 kg/m2) (%) 5.2
Po0.001
Po0.001
19.2 63.4 3.1 33.5 4.1 90.6 131.2 81.1 4.98 3.6 5.7 2.6 5.1 0.7 10.7 4.31 173.6 24.08 4.3
Po0.001 P ¼ 0.06 Po0.001 P ¼ 0.01 Po0.001 P ¼ 0.05 P ¼ 0.35 P ¼ 0.31 P ¼ 0.18 P ¼ 0.02 P ¼ 0.50 P ¼ 0.03 P ¼ 0.02 Po0.001 P ¼ 0.03 P ¼ 0.32
Po0.001
Po0.001
Po0.001 P ¼ 0.33 P ¼ 0.006 P ¼ 0.29 P ¼ 0.07 P ¼ 0.28 P ¼ 0.70 P ¼ 0.006 P ¼ 0.61 P ¼ 0.54 P ¼ 0.33 P ¼ 0.42 P ¼ 0.61 P ¼ 0.01 P ¼ 0.24 P ¼ 0.88
MRC, Medical Research Council; FEV1, forced expiratory volume in 1 s; ECG, electrocardiogram. a All means and percentages are age-adjusted. b FEV1, cholesterol and ECGs were measured in a sub-sample, see footnote Table 2.
gradients in cigarette smoking among men are observed in most studies from this period.12,27–29 However, a study of 1585 men working for a USA company in the late 1950s similarly observed an uneven distribution of smoking across the occupational hierarchy.30 The steepest part of the gradient in respiratory indicators lies between the clerk/ telephonist and manual grades, although a higher percentage of the former are ever-smokers. Other studies have observed that area of residence has an effect on bronchitis symptoms and FEV1 that is additional to exposure to tobacco smoke and which tends to increase with the number of cigarettes smoked.31 Also, a 1956 study of men in their 60s
found a strong, inverse social class gradient in chronic bronchitis among cigarette smokers, which was not explained by class variation in the amount of tobacco smoked.32 A greater proportion of men in the manual grades will have come from manual class backgrounds where exposure to air pollution and passive smoking both in childhood and adulthood are likely to have been higher.33,34 They may also have been heavier smokers and we know that filter tip use was lowest among them 38% compared with 65% among the clerk/ telephonists. With the exception of manual workers, physical activity at work was higher among GPO women and
ARTICLE IN PRESS Socioeconomic gradients in cardiorespiratory disease men in the higher grades due to the nature of their tasks. Almost without exception, studies have shown physical activity at work to be inversely associated with occupational grade, and research in the 1950s and 1960s proposed that the higher prevalence of CHD among men in higher social groups observed in some studies might be explained by differences in physical activity at work.35,36 Activity levels at work were similar for both engineers and manual staff who were also the groups with the lowest blood pressure and cholesterol, although manual men had much poorer respiratory and diabetes indicators than engineers. The direct association between occupational grade and cholesterol observed in GPO women reflects findings from contemporary studies with data for women,12,37 although an inverse gradient was observed among factory workers in Germany.26 When car ownership was used as the measure of socioeconomic position, the inverse association with blood pressure and the direct association with cholesterol in men replicates findings from most other studies of the time,27,29,32,36,38–40 although a study of 18,000 men in Oslo in 1972–1973 found an inverse social class gradient for serum cholesterol.28 No associations were observed between socioeconomic position and angina, possible MI, intermittent claudication or ischaemia for either sex. For women these observations reflect findings among Midspan women for angina and ischaemia.12 Although findings for occupational grade in men run counter to the socioeconomic gradients observed for angina and ischaemia in both Whitehall I and Midspan,12,27 the significant inverse association between car ownership and ischaemia more closely reflects these previous findings. For the first time in the late 1970s proposed diagnostic criteria for diabetes included the definition of impaired glucose tolerance.21 Using the new criteria, the prevalence of diabetes was lower, but IGT higher, among GPO men compared with Whitehall I men,41 findings which reflect the increase in diabetes with age and its inverse association with occupational grade.42 No significant associations were observed between markers of socioeconomic position and any diabetes indicator in women. However, the power of this study to detect differences is limited by the narrow occupational range and the large number of much younger women in the lower grade who are less likely to be affected by the wearing off of health-related selection into the labour force.43 There appear to be few data from the period on diabetes and IGT in women, although data from the Pittsburgh insulindependent diabetes mellitus registry for 1965–1976
693 showed no social class gradient in insulin-dependent diabetes in either sex.44 However, recent data have demonstrated an employment gradient in type 2 diabetes for women in the Whitehall II study and a strong inverse association between income and doctor-diagnosed diabetes for both sexes in the Health Survey for England.45,46 Among GPO men there was no consistent association between either occupational grade or car ownership and diabetes, although there were strong inverse associations between both socioeconomic measures and IGT and 2-h glucose. Similar results were found for both sexes in Midspan, although glucose levels were measured in non-fasting samples.12 Women students screened at the University of Glasgow between 1948 and 1968 were over 2 cm, and men students 0.3 cm, taller than GPO women and men. These differences reflect the higher socioeconomic position of students, especially women students, at a time when entry into tertiary education was more limited and so highly selective.47 Contrary to findings from Midspan, no inverse gradient between any measure of socioeconomic position and height was observed in women. However, in common with Midspan and Whitehall I, a steep, direct gradient in height was observed for all socioeconomic markers among men. There was no association between occupational grade and BMI among GPO participants of either sex. These findings run counter to the inverse gradient observed for BMI among women in previous studies and men in Whitehall I, but reflect the absence of a gradient in other studies, and in countries where the inverse socioeconomic gradient in CHD is still emerging.6,7,12,27,39 However, when using car ownership and gardening as measures of socioeconomic position, a significant positive association with BMI was observed, an unexpected finding given the lower 2-h glucose levels in these groups.
Methodological considerations Although the response rate at 79% was good, we have no information on the non-responders. As is well known, these employees are more likely to be of lower socioeconomic position and to be less healthy, whatever their socioeconomic position, than those who participate. Their exclusion will thus result in an underestimation of any socioeconomic gradient observed. We also know from study correspondence that some of the engineers were weekly paid and some salaried. Weekly paid engineers are likely to be lower down the occupational hierarchy than salaried engineers. However,
ARTICLE IN PRESS 694 as we have no way of distinguishing these two groups, all engineers have been classified in the highest grade. This differential misclassification may result in further underestimation of the socioeconomic gradients in men. Reports from other studies performed under the aegis of Donald Reid indicate that reliability and validity tests were performed on the measures used in this study. However, we have been unable to find data on these specific to the GPO study. The drawback of a study population recruited from a single public sector employer is that the occupational range of participants is more limited than in the populationbased observational studies carried out in the UK around the same time,12,32,39,40 although this homogeneity should have the advantage of reducing potential confounding due to occupational variation in future mortality analyses. Like any occupational-based cohort the GPO study will be subject to the healthy worker effect.48 However, it has the advantage of being one of the few studies of the period to include women and examine diabetes indicators. Examinations for the GPO study took place between October 1966 and April 1967, months when blood pressure is at its highest.49 This period also includes the coldest months of the year, which are characterized by high levels of morbidity and mortality from chronic bronchitis.50 However, this is unlikely to have affected our findings unless certain occupational grades were more likely to have been screened in the coldest months. Among the women, 95% of the supervisors and 94% of the clerk/telephonists were screened in October and November, while 84% of the men were screened in January, ranging from 94% of the engineers to 71% of the manual workers.
