Systematic Review: Polysubstance Prevalence Estimates Reported during Pregnancy, US, 2009–2020

Abstract

Introduction

The objective of this systematic review is to describe polysubstance studies and their prevalence estimates among pregnant people in the US.

Methods

This review was not subject to protocol preparation or registration with the International Prospective Register of Systematic Reviews (PROSPERO) because outcome data were not reported. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Checklist was followed. Four scientific literature databases were used to identify articles published from January 1, 2009 to June 3, 2020 reporting prenatal exposure to two or more substances in the US. A standardized process of title and abstract screening followed by a two-phase full-text review was used to assess study eligibility.

Results

A total of 119 studies were included: 7 case–control studies, 7 clinical trials, 76 cohort studies, and 29 cross-sectional studies. Studies varied with respect to study design, time period, region, sampling and participant selection, substances assessed, and method of exposure ascertainment. Commonly reported polysubstance prevalence estimates among studies of pregnant people included combinations with alcohol, marijuana, and/or tobacco/nicotine. The range of prevalence estimates was wide (alcohol 1–99%; marijuana 3–95%; tobacco/nicotine 2–95%).

Discussion

Polysubstance use during pregnancy is common, especially with alcohol, marijuana, and/or tobacco/nicotine. Future research to assess polysubstance use during pregnancy could help better describe patterns and ultimately help mitigate its effects on maternal and infant health outcomes.

Introduction

Substance use during pregnancy continues to be a major public health challenge. The National Survey on Drug Use and Health (NSDUH) indicated that in 2019 and 2020 between 18 and 21% of pregnant females reported using alcohol tobacco products, and/or illicit drugs in the past month (Substance Abuse and Mental Health Services Administration, 2020a). Polysubstance use (intentional or unintentional exposure to more than one substance) (Centers for Disease Control & Prevention, 2021) is a growing public health concern. According to the Survey of Key Informants’ Patients (SKIP) program, which includes individuals in treatment at a participating opioid use disorder (OUD) treatment center, more than 90% of participants who used opioids reported using at least one other non-opioid drug in the past month (Cicero et al., 2020).

Polysubstance use during pregnancy may complicate clinical assessments during pregnancy or delivery hospitalization. For example, neonatal abstinence syndrome (NAS), or the occurrence of drug withdrawal symptoms in newborns, is a treatable condition that can occur in the first few weeks of life, but more evidence is needed regarding how to tailor management based on which substance exposures occurred during pregnancy (Choo et al., 2004; Jones et al., 2013; Morris et al., 2020; Patrick et al., 2020). Moreover, the risk of overdose in the pregnant individual can increase when substances with central nervous system depressant effects (e.g., alcohol, barbiturates, benzodiazepines, opioids) are used in combination with one another (U.S. Food and Drug Administration, 2017).

For pregnant individuals, national prevalence estimates of polysubstance use in any combination are usually limited to sub-groups from surveys (e.g., NSDUH), administrative data [e.g., the Treatment Episode Data Set (TEDS)], or hospital discharge data [e.g., the National Inpatient Sample (NIS)]. There are several limitations to these approaches as definitions are limited to self-report or what is available in medical records, which often are not comprehensive (England et al., 2020; Jarlenski & Krans, 2021; Jarlenski et al., 2020; Washio et al., 2018). Additionally, survey data such as those from NSDUH need to be weighted appropriately to generate national prevalence estimates (Substance Abuse and Mental Health Services Administration, 2020b). A systematic review describing the estimated prevalence of polysubstance use during pregnancy across multiple types of study designs could inform future prevention research and recommendations for management and treatment. The objective of this systematic review is to describe polysubstance prevalence during pregnancy from studies published in the US from 2009 to 2020.

Methods

Protocol and Registration

This systematic review on prevalence estimates of polysubstance use among pregnant people in the US was not based upon a clinical study and was not subject to protocol preparation or registration with the International Prospective Register of Systematic Reviews (PROSPERO) because outcome data were not reported. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Checklist was followed to guide reporting methods and results (Online Appendix Tables A1 and A2).

