Biased estimates in epidemiological substance use research? Evidence from a community study with hair testing

• In Event: Poster Session 03
  In Poster Session: PS 03 Topic - Developmental Psychopathology

Abstract

Illicit drug use (e.g., hallucinogens, stimulants, cannabinoids) and nonprescription use of medical drugs (e.g., opioid) are widespread public health problems facing young people today. Drug use is particularly prevalent during the young adult period and, in many cases, it follows severe childhood adversity. To date, much valuable information about substance use comes from large-scale community-representative, national, and international surveys, which typically rely on self-reports of substance use. Similarly, most developmental studies rely on self-report data. Without additional objective verification, the accuracy of prevalence estimates from these studies is not fully understood, and estimates from models linking childhood adversity with young adult substance use could be biased. Therefore, large-scale representative studies using biological measures are needed to 1) estimate the true prevalence of substance use, 2) assess the convergence of self-reports and objective substance use quantification, and 3) identify individual and social risk factors of under- or overreporting drug use.

The current study leverages data from a large-scale community sample of young adults (N = 1003, mean age = 20.57 years (SD = 0.38), 50.2% female) and combines hair toxicology analysis with drug use self-reports. Hair toxicology analysis is a simple non-invasive method for generating precise quantification of drug use severity across the past months. So far, it has mostly been implemented in research with small, selective samples (e.g., clinical and high-risk samples), which are not informative about the general population.

We compared objective measures of substance use from hair analyses with self-reports for the same time period (i.e., the previous three months). A comprehensive list of illicit drugs was assessed, including cannabis, amphetamines, ketamine, cocaine, 3,4-methylenedioxymethamphetamine (MDMA/Ecstasy), and also non-medical use of codeine and opiate painkillers. We considered putative correlates of false negative and false positive drug use self-reports, including in the realms of socio-demographics, personality, psychopathology, social environmental characteristics, and drug use patterns.

Among the illicit drugs, cannabis was the most frequently detected substance in the hair samples, followed by MDMA/ecstasy, cocaine, ketamine, and amphetamine. Prevalence of heroin and 2-C drugs use was low (n < 5). Differences between hair test results and self-reports were significant across all illicit substances (Figure 1). For all illicit drugs, except amphetamines, self-reports underestimated the prevalence of use by about 30—60% compared to the hair-report standard. With regard to nonprescription medical drug use, codeine use was more prevalent than use of opiate painkillers, and although codeine use was underreported in the self-reports, the hair and self-report data provided equal prevalence rates for opiate painkillers use. Our study also provides several indices from test comparisons (e.g., test specificity and sensitivity). Analyses of risk factors of false self-reports show that poly-drug use increases the risk of self-report inaccuracy, whereas anti-social attitudes and behaviors increase self-report adequacy among those with a positive hair test result.

Our findings provide usable information for correcting estimates from self-report drug use studies and, thus, contribute to a better understanding of substance use prevalence and respective childhood risk factors that are typically identified based on drug use self-reports.

Authors

• Annekatrin Steinhoff, University of Zurich
  Presenting Author

• Laura Bechtiger, University of Zurich
  Non-Presenting Author

• Markus R. Baumgartner, University of Zurich
  Non-Presenting Author

• Josua Zimmermann
  Non-Presenting Author

• Denis Ribeaud, University of Zurich
  Non-Presenting Author

• Manuel Eisner, University of Cambridge
  Non-Presenting Author

• Lilly Shanahan, University of Zurich
  Non-Presenting Author

• Boris B. Quednow, University of Zurich
  Non-Presenting Author