Acknowledgements The GPO study was supported by a grant to Donald Reid. The diabetes component of the study was funded by a grant to Harry Keen from the Chief Scientist Fund, Department of Health. J.E.F. is supported by the MRC (Grant number G8802774) and M.J.S. by a grant from the British Heart Foundation. We would like to thank David Leon for providing access to the original records of the study archived at the London School of Hygiene and Tropical Medicine and for his comments on an earlier draft of the paper. We also thank all the GPO employees who participated in the study; the Post Office management; the Post Office Trade Unions; the Treasury Medical Service; the researchers—
J.E. Ferrie et al. John Colley, Patrick Hamilton, Nan Keen, Harry Keen, Donald Reid, and Geoffrey Rose—who designed, set up and worked on the GPO study; Peter Shiarly for constructing the study database; and Walter Holland and Barbara Hunt who helped us track down people and information.
References 1. Antonovsky A. Social class and the major cardiovascular diseases. J Chronic Dis 1968;21:65–106. 2. Davey Smith G. Socioeconomic differentials. In: Kuh D, BenShlomo Y, editors. A lifecourse approach to chronic disease epidemiology. Oxford University Press: Oxford, UK; 1997. p. 242–73. 3. Davey Smith G, Leon D, Shipley MJ, Rose G. Socioeconomic differentials in cancer among men. Int J Epidemiol 1991;30: 339–45. 4. Davey Smith G, Egger M. Socioeconomic differences in mortality in Britain and the United States. Am J Pub Health 1992;82:1079–81. 5. Mackenbach JP, Cavelaars AE, Kunst AE, Groenhof F. Socioeconomic inequalities in cardiovascular disease mortality; an international study. Eur Heart J 2000;21:1141–51. 6. Duncan BB, Rumel D, Zelmanowicz A, Mengue SS, dos Santos S, Dalmaz A. Social inequality in mortality in Sao Paulo State, Brazil. Int J Epidemiol 1995;24:359–65. 7. Gupta R, Gupta VP, Ahluwalia NS. Educational status, coronary heart disease, and coronary risk factor prevalence in a rural population of India. Br Med J 1994;309:1332–6. 8. Goodman N, Lane RE, Rampling SB. Chronic bronchitis: an introductory examination of existing data. Br Med J 1953;2: 237–43. 9. Holland WW. Epidemiological surveys of chronic respiratory disease in England. MD thesis, University of London, 1964. 10. Reid DD, Brett GZ, Hamilton PJS, Jarrett RJ, Keen H, Rose G. Cardiorespiratory disease and diabetes among middle-aged male civil servants. Lancet 1974;i:469–73. 11. Pocock SJ, Shaper AG, Cook DG, Phillips AN, Walker M. Social class differences in ischaemic heart disease in British men. Lancet 1987;2:197–201. 12. Hawthorne VM, Watt GC, Hart CL, Hole DJ, Smith GD, Gillis CR. Cardiorespiratory disease in men and women in urban Scotland: baseline characteristics of the Renfrew/Paisley (Midspan) study population. Scott Med J 1995;40:102–7. 13. Daunton MJ. Royal mail: the post office since 1840. London, UK: The Athlone Press; 1985. 14. Medical Research Council’s Committee on the aetiology of chronic bronchitis. Standardized questions on respiratory symptoms. Br Med J 1960;ii:1665. 15. Rose GA. The diagnosis of ischaemic heart pain and intermittent claudication in field surveys. Bull World Health Organ 1962;27:645–58. 16. Reid DD, Holland WW, Humerfelt S, Rose G. A cardiovascular survey of British postal workers. Lancet 1966;i:614–8. 17. Rose G. Standardisation of observers in blood pressure measurement. Lancet 1965;i:673–4. 18. Rose GA, Holland WW, Crowley EA. A sphygmomanometer for epidemiologists. Lancet 1964;13:296–300. 19. Hole DJ, Watt GC, Davey-Smith G, Hart CL, Gillis CR, Hawthorne VM. Impaired lung function and mortality risk in men and women: findings from the Renfrew and Paisley prospective population study. Br Med J 1996;313:711–5.
ARTICLE IN PRESS Socioeconomic gradients in cardiorespiratory disease 20. Blackburn H, Keys A, Simonson E, Rautaharju P, Punsar S. The electrocardiogram in population studies. A classification system. Circulation 1960;21:1160–75. 21. Keen H, Jarrett RJ, Alberti KG. Diabetes mellitus: a new look at diagnostic criteria. Diabetologia 1979;16:283–5. 22. Coleman D, Salt J. The British population—patterns, trends, and processes. Oxford, UK: Oxford University Press; 1992. 23. Pullinger J, Summerfield T (Eds.) Social Trends 28. London, UK: The Stationery Office; 1998. 24. Reid I. Social class differences in Britain. London, UK: Open Books Publishing Limited; 1977. 25. Kelly P. White-collar proletariat: the industrial behaviour of British civil servants. London, UK: Routledge & Kegan Paul; 1980. 26. Schwalb H, van Eimeren W, Kunzlmann G, Strobl G. [Coronary risk factors and social class. Screening in employees of a large industrial firm (author’s transl.)]. Basic Res Cardiol 1978;73:97–125. 27. Rose G, Marmot MG. Social class and coronary heart disease. Br Heart J 1981;45:13–9. 28. Holme I, Helgeland A, Hjermann I, Lund-Larsen PG, Leren P. Coronary risk factors and socioeconomic status. The Oslo study. Lancet 1976;ii:1396–8. 29. De Becker G, Kornitzer M, Thilly C. Coronary risk factors and socioeconomic status. Lancet 1977;i:488. 30. Pell S, D’Alonzo CA. Blood pressure, body weight, serum cholesterol, and smoking habits among executives and nonexecutives. J Occup Med 1961;3:467–70. 31. Holland WW. The natural history of chronic bronchitis. J Coll Gen Pract 1966;11:8–16. 32. Brown RG, McKeown T, Whitfield AG. Observations on the medical condition of men in the seventh decade. Br Med J 1958;45:555–62. 33. Kiernan KE, Colley JR, Douglas JW, Reid DD. Chronic cough in young adults in relation to smoking habits, childhood environment and chest illness. Respiration 1976;33:236–44. 34. Jarvis MJ, Strachan DP, Feyerabend C. Determinants of passive smoking in children in Edinburgh, Scotland. Am J Public Health 1992;82:1225–9. 35. Morris JN, Crawford MD. Coronary heart disease and physical activity of work; evidence of a national necropsy survey. Br Med J 1958;30:1485–96. 36. Cassel JC. Review of 1960 through 1962 cardiovascular disease prevalence study. Arch Intern Med 1971;128:890–5.
695 37. Banerjee B, Saha N. Social class and serum-cholesterols. Lancet 1970;i:1113. 38. Syme SL, Oakes TW, Friedman GD, Feldman R, Siegelaub AB, Collen M. Social class and racial differences in blood pressure. Am J Public Health 1974;64:619–20. 39. Evans JG, Tunbridge WM. Blood pressure and social class. Public Health 1981;95:161–4. 40. Howell RW. Social class and biological tests. Lancet 1970;760:1343. 41. Fuller JH, Shipley MJ, Rose G, Jarrett RJ, Keen H. Coronary heart disease risk and impaired glucose tolerance: the Whitehall Study. Lancet 1980;1:1373–6. 42. Connolly V, Unwin N, Sherriff P, Bilous R, Kelly W. Diabetes prevalence and socioeconomic status: a population based study showing increased prevalence of type 2 diabetes mellitus in deprived areas. J Epidemiol Community Health 2000;54:173–7. 43. Bartley M, Ferrie JE. Glossary: unemployment, job insecurity and health. J Epidemiol Community Health 2001;55: 776–81. 44. LaPorte RE, Orchard TJ, Kuller LH, Wagener DK, Drash AL, Schneider BB, et al. The Pittsburgh Insulin Dependent Diabetes Mellitus Registry: the relationship of insulin dependent diabetes mellitus incidence to social class. Am J Epidemiol 1981;114:379–84. 45. Kumari M, Head J, Marmot M. Prospective study of social and other risk factors for incidence of type 2 diabetes in the Whitehall II study. Arch Intern Med 2004;164:1873–80. 46. Shelton N. Diabetes. In: Sproston K, Primatesta P, editors. Health Survey for England 2003: risk factors for cardiovascular disease. London, UK: The Stationery Office; 2004. p. 221–40. 47. McCarron P, Smith GD, Okasha M, McEwen J. Life course exposure and later disease: a follow-up study based on medical examinations carried out in Glasgow University (1948–68). Public Health 1999;113:265–71. 48. Carpenter L. Some observations on the healthy worker effect. Br J Ind Med 1987;44:289–91. 49. Brennan PJ, Greenberg G, Miall WE, Thompson SG. Seasonal variation in arterial blood pressure. Br Med J 1982;285:919–23. 50. Holland WW, Gilerdale S. Epidemiology of chronic bronchitis. In: Scadding JG, Cumming G, editors. Scientific foundations of respiratory medicine. London, UK: William Heinemann Medical Books Ltd; 1981. p. 18–29.
