A large Medicaid-linked cohort measured well-child visits, ED use, and early developmental delay coding
This post is built for careful interpretation. Exposure was defined by meconium drug screening, not by self-report or ICD coding. That single design choice changes what the paper can detect, what it may miss, and how far the findings travel.
Why meconium testing changes how to interpret prenatal exposure research
What “late pregnancy exposure” does, and does not, mean in real life
Why claims-based developmental delay is a noisy proxy for development
What the utilization results suggest about healthcare engagement
How to talk about this prenatal cannabis exposure study without overclaiming
Why policy and systems can shape what gets documented, diagnosed, and coded
Prenatal Cannabis Exposure Study: What This Paper Found, What It Did Not, and Why Policy Context Matters
If you have reached this page because a headline felt louder than the data, you are in the right place.
Why this prenatal cannabis exposure study matters right now
Pregnancy and cannabis sits in a difficult place in public discourse. Many people are trying to do the right thing, while also trying to manage nausea, sleep disruption, anxiety, pain, or appetite changes. Meanwhile, the information environment is crowded with confident takes that do not scale down to a single patient’s reality.
This prenatal cannabis exposure study is useful because it is not built on self-report alone. The exposure definition is a meconium drug screen obtained during the birth hospitalization. That methodological choice changes what the study can detect, what it might miss, and how much the findings can be generalized.
Clinical note: Nothing in this post is medical advice. If you are pregnant, trying to conceive, or breastfeeding, the safest approach is to discuss any cannabis exposure with your obstetric clinician so your care plan can match your health history and risk profile.
TL;DR, in plain language
In this prenatal cannabis exposure study, infants with cannabis detected on meconium testing had similar well-child visit attendance and similar emergency department use through age 2. Developmental delay coding looked modestly lower at age 2, but that signal disappeared by age 3.
The most common misread is to jump from a short-lived statistical signal to a biological story. This paper does not support a developmental benefit claim. The more defensible interpretation is narrower: in this specific Medicaid-insured, risk-based screening cohort, late-pregnancy cannabis detection was not associated with higher early utilization or higher ICD-coded developmental delay by age 3.
Study design and setting: what was actually measured
The choice to compare cannabis-positive infants to infants who were tested and negative is a strength. It reduces, but does not eliminate, a common problem in pregnancy exposure research: screening-selection bias. In other words, the comparator group had a similar “reason” to be screened, even though the study does not fully describe the institutional triggers for sending meconium tests.
Still, it is important to name what the exposure variable is in this prenatal cannabis exposure study. It is a binary, late-pregnancy detection marker without dose, route, frequency, product type, or timing across trimesters. That is not a small caveat. It is the spine of interpretation.
Outcomes: utilization and developmental delay, defined by claims
Primary outcome: well-child care attendance
Well-child care (WCC) attendance was measured as a count of recommended visits in the first two years, operationalized using CPT and ICD coding. The mean number of visits was about six. This outcome tells you something about healthcare engagement, but it also tells you something about insurance continuity, transportation, caregiver bandwidth, scheduling friction, and policy-mediated follow-up.
Secondary outcome: emergency department visits
Emergency department utilization was measured through claims codes and revenue codes. Again, this is a health measure and an access measure at the same time. Many pediatric ED visits reflect caregiver concern in uncertain moments, after-hours limitations, and a lack of alternative urgent care options.
Developmental delay: an important nuance before the numbers
Developmental delay was defined using ICD-9 and ICD-10 codes listed by the authors. That is a practical choice in claims research, but it is not the same thing as standardized developmental testing. Claims-based developmental delay can be shaped by clinician threshold, referral patterns, caregiver persistence, and the frequency of well-child encounters.
If you read only one sentence in this section: this prenatal cannabis exposure study measures coded developmental diagnoses, not development itself.
