Neighborhood Deprivation and Contrasting Patterns of Prenatal Alcohol, Cannabis, and Nicotine Use

Abstract

Introduction:

Neighborhood context is an important social determinant of health behaviors, including substance use, but less is known about whether neighborhood deprivation is similarly or differentially related to prenatal use of alcohol, cannabis, and nicotine.

Methods:

Using electronic health record data from Kaiser Permanente Northern California, this retrospective observational study examined geographic variation in prenatal alcohol, cannabis, and nicotine use and evaluated whether neighborhood deprivation was associated with use of these substances during pregnancy. The sample included 173,578 pregnancies (November 2020-December 2024) among members with prenatal screening for substance use during early pregnancy as part of standard care. Neighborhood deprivation index (NDI) from geocoded census data was categorized into quartiles. Choropleth maps visualized tract-level prevalence of prenatal substance use, and modified Poisson regression models estimated associations between NDI quartiles and prenatal substance use using individual-level data. Data were analyzed July 2025-Feburary 2026.

Results:

Prevalence of prenatal alcohol, cannabis, and nicotine was 9.6%, 8.9% and 2.0%, respectively. Compared with pregnancies in the least deprived neighborhoods, those in the most deprived neighborhoods had a lower prevalence of prenatal alcohol use (aPR [95% CI] = 0.87 [0.84, 0.92]), but higher prevalence of prenatal cannabis (aPR [95% CI] =1.80 [1.70, 1.90]) and nicotine use (aPR [95% CI] = 1.96 [1.74, 2.20]).

Conclusions:

Neighborhood deprivation showed opposing associations across substances, with lower alcohol use but substantially higher cannabis and nicotine use in more deprived neighborhoods. Findings underscore the importance of considering neighborhood context and type of substance when developing preventive interventions.

INTRODUCTION

Prenatal alcohol, cannabis, and nicotine use are associated with adverse maternal and neonatal outcomes and remain important targets for prevention.^1-4 Neighborhood context is recognized as an important social determinant of health behaviors, including substance use.⁵ Prior studies have found that individuals in more socioeconomically deprived neighborhoods have higher rates of cannabis and tobacco use,^6-8 whereas associations with alcohol use are more heterogeneous.^7,9,10

Understanding whether neighborhood deprivation is associated with prenatal alcohol, cannabis, and nicotine use in similar or divergent ways may inform targeted prevention, yet few studies have compared socioeconomic gradients across substances within the same prenatal population.

Using electronic health record (EHR) data from a large, integrated healthcare system, this study examined geographic variation in prenatal alcohol, cannabis, and nicotine use and evaluated whether neighborhood deprivation is differentially associated with use of these substances during pregnancy.

METHODS

This retrospective observational study used EHR data from Kaiser Permanente Northern California (KPNC), a large healthcare system serving ~4.6 million members broadly representative of Northern California’s insured population. KPNC universally screens for substance use at entry to prenatal care using a self-report prenatal screening questionnaire (PSQ) and consented urine toxicology test.

The sample included pregnancies with onset between 11/2020-12/2024 among KPNC members who had a prenatal care office visit. Pregnancies not linkable to a valid self-report or urine toxicology test for prenatal substance use, and pregnancies for those whose address could not be geocoded were excluded. The final analytic sample included 173,578 pregnancies among 148,301 individuals with non-missing data on at least one of the three substances examined (Appendix Figure1; Appendix Table1). For each pregnancy, the associated residential address recorded in the EHR was geocoded.

This study followed STROBE guidelines and was approved by the KPNC IRB with a waiver of informed consent.

Measures

Prenatal cannabis use was defined as any self-reported use since pregnancy on the PSQ or a positive toxicology test at entry to prenatal care (Appendix Methods). Prenatal alcohol use was defined as any self-reported alcohol use since pregnancy on the PSQ. Prenatal nicotine use was defined as any self-reported tobacco use (e.g., cigarettes, cigars, hookah) or nicotine vaping since pregnancy on the PSQ (Appendix Methods).

Neighborhood deprivation index (NDI), derived from geocoded census data based on the American Community Survey for the same year as the PSQ screening date¹¹ was categorized into quartiles based on the distribution of the KPNC population, with the first quartile reflecting the least deprivation.