www.elsevierhealth.com/journals/pubh
ORIGINAL RESEARCH
Socioeconomic gradients in cardiorespiratory disease and diabetes in the 1960s: Baseline findings from the GPO study J.E. Ferriea,, M.J. Shipleya, E. Breezea, G. Davey Smithb a
Department of Epidemiology and Public Health, University College London, 1-19 Torrington Place, London WC1E 6BT, UK b Department of Social Medicine, University of Bristol, Bristol, UK Received 9 September 2005; received in revised form 1 February 2006; accepted 15 February 2006 Available online 7 July 2006
KEYWORDS Cardiorespiratory disease; Diabetes; Risk factors; Socioeconomic distribution; 1960s
Summary Objectives: To describe the socioeconomic distribution of risk factors for cardiorespiratory disease and diabetes in employed women and men in the late 1960s. Study design and methods: Cross-sectional data were collected from 3345 General Post Office (GPO) employees in London, via a questionnaire and clinical examination, between October 1966 and April 1967. Results: Our occupational grade classification conformed to expected patterns of greater car ownership and gardening among higher-grade women and men, and greater height in higher-grade men (highest–lowest grade 175.0–170.7 cm, Po0.001). A strong inverse grade gradient in bronchitis (2.1–9.4%, Po0.001) and a strong positive gradient in FEV1 (3.10–2.58 l, Po0.001) were observed in men, although smoking was less consistently associated with grade. There was no consistent inverse association between grade and any cardiovascular risk factor in either sex, but strong inverse gradients in prevalence of impaired glucose tolerance (IGT) (5.1–18.2%, Po0.001) and 2-h glucose (4.14–4.25 mmol/l, Po0.001) in nondiabetic men. Using car ownership as an alternative measure of socioeconomic position, findings in men were replicated for respiratory measures, IGT and 2-h glucose prevalence. Inverse gradients were additionally observed for blood pressure, cholesterol and electrocardiogram abnormalities. Conclusions: The GPO study confirms existing evidence of socioeconomic gradients in respiratory risk factors and provides new evidence of gradients in risk factors for diabetes in men. Although there was no conclusive evidence of an occupational gradient in any cardiovascular risk factor, car ownership was a good indicator of lower risk in men. No socioeconomic gradients in cardiorespiratory or diabetic risk factors were observed in women. & 2006 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved.
Corresponding author. Tel.: +44 207 679 5643; fax: +44 207 813 0288.
E-mail address: [email protected] (J.E. Ferrie). 0033-3506/$ - see front matter & 2006 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.puhe.2006.02.010
ARTICLE IN PRESS 686
Introduction In 1968 Antonovsky1 reviewed 35 studies of cardiovascular mortality and 21 studies of morbidity. He concluded that for diseases of the circulatory system, disease of the heart, arteriosclerotic and degenerative heart disease, and coronary heart disease (CHD) the evidence did not substantiate the widely held view that during the first half of the 20th century men in higher socioeconomic groups were more affected by cardiovascular disease than those lower down. The number of studies that reported direct class gradients was equal to the number that reported inverse class gradients and both were outnumbered by studies showing no clear gradient. The conclusions of Antonovsky’s review were endorsed by a more recent review of well-characterized studies with standardized measures of disease prevalence or incidence, carried out in the UK or the USA, with initial recruitment up to and including 1960. With one exception, the studies showed either no association or a nonsignificant inverse association between socioeconomic position and disease prevalence or incidence.2 Currently, steep inverse socioeconomic gradients (higher status, lower morbidity) are observed for most major causes of morbidity and mortality in industrialized countries and, increasingly, in less developed countries.3–7 While the emergence of the socioeconomic gradient in cardiovascular disease among men in industrialized countries has been relatively well documented, less work has examined the socioeconomic gradient in cardiovascular risk factors among women, or diabetes and impaired glucose tolerance (IGT) in either sex. Determinants of the inverse socioeconomic gradient in respiratory disease, which affected both sexes,8 have similarly been relatively well documented among men,9 but not among women. There were no women in the first Whitehall study10 or the British Regional Heart study,11 the two studies that are the source of most of what is known about the development of socioeconomic gradients in cardiovascular disease in Britain. The only early UK study of cardiorespiratory disease to include a substantial proportion of women, the Renfrew and Paisley or Midspan study, was conducted in an urban area of Scotland with an unusually high level of socioeconomic deprivation.12 From 1661, when the first British postmark was introduced, the Post Office has been a national institution. During the 1960s the General Post Office (GPO) was a government department, part of the Civil Service. It had a monopoly of all mail, telegraph and telecommunications services in the
J.E. Ferrie et al. UK and was one of its largest employers.13 Over a period spanning the end of 1966 to the beginning of 1967 a cohort of women and men employed by the GPO and aged 15–73 underwent a clinical examination. This cohort was recruited and surveyed under the aegis of Harry Keen, Donald Reid and Geoffrey Rose, who went on, shortly after (1968–1970), to recruit and survey the 19 000 men included in the first Whitehall study of white-collar civil servants. The GPO study was designed as a pilot for Whitehall I. However, possibly due to the untimely death of Patrick Hamilton, one of the investigators, the proximity of the studies, or the sufficiency of the data furnished by Whitehall I, data from the GPO study have never been published. In this paper we report socioeconomic distributions of risk factors for cardiorespiratory disease and diabetes amongst women and men in the GPO study. Mortality follow-up is available for a 38-year period and will be reported in another paper.
Methods Study sample The target population for the GPO study was women and men of working age employed by the telecommunications arm of the GPO in central London in late 1966. Of the 4230 invited, 3345 women and men completed a short questionnaire and participated in a clinical examination, a response rate of 79%. Letters of invitation delivered to potential participants at their place of work contained a copy of the questionnaire which participants were asked to complete and bring with them to their clinical examination. Clinical examinations took place during working hours between October 1966 and April 1967 and covered both day and night shifts. Participants were flagged for mortality at the National Health Service Central Register.
Questionnaire measures Questionnaires were checked for completeness during the clinic visit by a trained interviewer who sought missing information. Personal details: age was determined from the participant’s date of birth and date of examination. Marital status was pre-coded as married, single, widowed, or divorced. Measures of socioeconomic position: occupation was divided into four main categories: manual, clerk/telephonist, supervisor, and engineer (men only). Alternative socioeconomic
ARTICLE IN PRESS Socioeconomic gradients in cardiorespiratory disease indicators were measured by questions on car ownership and gardening (In an average week in the summer do you work as much as 2 hours in the garden?). Health-related behaviour: questions covered current tobacco use, smoking history and physical activity at work. Respiratory symptoms: chronic bronchitis was assessed using the Medical Research Council’s chronic bronchitis questionnaire (short form).14 Cardiovascular symptoms: angina of effort (Rose angina), possible myocardial infarction and intermittent claudication were determined using questions developed at the London School of Hygiene and Tropical Medicine and approved by the World Health Organization for use in field studies.15 Possible myocardial infarction (MI) was defined as a history of severe pain across the front of the chest lasting half an hour or more.16 The remainder of the questionnaire covered: hospital admissions, treatment for heart disease or blood pressure, medication, diabetes, symptoms of diabetes and family history of diabetes. For married women and formerly married women only, the questionnaire elicited details of pregnancies.
