Calendar Methods Resolve Prenatal Cannabis Timing Far Better Than Dose in Large Pregnancy Cohorts

Question: How do studies using the Timeline Follow-back method or other calendar-based self-report measures characterize the timing, dose, and frequency of prenatal cannabis exposure in large pregnancy cohorts?

Answer

Across the retrieved literature, calendar-based self-report is the field's default instrument for prenatal cannabis exposure, but it is used to build three axes of exposure with markedly unequal resolution. Timing is the axis these methods genuinely deliver: the largest retrieved pregnancy cohort (N=1,782 dyads, 221 exposed) assessed maternal use with the validated Timeline Follow-back and analyzed exposure presence, gestational timing, and frequency as separate analytic dimensions, finding first-trimester exposure associated primarily with white matter microstructural alterations and sustained throughout-pregnancy exposure associated with volumetric differences in cerebellar and subcortical regions [4]. Frequency is routinely collapsed into coarse ordered strata or dichotomies — for example >weekly versus <monthly — rather than modeled continuously, and those strata carry the reported dose-response signal, with high-frequency exposure associated with lower birthweight (Beta=-0.13) and low-frequency exposure not [1]. Quantified dose in mg THC per day is achievable with calendar methods and has been benchmarked against creatinine-normalized urine THC-COOH, but the retrieved studies that do this are in non-pregnant populations [3], and potency-adjusted standard THC units (1 unit = 5 mg THC) derived from an Enhanced Cannabis Timeline Followback likewise appear only outside pregnancy cohorts [8]. Taken together, these papers support a position of asymmetric measurement: the field can say when and how often with some credibility, can rarely say how much, and increasingly treats biochemical assay as the referent against which self-report is interpreted rather than replaced.

Calendar-based TLFB is the exception, not the norm, even among large pregnancy cohorts

The HEALthy Brain and Child Development Study is the only retrieved large pregnancy cohort that reports using the Timeline Follow-back method, and it uses it to generate the most granular exposure contrasts in the retrieved set — presence, gestational timing, and frequency within exposed infants — in 1,782 mother-infant dyads with 221 exposed pregnancies [4]. Other large cohorts rely on far thinner self-report: the Adolescent Brain Cognitive Development Study captures prenatal cannabis exposure through retrospective maternal report collected when children were 9–11 years old, with no timing or dose structure described [7], and other ABCD-based analyses use maternal self-report of prenatal substance exposure in coarse before/after-pregnancy-awareness and daily/weekly categories [10]. Nationally representative survey data reduce exposure further, treating trimester as a covariate that moderates the association between risk perception and use rather than as an exposure-dating instrument [5]. Administrative record-based ascertainment is thinner still: in a whole-population cohort of 970,470 Australian children, maternal substance use was recorded during the pregnancy period for only 1.2% of children, and ascertainment varied by data source [2].

Dose is characterized as a frequency stratum, not as an amount

Where pregnancy cohorts do attempt a dose-like characterization, they do it by binning frequency: exposed pregnancies in the CUDDEL study were stratified into >weekly versus <monthly use, and only the high-frequency stratum showed a birthweight decrement versus unexposed pregnancies [1]. That same study defines exposure status itself as self-reported use or urine THC-COOH positivity at any trimester, which means a single positive assay plus a single report collapses an entire pregnancy into one binary exposure variable [1]. True quantity is measured outside pregnancy: trained interviewers administered a calendar-based TLFB covering daily use over the past 30 days, summarized quantity as mg of THC per day, and tested its association with creatinine-normalized urine THC-COOH, motivated explicitly by the unreliability of self-reported cannabis measures and the paucity of TLFB–biomarker work in under-represented populations [3]. The Enhanced Cannabis Timeline Followback goes further, capturing quantity, frequency, and potency to compute weekly standard THC units administered at 3-month intervals [8]. No retrieved pregnancy cohort reports mg THC per day or potency-adjusted units.

