A spatial structural equation model with an application to area health needs

Abstract

Indices of population 'health need' are often used to distribute health resources or assess equity in service provision. This article describes a spatial structural equation model incorporating multiple indicators of need and multiple population health risks that affect need (analogous to multiple indicators-multiple causes models). More specifically, the multiple indicator component of the model involves health outcomes such as hospital admissions or mortality, whereas the multiple risk component models the impact on the need for area social and demographic indicators, which proxy population-level risk factors for different diseases. The latent need construct is allowed (under a Bayesian approach) to be spatially correlated, though the prior assumed for need allows a mix of spatially structured and unstructured influences. A case study considers variations in need for coronary heart disease (CHD) care over 625 small areas in London, using recent mortality and hospitalization data (the 'indicators') and measures of general ill-health, income and unemployment, which proxy variations in population risk for CHD.

Random-Effects Models for Migration Attractivity and Retentivity: A Bayesian Methodology

“…As well as these ‘multiple indicators’, we can extend the spatial attractivity scheme to have ‘multiple causes’ in a multiple indicators, multiple causes type of factor model, with the latter meaning house prices, unemployment rates, etc. So in the spatial prior for the random attractivities () we may account for measured causes, as Congdon (2009) illustrates in another form of application, namely an analysis of a latent small area health need index. Congdon (2009) adapts a spatial conditional auto‐regressive (CAR) prior modified for known predictors, as set out, for example, by Bell and Broemeling (2000).…”

“…So in the spatial prior for the random attractivities () we may account for measured causes, as Congdon (2009) illustrates in another form of application, namely an analysis of a latent small area health need index. Congdon (2009) adapts a spatial conditional auto‐regressive (CAR) prior modified for known predictors, as set out, for example, by Bell and Broemeling (2000).…”

Spatially explicit structural equation modeling

“…In cases where clear spatial causal relationships cannot be justified at all scales, it may be preferable to optimize the SE-SEM for a single lag distance as described in Description of the method: Fit and evaluate SEM models for each lag distance bin. The SE-SEM methodology outlined in this paper differs substantially from the alternative spatial SEM approaches available in the literature (Wang and Wall 2003, Liu et al 2005, Congdon et al 2007, Oud and Folmer 2008, Congdon 2010. Those methods have not achieved widespread usage in the natural sciences.…”

“…One common approach is to incorporate a distance measure as an observed or latent variable in a standard SEM model (Bailey and Krzanowski 2012). Other methods have largely been developed for health and sociometric data aggregated by administrative districts such as counties or city wards such as the modeling of spatially structured residuals accounting for relationships between adjacent districts (Congdon et al 2007, Oud and Folmer 2008, Congdon 2010. Also proposed are an extension of the common factor model to include neighborhood information (Wang and Wall 2003) and a hierarchical extension for simultaneous modeling of relationships between latent variables while accounting for spatial relationships (Liu et al 2005).…”