Clinical measures Participants were requested to fast overnight or for at least fours hours before their examination. Height was measured to the nearest 1/4 inch (6.3 mm) below, with the participant’s shoes removed. Weight was recorded using a standard lever balance to the nearest 1/2 pound (227 g) below, without shoes and with the participant wearing blouse/shirt, skirt/trousers, and underwear.16 A single reading of blood pressure was measured on the left arm with the participant seated. The diastolic pressure was recorded at both muffling (phase 4) and disappearance of the Korotkoff sound (phase 5). Phase 4 readings were used in the analyses. To reduce observer bias and variation, observers were specially trained17 and used the London School of Hygiene sphygmomanometer.18 Lung function was assessed in participants over 35 years of age as forced expiratory volume in 1 second (FEV1) using the Garthur ‘Vitalograph’.10 Following standard instructions the process was repeated until three technically satisfactory tracings were obtained and the mean used for analysis. FEV1 is presented in litres and as a percentage of that predicted. Predicted values of FEV1 were obtained from linear regressions on age and height among the 159 women and 134 men who had never smoked and who responded ‘no’ to questions about wheeze, breathing, and asthma.19
687 Predicted FEV1 for women ¼ 1:733
ð0:0303 ageðyearsÞÞ þ ð0:0123 height ðcmÞÞ; Predicted FEV1 for men
¼ 2:861 ð0:0312 ageðyearsÞÞ þ ð0:0443 height ðcmÞÞ.
Electrocardiograms (ECG) were restricted to participants over 35 years of age and taken using limb leads only with a Mingograph 31B. Five technically adequate complexes were recorded for each of the six limb leads.10 Each ECG was interpreted independently by two trained coders, and an arbitrated ruling was given in cases of disagreement.16 The Minnesota code system was used throughout.20 ECGs with Q/QS items (codes 1.1–3), ST/T items (codes 4.1–4 or 5.1–3) or left bundle branch block (code 7.1) present were combined to form the category ECG abnormality. On entry to the clinic each participant was given a glucose preparation equivalent to an oral load of 50 g anhydrous dextrose (235 ml of ‘Lucozade’). Two hours after the glucose load, two samples of capillary blood were taken from the participant’s ear lobe using a new sterile needle. One sample of 0.1 ml was drawn into a clean dry pipette and added to 0.9 ml of a 1% solution of sodium fluoride. Blood glucose was estimated by the ferricyanide reduction micromethod on an autoanalyser (Technicon method N-9a). Impaired glucose tolerance (IGT) was defined as a blood glucose reading of 96–199 mg/dl.21 The diabetic group was a combination of known diabetics plus those with a blood glucose level of X200 mg/dl. A second sample of approximately 0.5 ml of capillary blood was collected into a heparinized capillary tube for estimation of plasma cholesterol concentration using the standard Technicon method N24a.10
Statistical analysis All analyses were age standardized by 5-year age groups. When comparing women and men, the total study population was used as the standard, whereas the total sex-specific study population was used as the standard for analyses of risk factors by occupational category within the two sex groups. Tests of differences and of trend were carried out by assessing the significance of sex and occupational category in linear regression models with age as a covariate for continuous outcomes and by Cochran–Mantel–Haenszel tests of association, adjusted for age group, for dichotomous outcomes. All data were analysed using SAS version 6.12 for Windows (SAS Institute Inc., 1990).
ARTICLE IN PRESS 688
J.E. Ferrie et al.
Results The distribution of sociodemographic factors among the 1251 women and 2094 men in the GPO study, and the distribution of cardiorespiratory symptoms and risk factors, are shown by sex in Tables 1 and 2, and by occupational grade in Table 3. The bimodal distribution of women by age reflects their absence from work during the childbearing and rearing years, and in later working life. Women were much less likely to be married than men and occupational grade was inversely associated with marriage in women although there was a strong direct association in men. The occupational range among women was very narrow. With the exception of 13 women classified as manual workers, women were either clerk/ telephonists or supervisors. Car ownership and gardening were much less common in women
Table 1
Distribution of sociodemographic factors and health-related behaviours Women (N ¼ 1251) n
Age at screening 15–19 20–29 30–39 40–49 50–59 60+ Marital status Married (%) Widowed (%) Single (%) Other (%) Occupational grade Engineer (highest) (%) Supervisor (%) Telephonist/clerk (%) Manual (lowest) (%) Owns a car (%) 2 h gardening in summer (%) Smoking Current cigarette (%) Pipe/cigar (only) (%) Pipe/cigar (ex-cigarette) (%) Ex-cigarette (%) Never-smoker (%) Work activity Mainly sitting (%) Mainly standing (%) Mainly moving (%) a
(Po0.001), but direct associations between grade and car ownership and gardening were observed in both sexes. Among women there was little evidence of an association between occupational grade and bronchitis, FEV1 or smoking. Although smoking rates are non-linear by occupational grade in men, there are strong inverse grade gradients in bronchitis and FEV1. The only significant differences between grades for cardiovascular indicators among women were higher levels of cholesterol and possible MI among the supervisors. In men grade was inversely associated with height and inversely associated with blood pressure, but not with any other cardiovascular risk factor. There was no firm evidence of differences in risk factors for diabetes by occupational grade in women but in men there were strong inverse gradients in IGT and 2-h glucose among non-diabetics.
%a (SE)
170 340 206 323 188 23
13.6 27.2 16.5 25.8 15.0 1.8
433 49 667 101
36.7 (1.4) 5.3 (0.7) 49.1 (1.3) 8.9 (0.8)
Men (N ¼ 2094) n 85 509 463 493 430 114
P value—difference or heterogeneity
%a (SE) 4.1 24.3 22.1 23.5 20.5 5.4
Po0.001
1348 30 674 40
61.3 (0.9) 1.27 (0.2) 35.7 (0.9) 1.8 (0.3)
(1.0) (1.0) (0.3) (1.1) (1.4)
904 230 883 69 937 984
45.9 (1.0) 10.5 (0.6) 40.6 (1.0) 3.0 (0.4) 44.9 (1.1) 45.7 (1.0)
600 0 0 144 506
48.4 (1.4) 0.0 0.0 12.0 (1.0) 39.6 (1.4)
1181 39 113 325 436
55.9 (1.0) 1.9 (0.3) 5.2 (0.5) 15.0 (0.8) 21.9 (0.9)
Po0.001
1084 20 144
85.1 (1.0) 1.8 (0.4) 13.1 (1.0)
1218 66 805
57.0 (1.0) 3.4 (0.4) 39.6 (1.1)
Po0.001
0 170 1067 13 224 384
All percentages are age-adjusted.
