Abstract C16: Validation of a new medical discrimination scale among a diverse population of breast cancer survivors

Introduction: Experiences of medical discrimination may impact women's decisions regarding breast cancer treatment. However, measures of perceived discrimination generally are not specific to the medical context, and none are specific to breast cancer. The objective of this study was to determine the structure and psychometric properties of a 9-item scale assessing perceived discrimination among a racially/ethnically diverse sample of women receiving breast cancer care. This scale was developed based on existing measures of general discrimination and recent qualitative research with breast cancer survivors of various racial/ethnic backgrounds. Methods: Women ages 20 and older diagnosed with a first invasive primary breast cancer between 2006 and 2009 were sampled from all available case listings in the Greater Bay Area Cancer Registry. 542 of these women completed a telephone interview in English, Spanish, Mandarin, Cantonese, or Tagalog between 2011 and 2013. Women reported how often they perceived 9 types of discriminatory experiences while receiving breast cancer care using a 4-point scale (never, rarely, sometimes, often). Item-level descriptive statistics and reliability analyses were computed. A categorical confirmatory factor analysis (CCFA) was conducted to determine whether the 9 items reflected a single discrimination construct. Logistic regression was used to evaluate differential item functioning (DIF), examining whether the items performed differently by race after controlling for the level of perceived medical discrimination. A summed medical discrimination score was calculated from the items that supported the unidimensional discrimination construct according to the CCFA and did not show evidence of DIF. Correlations between the summed score and single item measures of perceived racial/ethnic discrimination and satisfaction with breast cancer care were used to assess convergent validity. Results: Reports of medical discrimination were fairly low, with no more than a third of any racial/ethnic group of women reporting discrimination on any item. As a result, responses were dichotomized into never vs. ever experiencing discrimination. A CCFA of all 9 items suggested improved model fit could be achieved by adding a correlated error term between two items with similar wording. After removing the item with the smallest factor loading from this item pair, a final CCFA produced adequate model fit (RMSEA=0.06, CFI=0.99, TLI=0.99). Standardized facto