Magnitude and causes of socioeconomic differentials in mortality: further evidence from the Whitehall Study.

STUDY OBJECTIVE: The aim was to explore the magnitude and causes of the differences in mortality rates according to socioeconomic position in a cohort of civil servants. DESIGN: This was a prospective observational study of civil servants followed up for 10 years after baseline data collection. SETTING: Civil service offices in London. PARTICIPANTS: 11,678 male civil servants were studied, aged 40-64 at baseline screening between 1967 and 1969. Two indices of socioeconomic position were available on these participants--employment grade (categorised into four levels), and ownership of a car. MEASUREMENTS AND MAIN RESULTS: Main outcome measures were all cause and cause specific mortality, with cause of death taken from death certificates coded according to the eighth revision of the ICD. Employment grade and car ownership were independently related to total mortality and to mortality from the major cause groups. Combining the indices further improved definition of mortality risk and the age adjusted relative rate between the highest grade car owners and the lowest grade non-owners of 4.3 is considerably larger than the social class differentials seen in the British population. Factors potentially involved in the production of these mortality differentials were examined. Smoking, plasma cholesterol concentration, blood pressure, and glucose intolerance did not appear to account for them. The pattern of differentials was the same in the group who reported no ill health at baseline as it was in the whole sample, which suggests that health selection associated with frank illness was not a major determinant. The contribution of height, a marker for environmental factors acting in early life, was also investigated. Whereas adjustment for employment grade and car ownership attenuated the association between short stature and mortality, height differences within employment grade and car ownership groups explained little of the differential mortality. CONCLUSION: The use of social class as an index of socioeconomic position leads to underestimation of the association between social factors and mortality, which may be reflected in public health initiatives and priorities. Known risk factors could not be shown to account for the differentials in mortality, although the degree to which this can be explored with single measurements is limited.