Neighborhood Associated Exposures and Socioeconomic Factors are Associated with Prostate Biopsy Outcomes

View Poster

INTRODUCTION

Although poorly understood, neighborhood associated socioeconomic factors like poverty, education, segregation, employment, crime, insurance status, and residential stability, are important predictors of prostate cancer (PCa) mortality. Chicago neighborhoods with the highest PCa mortality also have the highest rates of crime and poverty, and lowest levels of educational attainment. We analyze whether census tract-level socioeconomic factors are independently associated with prostate biopsy outcomes.

METHODS

From 2009-2014, we prospectively recruited 1485 men undergoing routine prostate specific antigen (PSA) screening (n = 566) or their first prostate biopsy (n = 919) for elevated PSA or abnormal digital rectal exam (DRE) at five Chicago hospitals. An exploratory factor analytic approach (EFA) tested 23 socioeconomic and environmental variables from 2010 US Census, Chicago Health Atlas, Chicago Police Department and Environmental Protection Agency. EFA resulted in two highly reliable indices capturing > 70% of variance among neighborhood variables; Neighborhood Associated Toxic Stressors (NATS) and Neighborhood Associated Advantage (NAA) with Cronbach's α of 0.94 and 0.84, respectively, were used as measures of disadvantage and affluence. Univariate and multivariate analyses examined the associations between clinical risk factors, NATS and NAA with PCa diagnosis. We also assess associations between NATS, NAA and Gleason grade ≥3+4 with ordinal logistic regression.

RESULTS

Included in the analysis were 647 (43.6%) Black men and 837 (56.4%) White/Other men. Frequency of cancer on biopsy was 66.9% (615). In a model adjusted for family history and age, men living in the most disadvantaged neighborhoods (highest NATS quartile; OR 1.54; 95% CI, 1.11-2.12; p=0.009) and Black race (OR 1.52; 1.14-1.51; p=0.004) had highest odds of PCa diagnosis. Men in the most affluent neighborhoods (highest NAA quartile; OR 0.70; 0.52-0.95; p=0.022) had decreased odds of PCa diagnosis. In a model additionally adjusted for DRE and PSA, NATS (OR 1.64; 1.22-2.18; p=0.001) and NAA (OR 0.71; 0.53-0.95; p=0.02) remain independently associated with PCa diagnosis. In fully adjusted ordinal regression models for Gleason grade 6-10, the strength and direction of association persist for NATS (OR 1.68; p=0.004) and NAA (OR 0.50, p=0.001).

CONCLUSION

Census tract level neighborhood exposures are independently associated with biopsy outcomes. Utilization of neighborhood measures can inform public policy and should be validated in biopsy cohorts.

Funding: 1R01MD007105-01 (Kittles), IK2CX000926-01 (Murphy), W81XWH-10-1-0532 pd22E (Murphy)