Fentanyl, Heroin, Methamphetamine, and Cocaine Analyte Concentrations in Urine Drug Testing Specimens

Abstract

Importance: The US is experiencing a protracted drug overdose crisis primarily associated with exposure to illicitly manufactured fentanyl (IMF), methamphetamine, and cocaine. Overdose risk and treatment responses may be directly affected by absolute drug exposure concentrations and drug use prevalence.

Objective: To quantify changes in absolute drug exposure concentrations from 2013 to 2023.

Design, setting, and participants: This cross-sectional study analyzed urine drug testing (UDT) results from urine specimens collected between January 1, 2013, and August 22, 2023, in 49 states and the District of Columbia. Urine specimens were obtained from patients aged 18 years or older who presented to substance use disorder treatment clinics. The UDT was ordered by clinicians based on medical necessity.

Exposures: Urine specimens were analyzed for the following drugs or metabolites (analytes tested in parentheses): fentanyl (fentanyl), heroin (6-monoacetylmorphine), cocaine (benzoylecgonine), and methamphetamine (methamphetamine) using liquid chromatography with tandem mass spectrometry.

Main outcomes and measures: Relative concentrations of fentanyl, heroin, cocaine, and methamphetamine. Creatinine-normalized drug concentration values were log-transformed prior to visualization and statistical analyses. The Mann-Kendall trend test was performed to examine trends over time. To estimate the geospatial and temporal patterns of drug concentration, a second series of models (1 for each drug) with an interaction effect for clinic location and collection year were fit.

Results: A total of 921 931 unique UDT samples were collected from patients (549 042 males [59.6%]; median [IQR] age, 34 [27-44] years). The adjusted fentanyl concentration in urine specimens was 38.23 (95% CI, 35.93-40.67) ng/mg creatinine in 2023 and 4.61 (95% CI, 3.59-5.91) ng/mg creatinine in 2013. The adjusted methamphetamine concentration was 3461.59 (95% CI, 3271.88-3662.30) ng/mg creatinine in 2023 and 665.27 (95% CI, 608.51-727.32) ng/mg creatinine in 2013. The adjusted cocaine concentration was 1122.23 (95% CI, 1032.41-1219.87) ng/mg creatinine in 2023 and 559.71 (95% CI, 524.69-597.06) ng/mg creatinine in 2013. The adjusted heroin concentration was 58.36 (95% CI, 48.26-70.58) ng/mg creatinine in 2023 and 146.59 (95% CI, 136.06-157.92) ng/mg creatinine in 2013. Drug concentrations varied across US Census divisions.

Conclusions and relevance: This cross-sectional study found that absolute concentrations of fentanyl, methamphetamine, and cocaine in urine specimens increased from 2013 to 2023, with a decrease in heroin concentration during that period. The findings suggest that exposure to these substances, as well as the illicit drug supply, has fundamentally changed in many parts of the US, highlighting the need to reinforce surveillance initiatives and accelerate efforts to treat individuals with IMF and/or stimulant exposure.

Figure 1.. Distribution of Concentrations of Fentanyl, Heroin, Cocaine, and Methamphetamine in Urine Specimens From 2013 to 2023

Heroin (6-monoacetylmorphine) and cocaine (benzoylecgonine) are represented by the indicated metabolites. Fentanyl and methamphetamine were both measured as parent drugs. Concentration values are represented as log₁₀-transformed values.

Figure 2.. Coefficients for Additive Linear Regression Models of Fentanyl and Heroin Concentrations

Linear regression models were performed on log-transformed concentration outcomes. Model coefficients and 95% CIs seen in tabular form were exponentiated to convert them to the original linear space of the concentration values and thus represent the fold-change between the given factor level and the reference level. Model coefficients shown in the plots are in the original log space.

Figure 3.. Coefficients for Additive Linear Regression Models of Cocaine and Methamphetamine Concentrations

Models were performed on log-transformed concentration outcomes. Model coefficients and 95% CIs were exponentiated to convert them to the original linear space of the concentration values and thus represent the fold-change between the given factor level and reference level.

Figure 4.. Spatiotemporal Patterns of Drug Concentration Change

Adjusted mean concentration values were based on regression models with an interaction term for collection year and US Census division. Error bars represent 95% CIs.