Validity of Self-Reported Cannabis Use Based on Urine Toxicology Measures in a Population-Based Pregnancy Cohort in Michigan.
Level 3 - non-randomized controlled study
Diagnostic validation study against an objective reference standard within a prospective cohort
PubMed 42116667 · doi:10.1080/10826084.2026.2670606
What was done
Researchers evaluated data from 606 pregnant participants recruited at their first prenatal visit across 22 clinics (2017-2023) in the Michigan Archive for Research on Child Health (MARCH) cohort. The study assessed the diagnostic validity of self-reported cannabis use against urine toxicology as the criterion standard among participants who provided self-report data within 30 days of urine collection.
What was found
Urinalysis identified a 19.1% prevalence of prenatal cannabis use, compared to 14.2% identified by self-report. Sensitivity, specificity, positive predictive value, and negative predictive value of self-report were 57.8%, 96.1%, 77.9%, and 90.6%, respectively. Differences between participants who disclosed and those who did not were significant only by race and ethnicity: self-report sensitivity was 45.5% among Non-Hispanic Black participants versus 73.0% among Non-Hispanic White participants.
Why it matters
Self-report misses more than 40% of biologically verified cannabis exposures during early pregnancy and introduces differential misclassification across racial groups. Studies relying solely on self-reported prenatal cannabis exposure risk substantial measurement bias.
Limits
The sample is derived from a single state cohort at first prenatal visits. Allowing up to 30 days between self-report and urine collection may introduce exposure-window discordance. Urinalysis has a limited biological detection window, and the abstract does not report specific cannabis metabolites, frequencies, or modes of use.
Cited by
- partial Approximately 15% of pregnant women in the United States report using cannabis in some form during pregnancy.