Main findings: the table-anchored results, without drama
Well-child visits through age 2
The incidence rate ratio for WCC visits comparing exposed to unexposed infants was 0.982 (95% CI 0.957 to 1.008), with P = 0.1705. The study did not find a statistically significant difference in well-child visit attendance over the first two years.
Emergency department visits through age 2
The incidence rate ratio for ED visits was 0.934 (95% CI 0.863 to 1.012), with P = 0.0962. Again, this prenatal cannabis exposure study did not show a statistically significant increase in ED utilization in the first two years.
Developmental delay at age 2
At two years, the adjusted odds ratio for developmental delay coding was 0.826 (95% CI 0.686 to 0.995), with P = 0.0437. The absolute rates were 15.02% in the cannabis-exposed group and 18.39% in the unexposed group.
Developmental delay extended to age 3
At three years, the adjusted odds ratio was 1.007 (95% CI 0.852 to 1.191), with P = 0.9304. The apparent two-year difference did not persist. That time instability should tighten, not loosen, the inference.
The cleanest summary of this prenatal cannabis exposure study is not “good news” or “bad news.” It is “no measurable increase in early utilization burden, and no higher developmental delay coding by age 3, in this specific cohort and context.”
Bias audit: where interpretation can quietly go off the rails
Selection bias: who gets tested is not random
Meconium testing was risk-based, not universal. That means the analytic population is not “all infants with prenatal cannabis exposure.” It is infants selected for testing under institutional criteria. The negative comparator group strengthens internal comparison, but external generalizability narrows.
Exposure misclassification: what meconium misses
Meconium primarily reflects later pregnancy exposure. Early pregnancy exposure may be missed, and the study does not quantify dose, frequency, route, potency, or product types. In a prenatal cannabis exposure study like this, non-differential misclassification tends to move estimates toward no difference.
Detection bias: coding depends on contact
Developmental delay defined by claims codes is vulnerable to surveillance effects. More visits can mean more opportunities for a code to appear. Differences in clinician threshold and referral pathways can produce patterns that look biological but are actually procedural.
Confounding: what is not in the dataset
The models included many relevant medical covariates, including gestational age, SGA, LBW, maternal mental health, lead levels, and more. Still, key social confounders are not reliably captured in claims data, including caregiver supports, housing stability, education, and structured interventions following a positive screen.
The policy context: mandated reporting can change the downstream story
The institutional requirement to notify Child Protective Services for positive cannabis screens is not a neutral background detail. It can change caregiver behavior, clinician behavior, follow-up intensity, and referral patterns. It can also change how quickly developmental concerns are noticed and how frequently they are coded.
One way to hold this in your mind: this prenatal cannabis exposure study can be read as a measurement of system response as much as a measurement of exposure biology. That does not invalidate the findings. It clarifies what the findings represent.
A practical interpretation guardrail: if an exposure triggers structured surveillance, then downstream “no difference” outcomes might reflect buffering by the system, not biological neutrality.
How this prenatal cannabis exposure study fits with the broader literature
This paper is not a global verdict on cannabis in pregnancy. It speaks to a particular exposure definition, a particular insurance population, and a particular follow-up window. Still, it aligns with other large observational work showing no increased risk of early developmental delays when outcomes are defined through early childhood and measured through clinical records or claims, while also differing from studies assessing later childhood behavioral outcomes with different exposure definitions.
External validation that points in a similar direction
Avalos and colleagues (Kaiser Permanente Northern California) reported that maternal prenatal cannabis use in early pregnancy was not associated with increased risk of early child developmental delays in their cohort, while emphasizing the need for better pattern-of-use measurement across pregnancy.
A contrasting signal in older children using different methods
Paul and colleagues (ABCD Study) reported associations between prenatal cannabis exposure and later childhood outcomes, including psychopathology measures, using retrospective maternal report and a different age window. Method differences matter. Age differences matter. And confounding structure matters.
