This cross-sectional analysis includes evaluation of definitive urine drug testing (UDT) results tested by liquid chromatography-tandem mass spectrometry (LC-MS/MS) from unique patient specimens across the U.S. and multiple health care specialties. The LC-MS/MS testing method is a laboratory-developed test with performance characteristics determined by Millennium Health, San Diego, California, which is certified by the Clinical Laboratory Improvement Amendments and accredited by the College of American Pathologists for high-complexity testing. Specimens were collected between April 12, 2023, and July 20, 2023 and included specimens were from patients aged 18 years or older. All specimens were collected from health care practices and had ordered and valid test results for xylazine. Over 160,000 specimens from more than 73,000 unique patients were selected for analysis.
The following drug categories were tested for in patient specimens (all analytes tested are in parentheses): xylazine (xylazine, 4-OH-xylazine), cocaine (benzoylecgonine), methamphetamine (methamphetamine), fentanyl (fentanyl, norfentanyl), heroin (6-monoacetylmorphine) marijuana (delta-9-tetrahydrocannabinol carboxylic acid [cTHC]), alcohol (EtG, EtS), gabapentin (gabapentin), Benzodiazepines (alpha-hydroxyalprazolam, 7-aminoclonazepam, lorazepam, nordiazepam, oxazepam, temazepam, 8-aminoclonazolam, etizolam, alpha-hydroxyetizolam, flualprazolam and flubromazolam) and Prescription Opioids (hydrocodone, norhydrocodone, hydromorphone, oxycodone, noroxycodone, oxymorphone, O-desmethyl-tramadol, and N-desmethyl-tramadol). If any parent drug or metabolite within a drug category was ordered and detected, the drug category was considered positive for that specimen. Specimens with reported prescriptions for any of the UDT drug categories above were excluded from analysis. In addition to the above drugs and drug classes, the following fentanyl analogs were also tested for in patient specimens (all analytes tested are in parentheses): carfentanil (carfentanil), acrylfentanyl (acrylfentanyl) and parafluorofentanyl (parafluorofentanyl).
We modeled UDT positivity rates using Binomial generalized estimating equations (GEE). GEE is a robust estimator for longitudinal data when the correlation structure of observations is unknown and is an alternative to standard generalized linear mixed effect models. An `Exchangeable` correlation structure was used for model fitting, and we calculated sandwich robust standard errors to account for the clustered nature of patient data (R geeglm function). Different model specifications were used depending on the analysis requirements (see figure and table legends for included covariates). Adjusted positivity rates (Least Square Mean probabilities) and adjusted odds ratios (aOR) were estimated. 95% CI values for aOR and positivity rates (predicted probability) were estimated.