A comprehensive methodology for assessing true differences in coping for African American patients with cancer (#276)
Health disparities exist across many health care and disease outcomes in the USA, many of which impact the African American population. Health disparity research and medical practice has primarily focused on disease morbidity and mortality; however, there is limited analysis of disparities in more subjective, yet important, constructs such as coping, quality of life, and psychosocial adjustment. If true disparities or differences are evident in morbidity and mortality, then adequate measures of psychosocial constructs are essential in order to establish comprehensive models of health and disparities. A step-wise methodology is proposed that is mixed-method, comprehensive, and applicable to a variety of measures within the psychosocial domain. This model includes qualitative analysis of items, measurement invariance analysis, and item response theory analysis. The first step is a thorough qualitative analysis of items – noting those items that African American and Caucasians identify as important, less relevant, confusing, or difficult to understand, and so on. This comparative analysis is followed by Measurement Invariance (MI) between groups in which successive levels from weak to strong invariance are tested in the following pattern: 1) structural invariance, 2) factorial invariance (invariance of the factor loadings, 3) invariance of slopes, and finally 4) invariance of residuals. Finally, item response theory analysis is performed with a focus on Differential Item Functioning. An example of this comprehensive disparity analysis will be presented using the Cancer Behavior Inventory (CBI), a 33-item measure of self-efficacy for coping with cancer, as a test measure. The CBI was tested across groups of African American (AA; N = 124) and White (W; N = 188) low-income cancer patients. Multigroup Confirmatory Factor Analysis (MGCFA) was performed as a function of patient ethnicity. The results indicated that the CBI exhibited structural but not factorial invariance. Subsequently, an Item Response Theory (IRT) approach to Differential Item Functioning (DIF) was conducted on the CBI across AA (N = 245) and W (N = 407) groups. DIF was flagged for three items. Though these items did not impact CBI scores substantially at the scale level, a discussion of how these items were tested in terms of their influence on scale scores will be presented. This project provides a modern, sophisticated methodological approach to testing disparities in health related constructs and allows for the interpretation of differences across racial groups.