Baseline covariates extracted from the EHR included: age (<18, 18-24, 25-34, >34 years), self‐reported race/ethnicity (Asian, Black, Hispanic, Native American, Hawaiian/Pacific Islander, Multiracial, White and Other), and parity (0, 1, ≥2, missing). Limited missing demographic data were coded as a separate category.

Statistical analysis

Census tract-level NDI and prenatal substance use prevalence were generated for tracts with ≥5 pregnancies (N=2,720) and visualized using ArcGIS Pro. Individual-level associations between NDI and prenatal substance use were examined using modified Poisson regression with robust variance to estimate adjusted prevalence ratios (aPRs) and 95% CIs, adjusting for individual-level sociodemographic characteristics and year and accounting for repeated pregnancies. Trend analyses treated NDI quartile continuously. Sensitivity analyses examined weekly/daily self-reported alcohol and nicotine use. Analyses used SAS 9.4.

RESULTS

Among 173,578 pregnancies from 148,301 unique individuals, 68.3% were non-White and 10.4% were aged <25 years. Overall prevalence of prenatal alcohol, cannabis, and nicotine was 9.6%, 8.9% and 2.0%, respectively (Appendix Table2).

NDI and prenatal alcohol, cannabis, and nicotine use varied substantially across census tracts (Figure1; Appendix Figure2). Alcohol use was more common in less deprived neighborhoods, whereas cannabis and nicotine use were more common in more deprived neighborhoods.

Figure 1.

Geographic variation in neighborhood deprivation index (NDI) and prenatal alcohol, cannabis and nicotine use

Notes. Choropleth maps show census tract-level NDI and prevalence of prenatal alcohol, cannabis, and nicotine use among pregnancies receiving prenatal care at KPNC. Tracts with fewer than five pregnancies were excluded. NDI was summarized into quartiles, and substance specific quintiles were used due to differences in prevalence. Darker shades indicate higher prevalence. Geographic patterns mirrored socioeconomic gradients, with higher alcohol use concentrated in less deprived areas and higher cannabis and nicotine use concentrated in more deprived areas. Tract-level prevalence ranged from 0.0%-55.6% for alcohol, 0.0%-54.3% for cannabis, and 0.0%-20.0% for nicotine.

In tract-level analyses, higher neighborhood deprivation was associated with lower prenatal alcohol use, with mean prevalence decreasing from 11.5% in the least deprived quartile to 8.9% in the most deprived quartile (Appendix Table3). In contrast, prenatal cannabis and nicotine use increased with neighborhood deprivation, rising from 4.6% to 12.8% for cannabis and from 1.2% to 2.6% for nicotine from least to most deprived NDI quartiles.

In multivariable individual-level models, these contrasting associations persisted after adjustment for sociodemographic characteristics (Table 1). Compared with pregnancies in the least deprived neighborhoods (1^st quartile), those in the most deprived neighborhoods had a lower prevalence of prenatal alcohol use (aPR [95%CI]= 0.87 [0.84,0.92] for the 4^th quartile), but substantially higher prevalence of prenatal cannabis (aPR [95%CI]=1.80 [1.70,1.90]) and nicotine use (aPR [95%CI]=1.96 [1.74,2.20]). Trend analyses showed significant linear associations between neighborhood deprivation and prenatal use of all three substances (Table 1). Results from sensitivity analyses examining weekly or daily alcohol and nicotine use (Appendix Table4) supported the robustness of the observed substance-specific patterns.

Table 1.

Associations between Quartiles of Neighborhood Deprivation Index and Prenatal Substance Use

	Alcohol	Cannabis	Nicotine
	aPR (95% CI) a	aPR (95% CI) a	aPR (95% CI) a
Quartile of Neighborhood Deprivation Index (NDI)
1st quartile-least deprived	1.00	1.00	1.00
2nd quartile	0.98 (0.94, 1.02)	1.33 (1.26, 1.40)	1.39 (1.24, 1.56)
3rd quartile	0.92 (0.88, 0.96)	1.60 (1.52, 1.69)	1.61 (1.44, 1.80)
4th quartile-most deprived	0.87 (0.84, 0.92)	1.80 (1.70, 1.90)	1.96 (1.74, 2.20)
P value for trend analysis b	<.001	<.001	<.001

Notes. Bolding indicates significance at p<0.05 level. NDI quartiles based on distributions among KPNC members; the 1^st NDI quartile indicates the lowest NDI (highest socioeconomic status), and the 4^th quartile indicating the highest NDI (lowest socioeconomic status).