0.0 15.3 83.5 1.16 18.8 32.7
Po0.001
Po0.001
Po0.001 Po0.001
ARTICLE IN PRESS Socioeconomic gradients in cardiorespiratory disease Table 2
689
Distribution of biological risk factors for cardiorespiratory disease and diabetes Women (N ¼ 1251)
MRC bronchitis (%) FEV1b(l) % Predicted FEV1 Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Cholesterolc (mm/dl) Rose angina (%) Possible myocardial infarction (%) Intermittent claudication (%) ECG abnormalityb (%) Diabetes (%) Impaired glucose tolerance (%) 2-h glucose (non-diabetics only) (mmol/l) Height (cm) Body mass index (kg/m2) Obese (X30 kg/m2) (%)
n
Mean or %a (SE)
37
3.1 (0.5) 2.14 (0.02) 94.9 (0.8) 120.0 (0.5) 75.8 (0.4) 5.18 (0.05) 4.5 (0.6) 3.6 (0.6) 3.5 (0.5) 2.6 (0.7) 0.3 (0.2) 6.5 (0.7) 4.14 (0.02) 161.2 (0.2) 24.0 (0.10) 7.9 (0.8)
51 41 44 14 3 84
Men (N ¼ 2094) n 72
69 109 47 48 17 197
P value—difference
Mean or %a (SE) 3.2 (0.4) 3.06 (0.02) 92.4 (0.6) 129.5 (0.4) 79.0 (0.3) 5.04 (0.04) 3.1 (0.4) 4.9 (0.5) 2.2 (0.3) 3.8 (0.5) 0.7 (0.2) 9.2 (0.6) 4.25 (0.02) 174.3 (0.1) 24.1 (0.07) 4.6 (0.5)
P ¼ 0.96 Po0.001 P ¼ 0.01 Po0.001 Po0.001 P ¼ 0.01 P ¼ 0.04 P ¼ 0.08 P ¼ 0.03 P ¼ 0.16 P ¼ 0.13 P ¼ 0.01 Po0.001 Po0.001 P ¼ 0.25 Po0.001
MRC, Medical Research Council; FEV1, forced expiratory volume in 1 s; ECG, electrocardiogram. a All means and percentages are age-adjusted. b FEV1 and ECGs were measured only on those subjects aged 35 years or greater (644 women, 1274 men). c Cholesterol was only measured on a sub-sample of the cohort (596 women and 941 men), mainly those aged 35 years or more.
Alternative measures of socioeconomic position Car ownership and gardening were also examined as measures of socioeconomic position (Tables 4 and 5). Among women there were no significant inverse associations between these alternative measures of socioeconomic position and any cardiorespiratory or diabetes risk factor, although blood pressure and IGT prevalence were higher in women with no car access. In men there were strong associations between car ownership and FEV1, smoking, blood pressure, cholesterol, ECG abnormalities, IGT, 2-h glucose in non-diabetics, height and body mass index (BMI), and the indication of an association with bronchitis. Gardening was associated with a lower prevalence of smoking and possible MI, and with higher FEV1 and height. A measure that combined car ownership and gardening performed no better than car ownership alone.
Discussion Although intuitive, our occupational grade classification conforms to the expected pattern of higher car ownership and garden access among the highergrade occupations for both sexes. Height, a marker
of growth and socioeconomic position in childhood, as well as a marker of socioeconomic position in adulthood, also showed a steep occupational gradient in men. A strong inverse grade gradient was observed for bronchitis and a strong positive gradient in FEV1 in men, although smoking was less consistently associated with grade. There was no consistent inverse association between occupational grade and any cardiovascular risk factor in either sex and an inconsistent positive association between occupational grade and diabetes in men, although there were strong inverse associations between grade and IGT prevalence and 2-h glucose levels among non-diabetic men. Analyses using car ownership as an alternative measure of socioeconomic status, replicated the findings for respiratory and diabetes risk factors and demonstrated strong inverse associations with blood pressure, cholesterol and ECG abnormalities in men. The differences in marital status and car ownership between women and men supervisors are a reminder of labour market conditions in the mid1960s. Just over 20% of women supervisors were married compared to 73% of men supervisors, despite the rise of married women in the workforce during the 1960s from 30% to 42%.22 Until 1946 women employed by the Post Office and Civil Service had to resign on marriage, a factor likely
33.1 36.5 4.1 50.8 8.6 17.5 29.9 48.8 n/a n/a 11.6 39.6 95.3 0.2 4.5 3.2 95.7 118.2 74.2 5.1 4.4 2.6 3.3 2.2 0.3 7.0 4.1 161.5 23.7 6.9
47.9 21.1 1.8 71.2 5.8 21.8 37.9 42.5 n/a n/a 11.5 46.0 40.7 10.7 48.6 1.6 93.9 120.7 76.2 5.4 2.6 7.0 5.2 2.5 0.0 4.8 4.2 161.1 23.7 6.9 P ¼ 0.37 P ¼ 0.32 P ¼ 0.10 P ¼ 0.10 P ¼ 0.004 P ¼ 0.41 P ¼ 0.02 P ¼ 0.22 P ¼ 0.87 P ¼ 0.32 P ¼ 0.21 P ¼ 0.40 P ¼ 0.56 P ¼ 0.90 P ¼ 0.98
Po0.001
P ¼ 0.31
P ¼ 0.26 P ¼ 0.07
Po0.001
Po0.001
29.2 3.7 67.1 2.1 93.2 128.7 76.8 5.0 3.1 4.0 1.6 2.6 1.2 5.1 4.1 175.0 24.3 4.7
53.1 2.7 4.9 15.6 23.8
70.0 1.6 27.5 1.0 61.1 57.3
33.7
Mean or %b
46.1 6.1 47.8 1.7 94.1 130.6 80.6 5.0 6.0 10.3 2.1 5.1 1.8 10.3 4.3 174.4 24.2 2.0
48.9 1.8 7.6 20.8 21.0
73.1 0.6 25.8 0.5 50.4 57.9
47.5
Supervisor (n ¼ 230)
95.0 0.9 4.1 4.9 92.1 131.8 82.7 5.1 3.1 5.0 3.1 4.2 0.1 13.0 4.3 173.7 24.1 5.0
62.0 1.2 5.7 13.7 17.4
56.3 1.6 38.6 3.5 28.7 35.7
42.4
Clerk/ telephonist (n ¼ 883)
12.7 13.5 73.8 9.4 80.7 127.9 77.1 4.9 0.0 7.7 1.2 7.0 0.8 18.2 4.2 170.7 24.1 9.5
51.9 0.2 0.0 21.0 26.9
63.7 0.0 36.8 0.0 19.4 19.9
45.6
Manual (n ¼ 69)
Po0.001 Po0.001 P ¼ 0.007 Po0.001 P ¼ 0.34 P ¼ 0.46 P ¼ 0.38 P ¼ 0.10 P ¼ 0.17 P ¼ 0.02 Po0.001 Po0.001 Po0.001 P ¼ 0.26 P ¼ 0.27
Po0.001
P ¼ 0.001
Po0.001 Po0.001
P o0.001
Po0.001
P-value for trenda
690
MRC, Medical Research Council; FEV1, forced expiratory volume in 1 s; ECG, electrocardiogram. a All means and percentages are age-adjusted. b FEV1, cholesterol and ECGs were measured in a sub-sample, see footnote Table 2.; n/a: The 13 manual grade women were not included in the analyses.
Age at screening (years) Marital status Married (%) Widowed (%) Single (%) Other (%) Owns a car (%) 2 h gardening in summer (%) Smoking Current cigarette (%) Pipe/cigar only (%) Pipe/cigar (ex-cigarette) (%) Ex-cigarette (%) Never-smoker (%) Work activity Mainly sitting (%) Mainly standing (%) Mainly moving (%) MRC bronchitis (%) % Predicted FEVc1 Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Cholesterolc (mm/dl) Rose angina (%) Possible myocardial infarction (%) Intermittent claudication (%) ECG abnormalityc (%) Diabetes (%) Impaired glucose tolerance (%) 2-h glucose (non-diabetics) (mmol/l) Height (cm) Body mass index (kg/m2) Obese (X30 kg/m2) (%)
Mean or %b
Engineer (n ¼ 904)
P-value for difference
Supervisor (n ¼ 170)
Clerk/ telephonist (n ¼ 1067)
Men
Women
Distribution of sociodemographic factors and risk factors by occupational grade in women and men
Occupational grade
Table 3
ARTICLE IN PRESS J.E. Ferrie et al.
