We conducted qualitative interviews with maternal healthcare providers as well as people who use cannabis during pregnancy, an approach that has been documented in our other studies (29, 33). The research team consisted of advanced public health students (RMN, SSG), research staff (EEG), faculty co-investigators (GFD, CVB, TMB), and a faculty member (RCC) acting as the Principal Investigator of the study. We reviewed COREQ (consolidated criteria for reporting qualitative research), a validated 32-item checklist for qualitative reporting throughout the study to document important aspects of our research team, methodology, findings, and analysis (Supplementary Table S7: COREQ checklist) (34, 35).

We conducted a phenomenological study using constructivist grounded theory that aimed to understand the perspectives of two groups who have experienced a shared phenomenon, respectively: maternal health providers who care for patients who use cannabis during pregnancy and non-Hispanic Black and Latina people who used cannabis during pregnancy (36, 37).

We used theoretical sampling based on grounded theory methodology (38, 39). This study is part of a larger project focused on patient and provider experiences regarding maternal cannabis use (40). We recruited patients from the Maternal and Developmental Risks from Environmental and Social Stressors (MADRES) cohort study who enrolled pregnant persons from three locations that predominantly serve patients with Medi-Cal, a California Medicaid health program that directly pays for medical services for people with limited income through federal and state taxes (40, 41). The study coordinator contacted patients from the larger MADRES cohort who were identified as using cannabis during pregnancy through medical record abstraction; had given birth in the last 0–2 years; were 21 years or older; and fluent in English or Spanish (Table 1). We recruited maternal health providers (e.g., physicians, doulas, midwives) via selective sampling using a convenience sample and snowball approach to reach practitioners in safety net health settings in Southern California who had experience caring for people who use cannabis during pregnancy within the past year (Table 2). This type of non-random sampling was based on team discussions about selecting participants that would provide insightful information regarding the research question (42, 43). Potential participants completed a HIPAA-compliant REDCap survey to confirm eligibility, sign an e-consent form, and schedule an interview.

The semi-structured interview guides drew upon existing qualitative and quantitative literature (Supplementary Table S1: Interview guide 1, Supplementary Table S2: Interview guide 2) (27, 44–46). We piloted and adapted the questions within the research team. We revised the interview guides as the interviews progressed to refine questions and pursue areas identified as theoretically relevant. This was part of an iterative process aimed at generating richer responses from participants (39, 47).

The 60-min interviews occurred remotely via HIPAA-compliant Zoom™ video calls from 16 November 2021 to 7 February 2022 (patients, n = 7) or from 15 March 2022 to 6 April 2022 (providers, n = 10). Each interview was conducted by 1–3 research team members, with one individual leading the discussion and the others co-leading and taking analytical notes to inform analysis. We followed up on questions with open-ended inquiries about topics introduced by the participants. This non-directive, open-ended approach of qualitative interviewing encouraged participants to elaborate beyond the original scope of the interview guide and allowed for unanticipated perspectives (48). Further, the interview guide was iteratively revised based on topics introduced by participants to generate better insights in subsequent interviews (39). Throughout data collection, the theoretical sampling strategy guided responsiveness to emerging theory to identify important concepts and guide decisions about when data collection was complete (38, 39).

We sent audio recordings of interviews to an external transcriptionist who de-identified transcripts and then uploaded the files to a HIPAA-compliant One Drive for team analysis. The team made summaries of emerging ideas in transcripts after each interview was completed as part of the initial analysis. As data collection continued, emerging ideas aligned with previously observed phenomena, meaning that theoretical saturation was achieved (38, 47).

We used grounded theory methodology to generate a conceptual framework (theory or explanation) by analyzing and comparing data on experiences and discussion around an area of interest (maternal cannabis use) across the data (38). This method is best used to learn about social processes when little is known about a phenomenon (39). Once the final interview was completed, the team reviewed and categorized the summarized emerging subject areas in each transcript to develop codebooks for patient and provider datasets. The codebooks consisted of thematic categories with definitions and examples discussed and revised as a team. We then uploaded the transcripts into ATLAS.ti™ data software program, Mac Version 22.1.0, and inputted the codebooks, which were tested on one transcript from each dataset as a team and revised as needed, resulting in 16 thematic codes for each patient and provider dataset (Supplementary Table S3: Codebook 1, Supplementary Table S4: Codebook 2) for analysis (49). Then, two team members independently analyzed each transcript and assessed consistent code application to ensure intercoder reliability. Final memos were made to capture themes resulting from code overlap and to facilitate deeper discussions of the data. Final insights from this process developed into the three results as they pertain to codes and final memos related to patient-provider relationships, priorities, and access to information.

This was a phenomenological study which used constructivist grounded theory to generate results. Constructivist grounded theory is a methodology where themes are inductively generated from the data and we develop specific theories and constructs as they relate to each other, co-occur, and build upon each other to add contextual richness to our data. As such, this methodology does not necessitate results with equal numbers of each group of participants, comparison within groups, or quotes from all participants and is instead focused on demonstrating results that best represent our themes. We did not include quotes from certain providers (i.e., doulas and midwives) because the quotes chosen were the best representation of the themes presented in results. While using theoretical sampling, we reached theoretical saturation for both patient and provider cohorts indicating that our results rigorously reflect and evaluate both perspectives (38, 47).