Utilization and severe outcomes are not the same question
Bandoli and colleagues examined prenatal cannabis use disorder and infant hospitalization and death in the first year of life, highlighting how a use disorder phenotype differs from a biomarker-positive late pregnancy exposure marker. These are distinct exposures and should not be blended in public conversations.
When you hear someone cite a single prenatal cannabis exposure study as definitive, a helpful question is: exposure definition, outcome definition, and age of follow-up, what were they?
What clinicians can responsibly say after reading this paper
If you want a single sentence that stays inside the evidence: this prenatal cannabis exposure study found no association between meconium-detected cannabis exposure and increased well-child absenteeism, ED utilization, or developmental delay coding by age 3, in a Medicaid cohort within a mandated CPS notification environment.
What you cannot responsibly say is that cannabis improves infant development. The two-year signal is modest, statistically fragile, disappears by three years, and lacks a plausible protective mechanism within this dataset.
What you also cannot say is that cannabis in pregnancy is “proven safe.” This study does not measure first-trimester exposure, dose-response relationships, product strength, route, or later neurobehavioral outcomes. It also does not adjudicate birth outcomes literature, which is a separate evidence stream.
If you are looking for a practical counseling framework, start with your goals and risks, then build a plan around safer alternatives, symptom control, and careful follow-up. Our pregnancy guide is a reasonable starting point.
A practical reader’s guide to not overreading results
The temptation with a topic like this is to treat each new paper as a referendum. A better approach is to treat each paper as a measurement instrument. This prenatal cannabis exposure study is a particular instrument. It measures late-pregnancy detection, under a policy architecture that can reshape follow-up and coding.
If you are deciding what to do in real life, the most protective step is not finding the “right” headline. It is building a support plan with your obstetric team, addressing nausea, sleep, stress, pain, or appetite with the lowest-risk options that work for your body, and returning to the plan when circumstances change.
FAQ: prenatal cannabis exposure study interpretation
Does this prenatal cannabis exposure study prove cannabis is safe in pregnancy?
No. The study measures a specific exposure definition, cannabinoids detected in meconium, which mainly reflects later pregnancy exposure. It does not measure dose, frequency, route, potency, or early pregnancy exposure. It also does not address many outcomes that may emerge later, such as executive function or attention regulation. The defensible conclusion is narrower: in this cohort and within this follow-up window, the measured outcomes were not worse by age 3.
Why does meconium testing change how to read the results?
Meconium testing is a biomarker-based exposure definition rather than self-report or diagnostic coding. It reduces certain biases but introduces others, particularly the timing issue. Meconium primarily reflects late second and third trimester exposure, so earlier pregnancy exposure can be missed. In a prenatal cannabis exposure study, that limitation matters because timing may relate to risk in ways the dataset cannot capture.
What did the study find about well-child visits?
The study did not find a statistically significant difference in well-child care visit counts through age two between the cannabis-exposed and unexposed groups. The mean number of visits was about six overall. This suggests similar engagement with routine pediatric care in this Medicaid-linked cohort. It does not prove equal access or equal quality of care, but it argues against a major utilization gap in this dataset.
What did the study find about emergency department visits?
The study did not find a statistically significant increase in emergency department utilization through age two among cannabis-exposed infants. ED use is a mixed signal that can reflect illness burden, caregiver concern, after-hours constraints, and access to outpatient care. The most cautious conclusion is that there was no measurable utilization burden difference detected here. It does not rule out differences in specific diagnoses or later outcomes.
What does “developmental delay by ICD codes” actually mean?
It means the outcome is based on diagnostic codes recorded in claims data, not standardized developmental testing. Claims-based developmental delay depends on how often a child is seen, clinician thresholds, referral patterns, and documentation practices. It can be influenced by surveillance or follow-up intensity. In a prenatal cannabis exposure study, ICD-coded developmental delay is best understood as a healthcare system signal, not a direct measurement of child neurodevelopment.
