P1-16 A latent class analysis of socioeconomic status and obesity in young adults from Cebu, Philippines

Abstract

Introduction The detection of adverse events following immunisation (AEFI) fundamentally depends on how these events are classified. Standard methods impose a choice between either grouping similar events together to gain power or splitting them into more specific definitions. We demonstrate a method of medically guided Bayesian information sharing that avoids grouping or splitting the data, and we further combine this with the standard epidemiological tools of stratification and multivariate regression. Methods All spontaneous reports of gastrointestinal AEFI in children under 18 years old in the WHO (Uppsala Monitoring Centre) VigibaseÓ were used to calculate reporting ORs for each AEFI and vaccine combination. After testing for effect modification these were then reestimated using multivariable logistic regression adjusting for age, gender, year and country of report. A medically guided hierarchy of AEFI terms was then derived to allow information sharing in a Bayesian model. Introduction Obesity is an important, global, public health problem. To promote prevention, more research is needed to understand exactly how shared environments impact obesogenic behaviours. Identifying spatial clusters of obesity is the first step towards a better understanding of its environmental drivers, and can immediately inform public health practice. To date, research has overlooked lower-income contexts, where obesity is emergent and environments are changing at an unprecedented pace. Methods Using data from a cohort of young adult Filipinos (21.5 y; n¼1808), we used the Kulldorff spatial scan statistic to detect areas in Metropolitan Cebu with a high sample prevalence of obesity. Cluster locations were then compared to the urbanicity of constituent neighbourhoods. We also tested whether clusters were explained by the spatial distribution of household-assets scores in the study participants. Results Significantly unusual clusters (rejection of H0: complete spatial randomness, at p<0.05) of overweight and obesity were detected for males and females. Clusters were primarily located in urban areas, but typically extended into peri-urban and even rural neighbourhoods. The exact location of clusters varied as a function of both sex and measure of obesity used. Clusters in males, but not females, were explained by the spatial distribution of socioeconomic status. Conclusions Where a young adult lives is a strong predictor of obesity risk in Cebu. Environmental drivers of obesity among young adults in Cebu may vary by gender. Using simple urban-rural classifications to contextualise obesity in lower income countries may be overly simple, and misdirect public health efforts. Introduction Socioeconomic status (SES) is a critical driver of human health, but in research practice it is rarely well-defined and inconsistently measured. Latent class analysis (LCA) is a potentially useful method of characterising SES, particularly when multiple SES indicators are available. We employed LCA to better understand how SES is related to obesity in ...

Using latent class analysis to develop a model of the relationship between socioeconomic position and ethnicity: cross-sectional analyses from a multi-ethnic birth cohort study

“…Other studies have also used LCA as a method to combine several dimensions of SEP into one indicator [ 10 – 14 ]. In this study the method provided a detailed description of the SEP profile of a multi-ethnic population of pregnant women with a level of detail that would not have been picked up using more traditional indicators of SEP, such as education or occupation.…”

“…LCA is a data driven approach and is increasingly being applied to health related data. LCA has been used to identify subgroups of study populations based on SEP in different studies in the UK [ 10 , 11 ] the USA [ 12 , 13 ] and the Philippines [ 14 ]. These studies found that between 3 and 7 classes best described the SEP profiles of study participants.…”

The need for, and value of, a spatial scan statistical tool for tackling social health inequalities

“…The most statistically significant area of excess was found in the west of Boston counting 15% more cases than would have been expected plus two other high areas in the east of Boston ( 16 ). In 2011, Dahly and Gilthorpe detected clusters of overweight and obesity located in urban areas ( 17 ), but extended into peri-urban and rural neighborhoods. More recently, in France, Kihal-Talantikite et al identified the locations of clusters of high risk for several health outcomes including infant mortality and end-stage renal disease (ESRD) at different regional levels ( 18–20 ).…”

“…Spatial approaches appear to be adequate to examine the spatial implications of neighborhood characteristics which may affect health inequalities ( 15 ). For instance, Dahly and Gilthorpe in 2011 concluded that clusters of high risk of overweight and obesity in males, but not females, were explained by the spatial distribution of socioeconomic status ( 17 ). More recently, studies revealed that spatial distribution of infant mortality risk in the Lyon metropolitan area may be explained by deprivation level, noise exposure level or green space level ( 19,20 ).…”