Realist identification of group-level latent variables for perinatal social epidemiology theory building.

We have previously reported in this journal on an ecological study of perinatal depressive symptoms in South Western Sydney. In that article, we briefly reported on a factor analysis that was utilized to identify empirical indicators for analysis. In this article, we report on the mixed method approach that was used to identify those latent variables. Social epidemiology has been slow to embrace a latent variable approach to the study of social, political, economic, and cultural structures and mechanisms, partly for philosophical reasons. Critical realist ontology and epistemology have been advocated as an appropriate methodological approach to both theory building and theory testing in the health sciences. We describe here an emergent mixed method approach that uses qualitative methods to identify latent constructs followed by factor analysis using empirical indicators chosen to measure identified qualitative codes. Comparative analysis of the findings is reported together with a limited description of realist approaches to abstract reasoning.

Neighborhood adversity, ethnic diversity, and weak social cohesion and social networks predict high rates of maternal depressive symptoms: a critical realist ecological study in South Western Sydney, Australia.

The aim of the study reported here is to explore ecological covariate and latent variable associations with perinatal depressive symptoms in South Western Sydney for the purpose of informing subsequent theory generation of perinatal context, depression, and the developmental origins of health and disease. Mothers (n = 15,389) delivering in 2002 and 2003 were assessed at two to three weeks after delivery for risk factors for depressive symptoms. The binary outcome variables were Edinburgh Postnatal Depression Scale (EPDS)> 9 and > 12. Aggregated EPDS > 9 was analyzed for 101 suburbs. Suburb-level variables were drawn from the 2001 Australian Census, New South Wales Crime Statistics, and aggregated individual-level risk factors. Analysis included exploratory factor analysis, univariate and multivariate likelihood, and Bayesian linear regression with conditional autoregressive components. The exploratory factor analysis identified six factors: neighborhood adversity, social cohesion, health behaviors, housing quality, social services, and support networks. Variables associated with neighborhood adversity, social cohesion, social networks, and ethnic diversity were consistently associated with aggregated depressive symptoms. The findings support the theoretical proposition that neighborhood adversity causes maternal psychological distress and depression within the context of social buffers including social networks, social cohesion, and social services.

Factors predicting patient satisfaction following major trauma.

INTRODUCTION: Patient satisfaction is an intuitively important outcome measure and has been previously linked to general health status. Previous research on patient satisfaction after injury has concentrated on satisfaction with medical care. This study aims to explore possible predictors of patient satisfaction with outcome following major trauma. METHODS: A cross-sectional survey involving consecutive adult patients involved in major accidental trauma from a major metropolitan trauma centre, over a 5-year period, was performed between 1 and 6 years post-injury. The outcome used was patient satisfaction with progress since the injury. Multiple logistic regression was used to develop a model of significant predictors of patient satisfaction. RESULTS: The survey was mailed to 728 eligible patients, 56 were excluded due to death or inability to complete the survey, 93 refused to participate and 90 were not contactable. One hundred and thirty-four patients did not respond and 355 completed surveys were returned. Patient dissatisfaction was found to be significantly associated with unemployment at the time of follow up (OR, 2.38; 95% CI, 1.38-4.08; p=0.004), having one or more chronic illnesses at the time of injury (OR, 2.57; 95% CI, 1.45-4.55; p=0.001), being involved in a motor vehicle accident (OR, 1.83; 95% CI, 1.02-3.30; p=0.04) and having an unsettled compensation claim (OR, 5.19; 95% CI, 2.80-9.65; p<0.0001). Patient satisfaction was not significantly associated with any measure of injury severity. CONCLUSIONS: Having an unsettled compensation claim after major trauma is the strongest predictor of patient dissatisfaction following major trauma, allowing for other factors.