Risk Factors for Tuberculosis Conversion in a State Prison

Abstract

A case-control study determined the risk factors for latent tuberculosis (TB) conversion among Oregon Department of Correction (ODOC) inmates from July 2000 - July 2001. The first objective was to identity the converters. These were inmates who tested negative for the Purified Protein Derivative (PPD) skin test on entry and subsequently tested positive on annual testing. The second objective was determining the risk factors for conversion by comparing the converters with randomly selected controls. The Correctional Information System (CIS) and Mental Health databases were accessed to obtain health and demographic information. With ninety-nine percent of PPD positive inmates on anti-tuberculosis medications, nearly all male inmates who tested positive from July 00-01 (n = 307) were identified through the ODOC pharmacy records. A medical chart review (276 of 307 or 90%) separated the converters (n = 72) from the reactors who tested positive on entry (n = 123) and the prior positives on medications (n = 81). The conversion rate was 5.0 per 1,000 person-years. Differences between the cases (converters) and controls were analyzed using multivariate logistic regression. The converters were 6 times more likely to be Latino (p < .005) vs. Caucasian, over 19 times less likely to live in medium vs. minimum (p < .001) or maximum vs. minimum (p < .001) security prisons, and over 5 times less likely to live in a medium vs. low (.012 < p < .031) or high vs. low (.002 < p < .007) density prison. They had 1.4-1.5 times fewer PPD skin tests (.002 < p < .009) and lived in 1.5-1.7 times fewer prisons (.005 < p < .017). Age, education, county of incarceration, number of incarcerations, and number of visitors were not found to be significant variables. The results revealed a low conversion rate compared to other U.S. prisons. Prison health officials should consider performing two-step skin testing in order to distinguish the booster phenomenon from intramural conversion.