ARTICLE IN PRESS Socioeconomic gradients in cardiorespiratory disease Table 4
691
Distribution of sociodemographic factors and risk factors by car ownership and gardening in women. P-value for Gardening No Car No car P-value for (N ¼ 224) (N ¼ 1020) difference or (N ¼ 384) gardening difference or (N ¼ 863) heterogeneity heterogeneity Mean or %a
Age at screening (years) Marital status Married (%) Widowed (%) Single (%)
Mean or %a
39.1
34.4
Po0.001
40.4
32.9
Po0.001
64.1 1.8 29.8
28.0 4.4 58.7
Po0.001
38.4 4.2 49.8
32.9 3.8 55.1
P ¼ 0.24
4.3
8.9
7.6
8.2
— 42.5
— 28.1
25.2 —
14.9 —
Po0.001
46.4 11.6
48.3 11.6
49.3 9.7
47.4 12.4
P ¼ 0.42
Never-smoker (%) 42.0 Work activity Mainly sitting (%) 91.9 Mainly standing (%) 1.9 Mainly on the move (%) 6.2 MRC bronchitis (%) 3.8 % Predicted FEV1b 95.3 Systolic blood pressure (mmHg) 116.8 Diastolic blood pressure (mmHg) 73.5 Cholesterolb (mm/dl) 5.1 Rose angina (%) 3.0 Possible myocardial infarction (%) 5.1 Intermittent claudication (%) 4.3 ECG abnormalityb (%) 1.4 Diabetes (%) 0.0 Impaired glucose tolerance (%) 4.2 2-h glucose (non-diabetics only) (mmol/l) 4.1 Height (cm) 160.8 Body mass index (kg/m2) 24.3 Obese (X30 kg/m2) (%) 8.1
40.1
41.0
40.3
85.4 1.1 13.6 3.2 96.0 119.4 75.1 5.1 5.4 3.8 3.2 2.6 0.0 5.5 4.1 162.4 24.2 8.0
87.5 1.8 10.7 2.9 94.6 118.4 74.4 5.9 3.5 3.0 3.7 2.0 0.4 7.4 4.1 161.0 23.6 6.6
Other (%) Owns a car (%) Does 2 h gardening in summer (%) Smoking Current cigarette (%) Ex-cigarette (%)
85.8 1.5 12.7 2.8 95.2 119.2 74.8 5.2 4.3 2.9 3.3 2.5 0.3 7.4 4.1 161.5 23.6 6.9
Po0.001 P ¼ 0.87
P ¼ 0.04
P ¼ 0.50 P ¼ 0.95 P ¼ 0.06 P ¼ 0.18 P ¼ 0.41 P ¼ 0.39 P ¼ 0.11 P ¼ 0.44 P ¼ 0.43 P ¼ 0.28 P ¼ 0.09 P ¼ 0.57 P ¼ 0.12 P ¼ 0.02 P ¼ 0.58
P ¼ 0.30
P ¼ 0.80 P ¼ 0.41 P ¼ 0.37 P ¼ 0.41 P ¼ 0.65 P ¼ 0.15 P ¼ 0.50 P ¼ 0.73 P ¼ 0.62 P ¼ 0.13 P ¼ 0.23 P ¼ 0.90 Po0.001 P ¼ 0.006 P ¼ 0.41
MRC, Medical Research Council; FEV1, forced expiratory volume in 1 s; ECG, electrocardiogram. a All means and percentages are age-adjusted. b FEV1, cholesterol and ECGs were measured in a sub-sample, see footnote Table 2.
to have had a greater effect in this older group of women. Only 20% of women supervisors were car owners compared to 50% of men supervisors, although by 1967 about 40% of households had access to a car. During the 1960s average incomes increased, the retail prices index remained static, and unemployment was very low.22,23 However, the equal pay act did not come into force until 1970,24 and there was extensive gender segregation in the labour market. As clerical work became increasingly feminised—by 1966 70% of all clerical employees were women—clerical salaries fell relative to most other sectors.25 Thus, in this cohort car
ownership—a good indicator of spending power— appears to be a better marker of socioeconomic status than occupational grade. Contrary to numerous previous observations of steep, inverse social gradients in chronic bronchitis mortality for both sexes,8 no significant differences in bronchitis, lung function or smoking by socioeconomic position were seen in women. Few studies of the time report on differences in smoking among women; the Midspan study reported a steep inverse gradient in smoking but among German factory workers smoking was more prevalent in the highest occupational group.26 Steep inverse
ARTICLE IN PRESS 692 Table 5
J.E. Ferrie et al. Distribution of sociodemographic factors and risk factors by car ownership and gardening in men. P-value for Gardening No Car No car P-value for (N ¼ 937) (N ¼ 1157) difference or (N ¼ 984) gardening difference or (N ¼ 1104) heterogeneity heterogeneity Mean or %a
Age at screening (years) Marital status Married (%) Widowed (%) Single (%) Other (%) Owns a car (%) Does 2 h gardening in summer (%) Smoking Current cigarette (%) Pipe/cigar (%) Ex-cigarette (%)
Mean or %a
38.1
40.3
Po0.001
43.0
36.0
Po0.001
71.4 1.4 26.2 1.0 — 58.7
58.8 1.4 37.1 2.6 — 37.8
Po0.001
75.4 1.5 22.6 0.4 56.5 —
55.6 1.4 40.8 3.3 34.2 —
Po0.001
51.1 9.1 17.0
60.7 5.8 14.3
52.5 9.8 17.1
59.9 5.1 13.9
20.6
21.1
54.0 3.1 42.9 3.0 93.5 130.7 79.2 5.01 3.1 3.8 2.1 3.7 0.6 8.9 4.26 174.6 24.30 4.8
62.0 3.2 34.8 3.9 90.6 129.8 80.3 5.10 3.5 6.6 2.4 4.4 1.0 10.0 4.24 173.8 24.13 4.7
Never-smoker (%) 22.8 Work activity Mainly sitting (%) 52.1 Mainly standing (%) 3.2 Mainly on the move (%) 44.7 MRC bronchitis (%) 2.6 % Predicted FEV1b 94.5 Systolic blood pressure (mmHg) 129.1 Diastolic blood pressure (mmHg) 78.1 Cholesterolb (mm/dl) 5.13 Rose angina (%) 2.9 Possible myocardial infarction (%) 4.7 Intermittent claudication (%) 1.8 ECG abnormalityb (%) 2.5 Diabetes (%) 1.0 Impaired glucose tolerance (%) 7.9 2-hour glucose (non-diabetics only) (mmol/l) 4.20 Height (cm) 174.9 Body mass index (kg/m2) 24.39 Obese (X30 kg/m2) (%) 5.2
Po0.001
Po0.001
19.2 63.4 3.1 33.5 4.1 90.6 131.2 81.1 4.98 3.6 5.7 2.6 5.1 0.7 10.7 4.31 173.6 24.08 4.3
Po0.001 P ¼ 0.06 Po0.001 P ¼ 0.01 Po0.001 P ¼ 0.05 P ¼ 0.35 P ¼ 0.31 P ¼ 0.18 P ¼ 0.02 P ¼ 0.50 P ¼ 0.03 P ¼ 0.02 Po0.001 P ¼ 0.03 P ¼ 0.32
Po0.001
Po0.001
Po0.001 P ¼ 0.33 P ¼ 0.006 P ¼ 0.29 P ¼ 0.07 P ¼ 0.28 P ¼ 0.70 P ¼ 0.006 P ¼ 0.61 P ¼ 0.54 P ¼ 0.33 P ¼ 0.42 P ¼ 0.61 P ¼ 0.01 P ¼ 0.24 P ¼ 0.88
MRC, Medical Research Council; FEV1, forced expiratory volume in 1 s; ECG, electrocardiogram. a All means and percentages are age-adjusted. b FEV1, cholesterol and ECGs were measured in a sub-sample, see footnote Table 2.
