Words vs. strands: Does the concordance of self-reported and hair-analyzed substance use of young adults change over time?

Background

Substance use is a widespread public health problem with implications for individual and societal well-being (Peacock et al., 2017). Traditionally, the assessment of substance use has heavily relied on self-reports (SR), which might be subject to biases such as recall errors and social desirability. We previously investigated the prevalence of illegal substance use among 20-year-olds with SR and toxicological hair analyses (HA) and demonstrated that SR underestimated young adults’ exposure to specific illegal substances (e.g., cocaine) and non-medical use of prescription drugs (e.g., codeine) up to 60% (Steinhoff et al., 2023). Similarly, approximately 10% of children aged 9-13 years old showed inconsistent hair toxicology results with reported substance use (Wade et al., 2022). 

However, longitudinal investigations using a large set of illicit substances within one cohort of adults are currently lacking. We thus aim to investigate the prevalence of specific substance use and non-medical use of prescription drugs with SR and HA across two timepoints.

Methods

Data was drawn from a large cohort study (z-proso). At the first timepoint (t1), the sample comprised 1,002 participants (Mage, t1 = 20.6 (.38), 50.2% female), whereas the cohort consisted of 886 participants (Mage, t2 = 24.5 (.38), 48.3% female) at the second timepoint (t2). All participants provided a minimum of 3 cm of hair and gave information on their substance use, sociodemographic characteristics, and psychological and behavioral attributes in a survey. HA focused on the detection of cannabinoids, ketamine, opiates/opioids, stimulants, and relevant metabolites. We used χ2-tests to analyze SR and HA and binary logistic regressions to test behavioral and psychological predictors of underreporting. 

Results

Findings from both timepoints suggest an underestimation of most illegal substances in SR between 23% and 74%. Participants consumed more cocaine, ketamine and opioid painkillers, and less codeine at t2, but similarly often cannabis, amphetamines, and ecstasy. Females consumed more opioid painkillers, whereas males used more illicit drugs, such as cannabis and cocaine. SR and HA showed significant stability and reliability in detecting substance use across the timepoints, implicating a robust consistency. There was also a significant association between false negatives at both timepoints indicating a persistent trend of underreporting. For cannabis and amphetamines there was a significant relationship between the false positives for each substance across time. Low delinquency and low ADHD symptoms were predictors of underreported cocaine use, while low internalizing symptoms forecasted underreporting in codeine use.

Conclusions

Our findings emphasize the importance of objective assessment methods, and possible biases in associations with SRs, potentially affecting the validity of study findings.