Publication type
Thesis/Degree/Other Honours
Author
Publication date
June 1, 2004
Abstract:
Objective of this PhD thesis is to investigate the application of econometric techniques developed elsewhere in the social sciences to three issues of health care policy. First, we focus on the design of policies for the prevention of mental illness. Usually, benefits from prevention programs lie far ahead in the future, which makes it important to target subgroups likely to suffer from longstanding health problems. Our objective is to supplement information on subgroup disparities in mental illness at one point in time with information on disparities in intertemporal fluctuations, or health dynamics, of mental illness in 11 waves of the British Household Panel Survey. Using two measures of health dynamics, we find that its extent varies across socio-economic categories with - mostly - greatest persistence observed in more disadvantaged groups.
Second, we investigate to what extent health authorities can influence performance of hospitals; in particular, we analyse over which performance objectives they have greatest control and at what administrative level influence is greatest. Multilevel models are used to attribute variation in 14 performance indicator to regional and district health authorities, and to random fluctuations at small area level. We find evidence of great variation in the extent to which health authorities can influence performance in different areas, even after adjusting for socio-demographic characteristics. District health authorities on a lower administrative level have greater influence than regional health authorities.
Third, we aim to provide a statistically more accurate picture of organisational performance in the presence of multiple objectives by measuring performance in 13 areas with multivariate-multilevel models. Such models provide a separate performance measure for each objective, but allow for correlations across them. They do not require weighting of objectives as some of the standard techniques. We find evidence of correlations across 13 objectives, suggesting that some are complementary, others subject to trade-off, and some redundant. The estimates generated when assessing performance with multivariate-multilevel models as compared to conventional techniques differ, with the magnitudes varying by objective.
Subjects
Notes
not held in Res Lib - bibliographic reference only
#508444