A multivariable analysis delineating hair color, hair dyeing, and hat wearing as predictors of level of cocaine and MDMA detection in human hair samples.

Research suggests that hair color, hair dyeing, and perspiration can bias hair test results regarding drug exposure, but research is needed to examine such associations in a multivariable manner. In this epidemiology study, adults were surveyed entering nightclubs and dance festivals in New York City, and 328 provided hair samples, which were analyzed using ultra-high-performance liquid chromatography-tandem mass spectrometry to determine the level of detection of cocaine and 3,4-methylenedioxymethamphetamine (MDMA). Reporting use was not an inclusion criterion for analysis. We used two-part multivariable models to delineate associations of hair color, past-year hair dyeing, and frequency of past-month hat wearing (which may increase perspiration) in relation to any vs. no detection of cocaine and MDMA as well as level of detection, controlling for hair length, self-reported past-year cocaine/ecstasy/MDMA use, and age, sex, and race/ethnicity. Those reporting having dyed their hair were at increased odds of having any level of cocaine detected (adjusted odds ratio [aOR] = 3.75, 95% CI confidence interval [CI]: 1.85-6.70), and compared to those with brown hair, those with blond(e) hair on average had lower levels of cocaine (ng/mg) detected (beta = -7.97, p = 0.025). Those reporting having dyed their hair were at increased odds of having any level of MDMA detected (aOR = 3.05, 95% CI: 1.44-6.48), and compared to those who reported never wearing a hat, those who reported wearing a hat daily or almost daily on average had lower levels of MDMA (ng/mg) detected (beta = -6.61, p = 0.025). This study demonstrates the importance of using multivariable models to delineate predictors of drug detection.