Regional disparities in mortality by heart attack: Evidence from France

Presenter: Carine Milcent, CNRS-PSE and Tsing Hua University

Abstract

Objectives:
In many countries, spatial disparities between local markets are large and raise some
major policy concerns. Whereas the focus of the attention is often the labour market,
disparities also occur on other markets such as health. This paper develops a new
approach to explain the spatial disparities in healthcare quality. In health literature,
most of paper focus on the marginal effect of some factors on mortality but do not
assess how some spatial variations in these factors can explain spatial disparities in
mortality. For instance, it is usually found that sex significantly affects mortality. If there
is no variation in the share of females across the territory though, the differences in
the local sex composition will not explain the disparities in mortality across locations.
The same arguments apply when factors vary across space such as local competition
indices. Because decision makers are often concerned with issues on spatial
disparities, we explore the determinants of spatial disparities in hospital healthcare
quality.

Methodology:
We distinguish four types of factors according to the literature on health. First, the
spatial disparities in mortality can be explained by some differences in the local
composition of patients (case-mix). Second, they can be caused by hospital
characteristics. Hospitals can be characterized by their ownership status which is
usually found to affect the hospital performances. Third, spatial disparities in mortality
can come from the local interactions between hospitals. In particular, the local
competition measured with an Herfindahl index is often found to have a significant
negative impact on mortality but what about its affect on spatial disparities in quality.
At last, they can be due to area factors as local employment.

Econometric Model:
We use a very flexible econometric specification building on Ridder and Tunali (1999)
and Gobillon, Magnac and Selod (2007). We first estimate a Cox duration model
stratified by hospital (i.e.each hospital has a specific baseline hazard) using the
stratified partial likelihood estimator.. Their effects are estimated properly accounting
for the hospital unobserved heterogeneity. We then go further and specify the hospital
hazards as the product of some hospital fixed effects and a baseline hazard. The
estimated hospital fixed effects are regressed on some hospital and regional
variables. We finally make a spatial variance analysis. Estimations are conducted on a
unique matched patient-hospital dataset from exhaustive French administrative
records over the 1998-2003 period.

Results:
We show that regional disparities in mortality are quite large. In particular, the raw
difference in the propensity to die within 15 days between the extreme regions
reaches 80%. After accounting for the individual variables, this difference drops to
47%. A variance analysis at the regional level shows that regional differences in
innovative treatments play a major role in explaining the regional disparities in
mortality. A local Herfindahl index computed with the number of patients in each
hospital also plays a significant significant role. This suggests that spatial differences
in the degree of competition between hospitals would partly explain spatial differences
in mortality. After hospital and geographic variables have been controlled for, some
unexplained regional disparities still remain.