gradients in cigarette smoking among men are observed in most studies from this period.12,27–29 However, a study of 1585 men working for a USA company in the late 1950s similarly observed an uneven distribution of smoking across the occupational hierarchy.30 The steepest part of the gradient in respiratory indicators lies between the clerk/ telephonist and manual grades, although a higher percentage of the former are ever-smokers. Other studies have observed that area of residence has an effect on bronchitis symptoms and FEV1 that is additional to exposure to tobacco smoke and which tends to increase with the number of cigarettes smoked.31 Also, a 1956 study of men in their 60s
found a strong, inverse social class gradient in chronic bronchitis among cigarette smokers, which was not explained by class variation in the amount of tobacco smoked.32 A greater proportion of men in the manual grades will have come from manual class backgrounds where exposure to air pollution and passive smoking both in childhood and adulthood are likely to have been higher.33,34 They may also have been heavier smokers and we know that filter tip use was lowest among them 38% compared with 65% among the clerk/ telephonists. With the exception of manual workers, physical activity at work was higher among GPO women and
ARTICLE IN PRESS Socioeconomic gradients in cardiorespiratory disease men in the higher grades due to the nature of their tasks. Almost without exception, studies have shown physical activity at work to be inversely associated with occupational grade, and research in the 1950s and 1960s proposed that the higher prevalence of CHD among men in higher social groups observed in some studies might be explained by differences in physical activity at work.35,36 Activity levels at work were similar for both engineers and manual staff who were also the groups with the lowest blood pressure and cholesterol, although manual men had much poorer respiratory and diabetes indicators than engineers. The direct association between occupational grade and cholesterol observed in GPO women reflects findings from contemporary studies with data for women,12,37 although an inverse gradient was observed among factory workers in Germany.26 When car ownership was used as the measure of socioeconomic position, the inverse association with blood pressure and the direct association with cholesterol in men replicates findings from most other studies of the time,27,29,32,36,38–40 although a study of 18,000 men in Oslo in 1972–1973 found an inverse social class gradient for serum cholesterol.28 No associations were observed between socioeconomic position and angina, possible MI, intermittent claudication or ischaemia for either sex. For women these observations reflect findings among Midspan women for angina and ischaemia.12 Although findings for occupational grade in men run counter to the socioeconomic gradients observed for angina and ischaemia in both Whitehall I and Midspan,12,27 the significant inverse association between car ownership and ischaemia more closely reflects these previous findings. For the first time in the late 1970s proposed diagnostic criteria for diabetes included the definition of impaired glucose tolerance.21 Using the new criteria, the prevalence of diabetes was lower, but IGT higher, among GPO men compared with Whitehall I men,41 findings which reflect the increase in diabetes with age and its inverse association with occupational grade.42 No significant associations were observed between markers of socioeconomic position and any diabetes indicator in women. However, the power of this study to detect differences is limited by the narrow occupational range and the large number of much younger women in the lower grade who are less likely to be affected by the wearing off of health-related selection into the labour force.43 There appear to be few data from the period on diabetes and IGT in women, although data from the Pittsburgh insulindependent diabetes mellitus registry for 1965–1976
693 showed no social class gradient in insulin-dependent diabetes in either sex.44 However, recent data have demonstrated an employment gradient in type 2 diabetes for women in the Whitehall II study and a strong inverse association between income and doctor-diagnosed diabetes for both sexes in the Health Survey for England.45,46 Among GPO men there was no consistent association between either occupational grade or car ownership and diabetes, although there were strong inverse associations between both socioeconomic measures and IGT and 2-h glucose. Similar results were found for both sexes in Midspan, although glucose levels were measured in non-fasting samples.12 Women students screened at the University of Glasgow between 1948 and 1968 were over 2 cm, and men students 0.3 cm, taller than GPO women and men. These differences reflect the higher socioeconomic position of students, especially women students, at a time when entry into tertiary education was more limited and so highly selective.47 Contrary to findings from Midspan, no inverse gradient between any measure of socioeconomic position and height was observed in women. However, in common with Midspan and Whitehall I, a steep, direct gradient in height was observed for all socioeconomic markers among men. There was no association between occupational grade and BMI among GPO participants of either sex. These findings run counter to the inverse gradient observed for BMI among women in previous studies and men in Whitehall I, but reflect the absence of a gradient in other studies, and in countries where the inverse socioeconomic gradient in CHD is still emerging.6,7,12,27,39 However, when using car ownership and gardening as measures of socioeconomic position, a significant positive association with BMI was observed, an unexpected finding given the lower 2-h glucose levels in these groups.
Methodological considerations Although the response rate at 79% was good, we have no information on the non-responders. As is well known, these employees are more likely to be of lower socioeconomic position and to be less healthy, whatever their socioeconomic position, than those who participate. Their exclusion will thus result in an underestimation of any socioeconomic gradient observed. We also know from study correspondence that some of the engineers were weekly paid and some salaried. Weekly paid engineers are likely to be lower down the occupational hierarchy than salaried engineers. However,
ARTICLE IN PRESS 694 as we have no way of distinguishing these two groups, all engineers have been classified in the highest grade. This differential misclassification may result in further underestimation of the socioeconomic gradients in men. Reports from other studies performed under the aegis of Donald Reid indicate that reliability and validity tests were performed on the measures used in this study. However, we have been unable to find data on these specific to the GPO study. The drawback of a study population recruited from a single public sector employer is that the occupational range of participants is more limited than in the populationbased observational studies carried out in the UK around the same time,12,32,39,40 although this homogeneity should have the advantage of reducing potential confounding due to occupational variation in future mortality analyses. Like any occupational-based cohort the GPO study will be subject to the healthy worker effect.48 However, it has the advantage of being one of the few studies of the period to include women and examine diabetes indicators. Examinations for the GPO study took place between October 1966 and April 1967, months when blood pressure is at its highest.49 This period also includes the coldest months of the year, which are characterized by high levels of morbidity and mortality from chronic bronchitis.50 However, this is unlikely to have affected our findings unless certain occupational grades were more likely to have been screened in the coldest months. Among the women, 95% of the supervisors and 94% of the clerk/telephonists were screened in October and November, while 84% of the men were screened in January, ranging from 94% of the engineers to 71% of the manual workers.
Acknowledgements The GPO study was supported by a grant to Donald Reid. The diabetes component of the study was funded by a grant to Harry Keen from the Chief Scientist Fund, Department of Health. J.E.F. is supported by the MRC (Grant number G8802774) and M.J.S. by a grant from the British Heart Foundation. We would like to thank David Leon for providing access to the original records of the study archived at the London School of Hygiene and Tropical Medicine and for his comments on an earlier draft of the paper. We also thank all the GPO employees who participated in the study; the Post Office management; the Post Office Trade Unions; the Treasury Medical Service; the researchers—
J.E. Ferrie et al. John Colley, Patrick Hamilton, Nan Keen, Harry Keen, Donald Reid, and Geoffrey Rose—who designed, set up and worked on the GPO study; Peter Shiarly for constructing the study database; and Walter Holland and Barbara Hunt who helped us track down people and information.