Information Sources and Searches

The authors and a librarian from the Centers for Disease Control and Prevention (CDC) developed the search strategy (Table 4). Using a comprehensive approach, the librarian searched for articles published in four scientific databases (i.e., CINAHL, Embase, Medline, and PsycINFO) from the date of inception of each database to June 3, 2020. Search terms, based on substances included in the 2019 NSDUH (Substance Abuse and Mental Health Services Administration, 2020a), included “pregnancy” and variations on and combinations of the following terms: polysubstance, substance use, alcohol, amphetamine, cocaine, hallucinogen, inhalant, cannabis, nicotine, opioid, and stimulant. Additional articles published after 2009 included one article identified through relevant citations, four articles through PubMed email alerts related to opioid, tobacco, other substance use, and pregnancy, and thirteen articles from the full text of articles of the original database results. Database results were imported, deduplicated, and filtered by articles published on or after 2009 in an EndNote Library to capture literature from the most recent past decade. A flowchart of the study selection process is shown in Fig. 1.

Fig. 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram of studies included in a systematic review describing polysubstance prevalence estimates reported during pregnancy, U.S., 2009–2020

Eligibility Criteria, Data Collection, and Study Selection

Two independent reviewers screened 3377 titles and abstracts for eligibility utilizing a standardized process in Microsoft Excel and Covidence (Covidence systematic review software, 2020). Full-text articles in the English language were eligible for inclusion during title and abstract screening if they reported the use of at least one substance during pregnancy in the US as studies that focused on one substance could have referred to other substance exposures during pregnancy within the full text. During title and abstract screening, articles were excluded for the following reasons: animal-related; basic science/in vitro-related; data only available as an abstract or poster; duplicate of an article already included in search results; not original data (e.g., review article, guidelines, recommendations); non-English language; study location outside of the US; non-pregnant study population; or no mention of at least one substance. Discrepancies between reviewers were resolved among the reviewers or by a third reviewer. Title and abstract screening resulted in 961 articles for full-text review.

A two-phase full-text review was conducted to further assess study eligibility in a standardized process in Microsoft Excel. For phase 1, the full text of articles that passed title and abstract screening were retrieved and independently reviewed for eligibility. For full-text review, polysubstance use was defined as the use of two or more substances from different drug classes at any time during the index pregnancy. Phase 1 exclusions were based on a more detailed list of criteria and were prioritized in the following order: non-pregnant study population; the article did not report two or more substances from different drug classes or did not clearly define at least one substance for a given polysubstance combination; the study was not written in English; the study location was outside of the US; the data were only available as an abstract, poster, dissertation, or commentary; the article was a review; the study did not include enough data to calculate a prevalence estimate for polysubstance use; the article was a duplicate. Phase 1 full-text screening resulted in 186 articles for phase 2 of full-text review.

Phase 2 of full-text review was conducted by two reviewers using a standard form in Microsoft Excel. Data were abstracted by an individual reviewer and included general study characteristics (i.e., study location, study time period, strengths and limitations), study methods [i.e., study design, sampling and/or survey methods, how pregnancy was determined, substances measured, how exposure was ascertained (e.g., self-report, abstracted from the medical record, maternal or infant biomarkers)], study population (e.g., inclusion and exclusion criteria, race and ethnicity), and polysubstance prevalence estimates. Race and ethnicity categories were based on the Office of Management and Budget minimum categories for data collection standards (Office of the Assistant Secretary for Planning & Evaluation, 2011). Study locations were based on the US Census Bureau regions (U.S. Census Bureau, 2021).

Prevalence estimates were abstracted for any combination of two or more defined substances from different drug classes (e.g., alcohol and tobacco/nicotine) or for at least one defined substance in combination with a group of substances (e.g., alcohol and illicit drugs). The base substance was the primary substance reported defining the study population and served as the denominator for polysubstance prevalence calculations (e.g., people who used alcohol during pregnancy). The secondary substance was the other substance reported and served as the numerator of polysubstance prevalence calculations (e.g., tobacco/nicotine use among people who used alcohol during pregnancy). Although most studies reported polysubstance prevalence estimates from the perspective of a base substance category used during pregnancy, some studies reported overall use patterns without reference to a base substance (referred to as the “no base substance” category). To better describe more specific subpopulations of polysubstance groups, if either methadone or buprenorphine products were specified from the “opioids” base substance category, these prevalence estimates were moved to a separate base substance category for “medications for opioid use disorder (MOUD).” Additionally, prevalence estimates for “amphetamine/methamphetamine” and “cocaine” were separated from the “other” base substance category, and the “barbiturates” and “benzodiazepines” were separated from the “other” secondary substance category.