Most providers described the prevalence of judgement towards cannabis in medicine today. One provider recounted their appraisal of patient disclosures, noting that most patients won't disclose cannabis use to their provider in fear of judgement:

Providers further explained this judgmental history via discussion of enacted screening practices that cause patients who disclose cannabis use to face negative repercussions. They explained that provider views of cannabis use leading to other substances (i.e., the gateway theory) meant patients attempting to communicate openly with providers are not only judged but subjected to punitive treatment as well:

One provider recounted socioeconomic differences in screening and care: at different care sites, lower-income patients were monitored for substances without any evidence of use, but higher-income ones were spared this surveillance. They detailed provider assumptions about patients' behavior and how that impacted which ones received more oversight:

They detailed the judgement and lack of understanding exhibited by many providers towards pregnant patients who are using cannabis. They called for divesting from this thinking via examination of patient motivations for using cannabis:

Another participant described the patient stereotypes many providers hold and emphasized the importance of exposure to patients who use cannabis to combat them. Without the opportunity to directly challenge preconceived notions regarding patients' views on pregnancy and their baby, they anticipated that the judgement towards cannabis use in pregnancy would continue:

Patients and providers described a dissonance between their priorities as well as perceptions of risks and benefits of cannabis use. While patients wanted inclusion of their reasons for turning to cannabis in care conversations, providers expressed interest in following science that tends to focus on fetal outcomes rather than maternal ones. One provider detailed concern about the long-term impacts of prenatal cannabis use on fetal well-being and thus saw their role as using patient education to protect only the baby:

Yet, from the patient perspective, they felt “misjudged” by providers who prioritized potential risk of cannabis exposure to their fetus over the challenges they currently faced during pregnancy. One individual, who was using cannabis to deal with symptoms such as nausea and lack of appetite, shared difficulties with managing not only their pregnancy but also the perceptions of their providers as they explored the nuance of caring for self and baby in a health system with limited resources:

Another participant expanded on what was lacking in their communication with providers as they trusted their own reasons for cannabis use. While their conversations were oriented towards the baby, this mother expressed a need to shift some focus to her own care and how to navigate that via harm reduction-based symptom management strategies:

Another provider described how mothers face punitive measures because of their cannabis use in pregnancy rather than being offered support. As providers decide from an individual standpoint to conduct precautionary screening and reporting of patients for using cannabis in pregnancy, they tie care to agencies that can potentially perpetuate harm. As a result, the relationship between patient and provider can be damaged:

One provider similarly emphasized the support needed by patients and their role in this. Knowing why patients are turning to cannabis and counseling or providing resources specific to this shifts the focus to encompass both maternal and fetal needs:

Both patients and providers detailed seeking out information about cannabis use during pregnancy on their own. Patients described feeling compelled to do their own research amidst acknowledgment that healthcare providers were not open to their cannabis use. Reliance on peer knowledge and social media were commonly described among participants seeking information:

Another participant recounted turning to their social networks to discuss maternal cannabis use and how these conversations were centered around the mother's experience:

Providers acknowledged seeking out information on maternal cannabis use because they felt there was limited data on the subject but did not trust the knowledge patients shared about cannabis. One healthcare provider noted the challenge of navigating conflicting information when discussing patients doing their own research on maternal cannabis use:

Another provider acknowledged that healthcare settings could take advantage of patients' drive to educate themselves and their social circles via peer-to-peer education programs, harnessing the positives of patients' desires and improving collective well-being:

This research has limitations. First, our findings may not be generalizable as we interviewed patients and providers who are based in California where cannabis is medically and recreationally legal. Second, maternal health stakeholders were not randomly selected and may not be representative of all providers, especially because of the limited maternal health roles they held. Third, though providers all cared for BIPOC patients, they had limited insights specific to this group as well as the role of their own identities outside of it, content we wish to explore in future projects focusing more on race and racism (75, 76). Fourth, despite patients giving informed consent to information from their medical records, some potential participants were surprised to be contacted for a cannabis study and refuted any use. Because younger, less educated, publicly insured, patients of color are more likely to be screened for substance use during prenatal care visits and face ensuing sociolegal ramifications, future studies would benefit from using self-reporting of cannabis use, despite greater identification of cannabis use via urine toxicology testing vs. self-report (30, 77). Fifth, this study may overrepresent people confident in their cannabis decision making and disclosure to providers about using during pregnancy.

Despite these limitations, this study includes a qualitative exploratory approach to understand perspectives, behaviors, motivations, and challenges on a taboo public health subject for both pregnant patients and maternal health providers. This supports ongoing research to better understand the stigma and barriers faced by those using cannabis while engaging in perinatal care. Further, our preliminary findings support future interventions that aim to inform providers and patients, protect patients from discrimination and criminalization during pregnancy, and preserve continuity of care.