^aAdjusted prevalence ratios (aPR) and 95% confidence intervals (95% CI) estimated from modified Poisson models with an exchangeable correlation matrix using individual-level data. Separate models were fit for each prenatal substance use outcome, each excluding those with missing data on the substance examined (that is, N=173,196, N=173,032 and N=173,493 for the model examining alcohol, cannabis and nicotine, respectively). All models adjusted for age (14-17, 18-24, 25-34, ≥35), race and ethnicity (Asian, Black, Hispanic, Native American, Hawaiian/Pacific Islanders, Muti-racial, White and Other), parity (0, 1, ≥2, missing), and year while accounting for multiple pregnancies per individual.

^bThe trend analysis was also conducted by refitting the models including NDI quartiles as a continuous variable, with results suggesting significant linear relationships between NDI quartiles and prenatal use of all three substances.

DISCUSSION

In this large, population-based study, neighborhood deprivation showed contrasting associations across substances, with lower prenatal alcohol use but higher cannabis and nicotine use in more deprived neighborhoods. These findings do not support assumptions that prenatal substance use is uniformly more common in socioeconomically disadvantaged communities; rather, socioeconomic patterns differed by substance.

Findings are consistent with prior research showing higher cannabis and tobacco use in more socioeconomically deprived neighborhoods and more heterogeneous associations between neighborhood deprivation and alcohol use. This study extends prior work by examining alcohol, cannabis, and nicotine within the same prenatal population and screening system, using current early-pregnancy substance use measures and an objective measure of cannabis use. Higher prenatal alcohol use in less deprived neighborhoods is consistent with some prior studies showing that prenatal alcohol use in the US is inversely associated with neighborhood deprivation status.¹² Differences in social norms, stigma, perceived acceptability, and willingness to disclose use during prenatal screening may also contribute to the observed patterns across substances. For example, moderate alcohol use may be more socially acceptable or more readily disclosed in less deprived neighborhoods, whereas cannabis and nicotine use may be shaped by different community norms, stressors, retail environments, and histories of surveillance or reporting. Cannabis and nicotine use may be more concentrated in disadvantaged communities due to structural and social conditions that operate differently by substance, including economic hardship, chronic stress, limited access to smoke-free housing, product availability, cost, targeted marketing, and local retail density.^13,14

Strengths include universal substance use screening at entry to prenatal care, reducing reliance on postpartum recall; assessment of alcohol, cannabis, and nicotine within the same population and screening system; and linkage of geocoded residential addresses to neighborhood deprivation.

Limitations include reliance on self-report for alcohol and nicotine use, whereas cannabis use was measured using self-report and urine toxicology; underreporting may therefore have differed by substance and neighborhood deprivation. Substance use was measured in early pregnancy and may have occurred before pregnancy recognition. Because the study began during the COVID-19 pandemic, care-seeking, screening completion, or substance use may have changed over time; however, screening remained standard care, and models adjusted for year. Findings may not generalize to uninsured populations, other regions, or the U.S. population overall. Neighborhood deprivation is area-level and may not reflect individual socioeconomic status. Cannabis toxicology testing was unavailable for a small proportion of otherwise eligible pregnancies, which may have underestimated cannabis use. Future studies should examine broader prenatal polysubstance use patterns, including opioid use and co-occurring cannabis and nicotine use.

From an equity perspective, results highlight that socioeconomic patterns of prenatal substance use differ by substance. These findings do not support selective screening based on neighborhood characteristics; universal, non-stigmatizing prenatal screening for all substances should remain standard. Instead, neighborhood context may inform population-level prevention, resource allocation, outreach, and health system planning. As cannabis use increases and products become more widely available, higher prenatal cannabis use in more deprived neighborhoods has important implications for equity-focused prevention, counseling, and outreach. Tailoring prevention strategies to neighborhood conditions may improve interventions addressing prenatal substance use.