Occupation-related suicide: The impact of the Census Act and the Population (Statistics) Act on research based on individual-level mortality data
, MRes thesis, University of Salford.
A comprehensive review of the literature reveals a deficit in explanation for the differences in suicide-risk between occupations. Two studies of occupation-related suicide in England and Wales (Kelly et al, 1995; Kelly & Bunting, 1998) are unusual in reporting suicide-data for multiple occupations. These studies rank a series of occupations according to their associated suicide risk, which is measured in terms of the Provisional Mortality Ratio for suicide. The studies do not incorporate controls for demographic covariates of occupation, apart from gender. An American study (Stack, 2001), by contrast, is based on a logistic regression model of analysis, with controls introduced for age, gender, marital status, and race. Comparing the bivariate logistic regression odds ratios with the respective multivariate logistic odds ratios for a series of occupations demonstrates the importance of incorporating demographic controls in the assessment of occupational risk of suicide.
Examination of the overlap between the findings of these studies demonstrates a series of inconsistencies, and suggests that this may partly due to different methodologies. This highlights the importance of rendering studies to a common theoretical base in order to facilitate valid comparisons, and provides justification for development of a statistical model analogous to that employed in Stack’s American study, with the objective of testing it using mortality data for England and Wales. Testing this model depends for its viability on access to individual-level (as opposed to aggregate) mortality data. The only potential source is identified as the Longitudinal Study, which links Census and vital-event data for 1% of the population of England and Wales from 1971, and is held by the Office for National Statistics (ONS). The Centre for Longitudinal Studies Information and User Support (CeLSIUS) serves as the user gateway to the Longitudinal Study. My application for access to the Longitudinal Study database revealed that Census-based data and vital events data (birth and death registration, widow(er)hoods, cancer registrations, migration, enlistments, entries to long-stay hospitals) are protected in England and Wales by the Census Act and the Population (Statistics) Act. Secondary analysis of this data is restricted to officers of the Office of National Statistics. In the United States, by contrast, access to population data (such as the National Mortality Detail Files, which served as the source of data for Stack’s study) is not only permissible, but the accessibility of such data (which is held by the National Centre for Health Statistics) to the public is mandated.
A discussion of the respective legislation underlying the different policies in England and Wales and the United States relating to privacy of mortality data highlights a trade-off between the privacy of the dead and social scientific research, and gives rise to the question: To what extent do the dead need protection?
Actions (login required)
||Edit record (repository staff only)