Abstract 4137: Latent class model characterization of neighborhood SES

Abstract

Abstract Background: Breast cancer is the most common cancer in women and a significant source of mortality. Neighborhood-level socioeconomic status (NSES) has been shown to play a key role in health; poorer health outcomes are observed in deprived neighborhoods even after controlling for individual level SES. Commonly used NSES indexes are difficult to interpret. The relative impact of different factors within the index cannot be evaluated. Latent class models, which use latent variables to create distinct classes, allow for characterization of NSES and estimation of the effects of specific neighborhood characteristics on cancer outcomes. Methods: We used data from women diagnosed with breast cancer from a teaching hospital in Philadelphia, PA. Census information at the Zip Code Tabulation Area (ZCTA) level was used to obtain NSES variables. Latent class analysis and comparisons of model fit statistics were used to determine the optimal number of classes. Results: Complete ZCTA-level data was available for 1,664 breast cancer patients. In this population, NSES was best represented by two separate latent variables, each with 2-classes (LC2). When the class variables were compared to a continuous NSES index (NSI), the correlation was 0.85. However, LC2 demonstrated stronger association with race, stage, disease subtype, tumor size, and histologic grade. Conclusions: We classified neighborhoods based on correlated socioeconomic census-level variables. Latent variables identify specific characteristics associated with living in a neighborhood with low versus high advantage and disadvantage and thus may improve our understanding of critical breast cancer prognostic factors and the targeting of cancer control efforts. Further research will provide overall structural latent models that incorporate information on tumor biology, prognostic variables and survival outcomes to aid in the identification of vulnerable populations. Table 1. Neighborhood latent class model (LC2) and SES Index (NSI) Table 1.Neighborhood latent class model (LC2) and SES Index (NSI)LC2 *ModelNSI Model *Prognostic Factor Stage24.5511.61 Race366.44331.01 Subtype18.4315.48 Size22.532.91* Chi-square statistic for association of Prognostic Factor with LC2 class or with NSI quartile. Higher number indicates higher strength of association. Citation Format: Aimee Palumbo, Yvonne Michael, Terry Hyslop. Latent class model characterization of neighborhood SES. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 4137. doi:10.1158/1538-7445.AM2014-4137