Skip to content

2012 Conference Presentation

Economics EnglandUnited Kingdom

7 September 2012

Estimates of life-time long-term care costs and their implications for insurance benefits/payouts and premiums

Adam Steventon, The Nuffield Trust, United Kingdom
Robert Adams, The Nuffield Trust, United Kingdom


Objective: Many economic evaluations rely on subjects’ reports of their use of health and social care services, due to the breadth and depth of information required for a comprehensive evaluation. However, administrative data sets containing information about social care packages are now beginning to be used in research. Recent advances in data linkage mean that these data sets are being linked at the person level to secondary, primary and community health care data, so that longitudinal profiles of social and health care use are available spanning several years. Evaluators risk different biases in their data depending on the choice of data source but, as opportunities to compare data sets have been rare, the directions of the biases are not well understood. In particular, there are doubts about the accuracy of social care administrative data. We aimed to compare self-report and administrative data on the use of health and social care and further, to explore the plausibility of hypotheses that exist to explain discrepancies (such as the “telescoping” of more distant events into the recall period).

Data and methods: As part of the evaluation of two randomised trials, data were collected for a population with social care needs and a population with long-term conditions in three areas of England (Cornwall, Kent and Newham). Selfreport data were collected from a total of 2,762 patients regarding the use of secondary health care, general practice, advanced community nurses and social care services such as domiciliary care, meals services, day care and equipment. Comparable data were also collected from administrative data systems of 244 general practices, 4 primary care trusts and 3 local authority social services departments and linked at the person level using encrypted identifiers. We assessed chance-corrected levels of agreement between the two data sets and used multivariate regression to identify characteristics associated with relative under- and over-reporting. We used the administrative data to calculate variables over several time periods in order to assess the role of telescoping in explaining discrepancies.

Results: The presentation will describe the results of the analysis, addressing the characteristics associated with overand under-reporting of use of a number of different aspects of health and social services. As this important evaluation is likely to affect policy, the results are presently under embargo by the sponsor but will be available by the time of the presentation.

Policy implications: These are under embargo but will be available by the time of the presentation.

Skip to toolbar