Timing is operationalized as trimester bins or as change relative to pregnancy recognition

The dominant timing construct is a trimester-level contrast, and its reported effects are anatomically specific: first-trimester exposure mapped onto white matter microstructure and sustained exposure onto regional volumes, in the same cohort [4]. An alternative and coarser construct is continuation versus cessation relative to pregnancy awareness, used to show that tobacco exposure both before and after awareness contributed to adiposity risk with no significant benefit from cessation after awareness [10]. The pattern-based approach — abstained, persistent, intermittent, discontinued across all three trimesters — has been implemented prospectively in a population cohort for alcohol and cigarettes, showing that persistence rather than any use carries the association with maternal symptom trajectories [9], but the retrieved literature does not apply this four-category scheme to cannabis. Survey evidence indicates later pregnancy is a period of elevated use risk among women with low perceived risk and depression [5], which makes the timing axis substantively, not just methodologically, important.

Biochemical anchoring is used to correct and complement self-report, not to replace it

The clearest design innovation is requiring both channels: CUDDEL classified exposure by self-report or urine THC-COOH positivity and adjusted for urinary cotinine, concluding that cotinine assessment is especially valuable where blunt use or secondhand exposure is common [1]. Where self-report is absent, assay-based exposure dating becomes the exposure measure itself — maternal urine collected at 8–14 weeks' gestation was assayed for COOH-THC and levels were associated with newborn hippocampal, insular, and caudate connectivity [6]. The calendar method's own metric has been explicitly validated against this referent in a non-pregnant clinical population, where TLFB-derived mg THC per day was tested against creatinine-normalized urine THC-COOH with stratification by gender at birth [3].

Conflicts and gaps

The field disagrees on where to cut the frequency variable: >weekly versus <monthly in one cohort [1], presence/timing/frequency-within-exposed in another [4], and four-level pattern categories in a third that studied alcohol and cigarettes rather than cannabis [9]. These are not interchangeable, so "high-frequency" means different things across papers. The HBCD report calls the Timeline Follow-back "validated" [4], yet no retrieved paper reports a pregnancy-specific validation of TLFB against a biomarker within a pregnancy cohort — the concordance evidence in the retrieved set comes from persons with HIV [3]. Dose is the largest gap: mg THC per day and standard THC units are demonstrated as feasible metrics [3] [8] but are not reported for any pregnancy cohort here, and product potency — central to standard-unit calculation — is not addressed in any retrieved prenatal study. Retrospective recall collected a decade after birth [7] and record-based ascertainment capturing 1.2% of pregnancies [2] represent the low-resolution floor against which prospective calendar methods should be judged, but no retrieved paper directly compares the two on the same pregnancies.

What would change the answer

A within-pregnancy validation study would settle the concordance question: administer TLFB prospectively each trimester alongside repeated urine THC-COOH and cotinine in the same cohort, and report sensitivity and specificity of TLFB-defined timing and frequency against assay-defined exposure. Extending HBCD's TLFB data to mg THC per day and potency-adjusted standard THC units [4], using the Enhanced Cannabis Timeline Followback quantity-frequency-potency instrument already applied in the CannTeen study [8], would move prenatal dose onto the same metric as the general cannabis literature. Applying the abstained/persistent/intermittent/discontinued trajectory scheme validated on trimester-by-trimester prospective reports [9] to TLFB day-level cannabis data would replace arbitrary frequency dichotomies with empirically derived patterns. Finally, linking TLFB-derived gestational-week timing to meconium or newborn hair cannabinoid concentrations would test whether week-level dating, rather than trimester bins, is recoverable — the only way to know whether the timing-specific brain findings [4] reflect the exposure window or the imprecision of the instrument.

