In this multicenter cohort of young children presenting to the PED after unintentional cannabis ingestion, cases increased substantially over time, with most exposures occurring in children under 2 years of age and primarily involving edible products. Clinical acuity was high, with frequent triage assignment of ESI 1 and 2, substantial use of resuscitation resources, and high rates of hospital admission, including critical care. We found high rates of involvement of SW and frequent CPS reporting in our population; there was no association between deprivation index or racial ethnic background to SW or CPS engagement or disposition planning.

Our findings align with accumulating evidence that pediatric unintentional cannabis ingestions have increased in the post-legalization era and that edibles are a driver of pediatric exposures [2, 3, 6, 13, 36] Edible products are often packaged in ways that resemble conventional snacks and stored in homes where adult users are present, factors that likely amplify risk in young children who are in the oral exploration stage [11, 12, 15]. Regulatory steps such as child-resistant packaging and dosing standards may mitigate risk, yet data suggest these measures have not been sufficient to reverse the upward trajectory of exposures; notably, there are no federal laws regulating safe storage of cannabis products [18–20].

The high proportion of children triaged as ESI 1 or 2, frequent trauma bay activations, and substantial admission rates underscore the significant resource burden that unintentional cannabis places on pediatric emergency and inpatient systems. Although no standardized THC dose threshold for toxicity in children has been established, higher doses are associated with severe and prolonged toxicity [19]. it is reasonable to infer that younger children are at greater risk due to lower body weight and the potential to ingest large quantities of highly concentrated edible products. These findings underscore the significant clinical burden of pediatric unintentional cannabis ingestion and highlight the growing strain on pediatric healthcare systems as cannabis becomes more accessible nationwide.

Despite prior literature documenting disparities in SW and CPS involvement across pediatric evaluations and in ingestion cohorts (e.g., higher referral rates by race/ethnicity in some settings) [23, 27, 28], our multicenter analysis did not detect associations between deprivation index or race/ethnicity and SW or CPS activation or disposition outcomes. Instead, SW and CPS involvement appeared consistent across groups. In the absence of national standards for mandatory reporting in cases of cannabis ingestions, we hypothesize that reporting practices in our study population may be driven by clinical severity, perceived risk to the child, or evolving institutional norms. It is also plausible that societal and legal ambiguity surrounding cannabis, particularly in the transition from medical to recreational legalization, contributes to a lower threshold for reporting. As social acceptance and regulatory frameworks stabilize, these practices may evolve.

These findings carry important implications for both clinical care and public health policy. The frequent involvement of SW and CPS in our study, compared to variable rates in prior studies [2, 7, 8, 13], supports consideration of standardized pathways to guide social evaluation in cases of unintentional cannabis ingestion. This may help reduce institutional variation and support equitable decision-making. Notably, providers may consider the circumstances surrounding individual ingestion cases to determine if reporting is necessary, rather than universal mandatory reporting [37–39]. Given that most exposures occurred at home and were linked to edible products, integrating targeted developmentally appropriate safe-storage counseling into emergency and primary care encounters is essential; this should include emphasis on locked, hidden storage and non-appealing, child-resistant packaging. Collaboration between clinicians, public health agencies, and regulatory bodies is critical as cannabis products evolve, particularly with respect to packaging standards, clear dosing units, and restrictions on child-appealing formulations. Finally, the substantial proportion of children presenting with high acuity and requiring admissions suggest that pediatric emergency systems should anticipate continued resource demands related to cannabis ingestion and incorporate this trend into planning for staffing, training, and clinical protocols as legalization and product availability expand.

It is important to note that there are several limitations to this study. First, while we found no difference of sociodemographic factors on safe disposition planning, CPS evaluation continues beyond discharge and can vary by state and region; for example, this could mean an eventual home safety evaluation or placement back with a primary caregiver despite initial placement with a relative on discharge. It is unclear if these ultimate decisions are affected by sociodemographic factors. Additionally, families may call poison control for recommendations prior to hospital evaluation; local policies can vary in terms of poison control center mandatory reporting and recommendations for emergency evaluation. As such, some children with unintentional ingestions may be missed in this cohort if they did not present for PED care. While we have considered certain sociodemographic variables in our analysis, they may be other factors unaccounted for that affect the social management and disposition of these children, such as prior CPS involvement due to concerns of abuse or neglect. Additionally, some cases may be missed due to synthetic cannabinoid products, which may not prompt positive THC results on conventional UDS. Similarly, in patients presenting with a known ingestion, a UDS may not have been obtained and thus could be missing from the data set. There may also be data that is missing or incorrectly listed in the EMR. It is also noted that although the overall cohort exceeded 200 patients, several variables of interest, including non-English language status and absence of SW consultation, occurred infrequently. The resulting wide confidence intervals suggest limited power to detect potential associations. Future studies with larger samples are needed to further evaluate these relationships. Additionally, site-level clustering was not incorporated into the regression models. Because several outcomes were highly prevalent (e.g., social work consultation occurred in > 95% of cases) and the overall sample size was modest (N = 266), meaningful site-level variation was considered unlikely. However, residual cluster-level confounding related to unmeasured differences across sites cannot be excluded. Finally, this study was limited to two PED; these results may not be generalizable to the larger population.