References 1. Antonovsky A. Social class and the major cardiovascular diseases. J Chronic Dis 1968;21:65–106. 2. Davey Smith G. Socioeconomic differentials. In: Kuh D, BenShlomo Y, editors. A lifecourse approach to chronic disease epidemiology. Oxford University Press: Oxford, UK; 1997. p. 242–73. 3. Davey Smith G, Leon D, Shipley MJ, Rose G. Socioeconomic differentials in cancer among men. Int J Epidemiol 1991;30: 339–45. 4. Davey Smith G, Egger M. Socioeconomic differences in mortality in Britain and the United States. Am J Pub Health 1992;82:1079–81. 5. Mackenbach JP, Cavelaars AE, Kunst AE, Groenhof F. Socioeconomic inequalities in cardiovascular disease mortality; an international study. Eur Heart J 2000;21:1141–51. 6. Duncan BB, Rumel D, Zelmanowicz A, Mengue SS, dos Santos S, Dalmaz A. Social inequality in mortality in Sao Paulo State, Brazil. Int J Epidemiol 1995;24:359–65. 7. Gupta R, Gupta VP, Ahluwalia NS. Educational status, coronary heart disease, and coronary risk factor prevalence in a rural population of India. Br Med J 1994;309:1332–6. 8. Goodman N, Lane RE, Rampling SB. Chronic bronchitis: an introductory examination of existing data. Br Med J 1953;2: 237–43. 9. Holland WW. Epidemiological surveys of chronic respiratory disease in England. MD thesis, University of London, 1964. 10. Reid DD, Brett GZ, Hamilton PJS, Jarrett RJ, Keen H, Rose G. Cardiorespiratory disease and diabetes among middle-aged male civil servants. Lancet 1974;i:469–73. 11. Pocock SJ, Shaper AG, Cook DG, Phillips AN, Walker M. Social class differences in ischaemic heart disease in British men. Lancet 1987;2:197–201. 12. Hawthorne VM, Watt GC, Hart CL, Hole DJ, Smith GD, Gillis CR. Cardiorespiratory disease in men and women in urban Scotland: baseline characteristics of the Renfrew/Paisley (Midspan) study population. Scott Med J 1995;40:102–7. 13. Daunton MJ. Royal mail: the post office since 1840. London, UK: The Athlone Press; 1985. 14. Medical Research Council’s Committee on the aetiology of chronic bronchitis. Standardized questions on respiratory symptoms. Br Med J 1960;ii:1665. 15. Rose GA. The diagnosis of ischaemic heart pain and intermittent claudication in field surveys. Bull World Health Organ 1962;27:645–58. 16. Reid DD, Holland WW, Humerfelt S, Rose G. A cardiovascular survey of British postal workers. Lancet 1966;i:614–8. 17. Rose G. Standardisation of observers in blood pressure measurement. Lancet 1965;i:673–4. 18. Rose GA, Holland WW, Crowley EA. A sphygmomanometer for epidemiologists. Lancet 1964;13:296–300. 19. Hole DJ, Watt GC, Davey-Smith G, Hart CL, Gillis CR, Hawthorne VM. Impaired lung function and mortality risk in men and women: findings from the Renfrew and Paisley prospective population study. Br Med J 1996;313:711–5.
ARTICLE IN PRESS Socioeconomic gradients in cardiorespiratory disease 20. Blackburn H, Keys A, Simonson E, Rautaharju P, Punsar S. The electrocardiogram in population studies. A classification system. Circulation 1960;21:1160–75. 21. Keen H, Jarrett RJ, Alberti KG. Diabetes mellitus: a new look at diagnostic criteria. Diabetologia 1979;16:283–5. 22. Coleman D, Salt J. The British population—patterns, trends, and processes. Oxford, UK: Oxford University Press; 1992. 23. Pullinger J, Summerfield T (Eds.) Social Trends 28. London, UK: The Stationery Office; 1998. 24. Reid I. Social class differences in Britain. London, UK: Open Books Publishing Limited; 1977. 25. Kelly P. White-collar proletariat: the industrial behaviour of British civil servants. London, UK: Routledge & Kegan Paul; 1980. 26. Schwalb H, van Eimeren W, Kunzlmann G, Strobl G. [Coronary risk factors and social class. Screening in employees of a large industrial firm (author’s transl.)]. Basic Res Cardiol 1978;73:97–125. 27. Rose G, Marmot MG. Social class and coronary heart disease. Br Heart J 1981;45:13–9. 28. Holme I, Helgeland A, Hjermann I, Lund-Larsen PG, Leren P. Coronary risk factors and socioeconomic status. The Oslo study. Lancet 1976;ii:1396–8. 29. De Becker G, Kornitzer M, Thilly C. Coronary risk factors and socioeconomic status. Lancet 1977;i:488. 30. Pell S, D’Alonzo CA. Blood pressure, body weight, serum cholesterol, and smoking habits among executives and nonexecutives. J Occup Med 1961;3:467–70. 31. Holland WW. The natural history of chronic bronchitis. J Coll Gen Pract 1966;11:8–16. 32. Brown RG, McKeown T, Whitfield AG. Observations on the medical condition of men in the seventh decade. Br Med J 1958;45:555–62. 33. Kiernan KE, Colley JR, Douglas JW, Reid DD. Chronic cough in young adults in relation to smoking habits, childhood environment and chest illness. Respiration 1976;33:236–44. 34. Jarvis MJ, Strachan DP, Feyerabend C. Determinants of passive smoking in children in Edinburgh, Scotland. Am J Public Health 1992;82:1225–9. 35. Morris JN, Crawford MD. Coronary heart disease and physical activity of work; evidence of a national necropsy survey. Br Med J 1958;30:1485–96. 36. Cassel JC. Review of 1960 through 1962 cardiovascular disease prevalence study. Arch Intern Med 1971;128:890–5.
695 37. Banerjee B, Saha N. Social class and serum-cholesterols. Lancet 1970;i:1113. 38. Syme SL, Oakes TW, Friedman GD, Feldman R, Siegelaub AB, Collen M. Social class and racial differences in blood pressure. Am J Public Health 1974;64:619–20. 39. Evans JG, Tunbridge WM. Blood pressure and social class. Public Health 1981;95:161–4. 40. Howell RW. Social class and biological tests. Lancet 1970;760:1343. 41. Fuller JH, Shipley MJ, Rose G, Jarrett RJ, Keen H. Coronary heart disease risk and impaired glucose tolerance: the Whitehall Study. Lancet 1980;1:1373–6. 42. Connolly V, Unwin N, Sherriff P, Bilous R, Kelly W. Diabetes prevalence and socioeconomic status: a population based study showing increased prevalence of type 2 diabetes mellitus in deprived areas. J Epidemiol Community Health 2000;54:173–7. 43. Bartley M, Ferrie JE. Glossary: unemployment, job insecurity and health. J Epidemiol Community Health 2001;55: 776–81. 44. LaPorte RE, Orchard TJ, Kuller LH, Wagener DK, Drash AL, Schneider BB, et al. The Pittsburgh Insulin Dependent Diabetes Mellitus Registry: the relationship of insulin dependent diabetes mellitus incidence to social class. Am J Epidemiol 1981;114:379–84. 45. Kumari M, Head J, Marmot M. Prospective study of social and other risk factors for incidence of type 2 diabetes in the Whitehall II study. Arch Intern Med 2004;164:1873–80. 46. Shelton N. Diabetes. In: Sproston K, Primatesta P, editors. Health Survey for England 2003: risk factors for cardiovascular disease. London, UK: The Stationery Office; 2004. p. 221–40. 47. McCarron P, Smith GD, Okasha M, McEwen J. Life course exposure and later disease: a follow-up study based on medical examinations carried out in Glasgow University (1948–68). Public Health 1999;113:265–71. 48. Carpenter L. Some observations on the healthy worker effect. Br J Ind Med 1987;44:289–91. 49. Brennan PJ, Greenberg G, Miall WE, Thompson SG. Seasonal variation in arterial blood pressure. Br Med J 1982;285:919–23. 50. Holland WW, Gilerdale S. Epidemiology of chronic bronchitis. In: Scadding JG, Cumming G, editors. Scientific foundations of respiratory medicine. London, UK: William Heinemann Medical Books Ltd; 1981. p. 18–29.