References

[1] Constantino-Pettit, A., Trammel, C., Agrawal, A., Smyser, C., Carter, E., Bogdan, R., Rogers, C.. "Associations Between Prenatal Cannabis Exposure and Birth Outcomes: Results from a Prospective Cohort Study". medRxiv. 2026-03-03T00:00:00.000Z. URL: https://doi.org/10.64898/2026.03.01.26347369

[2] Powell, M., Pilkington, R., Ahmed, T., Hanly, M., Newton, B., Lynch, J., Dobbins, T., Stewart, J., Cretikos, M., Havard, A., Falster, K.. "Prevalence of maternal substance use during pregnancy and first two years of life: A whole-population cohort of 970,470 Australian children born 2008-2017". medRxiv. 2024-12-10T00:00:00.000Z. URL: https://doi.org/10.1101/2024.12.09.24318518

[3] Porchia, D. D., Li, Y., Truver, M., Goldberger, B., Porges, E. C., Wang, Y., Li, Z., Wu, S., Cook, R. L.. "Gender differences in correlations of self-reported marijuana use with a creatinine-normalized urine biomarker among persons with HIV". medRxiv. 2025-03-15T00:00:00.000Z. URL: https://doi.org/10.1101/2025.03.14.25323995

[4] Shah, L., Planalp, E., McDonald, R., Regner, C., Atluru, S., Merhar, S., Alexander, A., Ossorio, P., Poehlmann, J., Dean, D.. "Associations of Prenatal Cannabis Exposure and Neonatal Brain Development in the HBCD Cohort". medRxiv. 2026-10-19T00:00:00.000Z. URL: https://doi.org/10.64898/2026.03.02.26347436

[5] Laila Volpe, Luise Glamann, Katherine Foster, Leah Mayo, Tory Spindle, Katrina Mark, Katrijn Houben, Daniel Roche. "Under What Circumstances Do Pregnant Women Use Cannabis and Alcohol? The Contributions of Risk Perception, Mental Health, and Individual Context". psyarxiv. 2025-05-14T20:35:32.257Z. URL: https://osf.io/mh85j_v1

[6] Chaela Nutor, Jocelyn Stanfield, Patricia Brennan, Dana Barr, Anne Dunlop, Cassandra Hendrix. "Maternal Early Pregnancy Tetrahydrocannabinol (THC) Metabolites Correlate with Newborn Resting-State Functional Connectivity". psyarxiv. 2026-05-14T15:24:23.939Z. URL: https://osf.io/3e9md_v1

[7] Ami Ikeda, Chaela Nutor, Chelsie Benca-Bachman, Rameez Syed, Valerie Knopik, Rohan Palmer, Patricia Brennan. "Prenatal Cannabis Exposure, Genetic Predispositions, and Autism Spectrum Disorder Traits in the Adolescent Brain Cognitive Development Study". psyarxiv. 2024-12-20T14:49:31.701Z. URL: https://osf.io/wpng5_v1

[8] Lees Thorne, R., Lawn, W., Petrilli, K., Trinci, K., Borissova, A., Ofori, S., Mokrysz, C., Curran, H. V., Hines, L., Freeman, T. P.. "Estimating thresholds for risk of cannabis use disorder using standard THC units". medRxiv. 2025-05-21T00:00:00.000Z. URL: https://doi.org/10.1101/2025.05.21.25328059

[9] Palmer, K. N., Law, K. Y. T., Massey, S. H., Lussier, A. A.. "Patterns of substance use during pregnancy are differentially associated with maternal mental health across the perinatal period". medRxiv. 2026-10-05T00:00:00.000Z. URL: https://doi.org/10.64898/2026.10.01.26364530

[10] Li, R., Wigley, I. M., Suuronen, I., Jolly, A., Tuulari, J. J.. "Effects of Prenatal Substance Exposure on Longitudinal Tri-Ponderal Mass Index Trajectories from Pre- to Early Adolescence in the ABCD study". medRxiv. 2024-11-26T00:00:00.000Z. URL: https://doi.org/10.1101/2024.11.23.24317835