Why was developmental delay lower at age 2 but not different at age 3?
The simplest explanation is that the two-year finding is not stable. When a signal disappears over time, it often suggests confounding, detection bias, or a transient documentation artifact rather than a durable biological effect. The dataset cannot identify dose, timing, or mechanisms that would plausibly create a true protective effect. That is why the study should not be interpreted as cannabis improving development.
How could CPS notification affect study outcomes?
If a positive screen triggers mandated reporting and structured follow-up, it can increase contact with services, change clinician attention, and alter referral pathways. That can influence what gets detected and what gets coded. The paper does not include CPS intervention data, so this remains an interpretation hypothesis rather than a proven mechanism. Still, policy context is central when reading any prenatal cannabis exposure study tied to mandated notification.
Who do these findings apply to, and who do they not apply to?
The findings apply most directly to Medicaid-insured infants in a health system using risk-based meconium screening, within a mandated CPS notification context, and within early childhood follow-up. They do not generalize cleanly to privately insured populations, universal screening environments, different state policies, or high-dose daily use contexts. They also do not generalize to first-trimester exposure patterns because the exposure definition is late-weighted. Generalizability is constrained by design.
How should a pregnant patient interpret this prenatal cannabis exposure study?
As a reminder that outcomes depend on what is measured, when it is measured, and how the system responds to exposure. This study does not support a developmental benefit narrative, and it does not provide reassurance about all exposure patterns. It does suggest that, in this setting, measured early utilization and ICD-coded developmental delay were not worse by age 3. A patient-facing next step is a supportive, individualized conversation with an obstetric clinician focused on symptom management and risk reduction.
What kind of future research would answer the harder questions?
A multi-state cohort with universal screening, quantified metabolite levels, and dose stratification would reduce exposure misclassification. Linking to CPS or plan-of-safe-care intensity could test system-mediated pathways directly. Standardized developmental testing at later ages would improve outcome sensitivity beyond claims coding. That design would help separate biological exposure effects from policy and surveillance effects in a prenatal cannabis exposure study.
Next Steps: Evidence and Care Pathways
If this prenatal cannabis exposure study prompted more questions than answers, that is appropriate. Pregnancy research is rarely binary. Below are three structured pathways, depending on what you are trying to understand.
For a comprehensive evidence overview
A broad, clinically grounded synthesis of what is known, what remains uncertain, and how to interpret mixed literature:
Expecting and Experimenting with Cannabis
For risk–benefit framing and counseling language
A structured review designed to help clinicians and patients discuss pregnancy and cannabis without exaggeration or dismissal:
Cannabis and Pregnancy: Risks, Benefits, and Care
For pediatric and longer-term safety context
A broader discussion of child development research and how early-life outcomes are measured:
Pediatric Cannabis Safety: What to Expect
If you are pregnant and navigating symptoms in real time, individualized medical guidance is more protective than generalized internet consensus.
References
1) Raffa BJ, Lanier P, Yang Y, Lin FC, Seashore C, Schilling S. Healthcare utilization and developmental delay among infants exposed to cannabis in utero. Academic Pediatrics. 2026. DOI: 10.1016/j.acap.2026.103224.
PDF |
Full text |
PubMed
2) Avalos LA, et al. Early Maternal Prenatal Cannabis Use and Child Developmental Delays. JAMA Network Open. 2024. JAMA Network Open
3) Bandoli G, et al. Prenatal cannabis use disorder and infant hospitalization and death in the first year of life. Drug and Alcohol Dependence. 2023;242:109728. DOI: 10.1016/j.drugalcdep.2022.109728. ScienceDirect | PubMed
4) Paul SE, et al. Associations Between Prenatal Cannabis Exposure and Childhood Outcomes: Results From the ABCD Study. JAMA Psychiatry. 2021;78(1):64-76. DOI: 10.1001/jamapsychiatry.2020.2902. JAMA Psychiatry | PubMed