To finalize the phase 2 full-text review, a secondary reviewer conducted an unblinded review of abstracted data on study characteristics to ensure accuracy and completeness, and a blinded review of abstracted polysubstance prevalence estimates. Discrepancies were resolved between the two reviewers. Articles were excluded during phase 2 for the following reasons: pre-pregnancy or postpartum periods were included in prevalence estimates, or it was not possible to confirm estimates were solely based on pregnancy period; exposure was reported by a caregiver other than the birthing parent; the study design was a case series or data were from focus groups; article only reported the quantity of substance consumed (e.g., number of cigarettes smoked; days of drinking) but did not report the number of people who used each substance. Articles using the same data source in the same study years were included if they included a unique study population and/or unique polysubstance combination and prevalence estimates. Among cohort studies with overlapping study populations and study years, the studies that reported the least number of unique polysubstance combinations were excluded. Excluded articles were categorized based on the prioritized list mentioned previously, but articles could fall under more than one exclusion criteria category (Fig. 1).

Results

All decisions to exclude any article during the screening process were made by group consensus among coauthors. Descriptive analysis of abstracted data from 119 eligible articles (Table 1) was conducted using SAS software (SAS Institute, version 9.4).

Table 1.

Characteristics of studies that reported polysubstance prevalence estimates among pregnant people, US, 2009–2020 (N = 119)

Article ID	Author(s), year	Study design	First year of study time period	Region	Setting	Total sample sizea	How pregnancy was determined
1	Alhusen et al. (2013)	Cohort	2009	South	Three urban clinics providing obstetrical care	Total sample size = 166; total subsample of women who used marijuana = 64/166	Medical record
2	Aliyu et al. (2009)	Cohort	1989	Midwest	Statewide	Total = 655,758; total who reported drinking alcohol = 14,444	Vital records
3	Allen et al. (2014)	Cohort	1993	Northeast	Hospital-based prenatal clinics in a low socioeconomic status area. Mercer Medical Center in Trenton, NJ and Medical College of Pennsylvannia	Total sample = 114; total with prenatal cocaine exposure = 48	Identified at prenatal care
4	Almario et al. (2009)	Cohort	2000	Northeast	Outpatient methadone treatment center	Total women = 258	Medical record
5	Ashford et al. (2019)	Cohort	Not reported	South	Academic and private prenatal clinics	Total sample of pregnant women in first trimester who used tobacco within 30 days = 138; total who only used tobacco (vs tobacco + cannabis) = 100; total who used both tobacco and cannabis = 38	Medical record
6	Bada et al. (2012)	Cohort	1993	Includes states across ≥ 2 regions	Brown University, University of Miami, University of Tennessee, Memphis, and Wayne State University	Total who used cocaine = 350; total with high prenatal cocaine exposure (PCE) = 115/350; total with some PCE = 235/350	Identified mothers of living children
7	Baer et al. (2017)	Cohort	2005	Pacific West	N/A	4588	Medical record
8	Baewert et al. (2012)	Clinical Trial	2005	Includes states across ≥ 2 regions	Rural (VT, TN) and urban (MD, PA, MI, RI) universities in the US: Johns Hopkins University School of Medicine (lead site), Thomas Jefferson University, Vanderbilt University School of Medicine, Wayne State University, University of Vermont, Alpert School of Medicine at Brown University	Total pregnant who used opioids = 94	Part of study criteria
9	Bailey et al. (2012)	Cohort	Not reported	South	6 rural medical practices providing prenatal care	265	Identified at prenatal care
10	Bakhireva et al. (2019)	Cohort	2013	West	Metropolitan clinics	42	Medical record
11	Barlow et al. (2010)	Clinical Trial	2006	South		322	Not reported
12	Beatty et al. (2012)	Cohort	Not reported	Midwest	Urban hospital	60	Delivered at hospital site
13	Bolin et al. (2019)	Cohort	Not reported	South	UAMS Women’s Mental Health Program, a tertiary referral center with primary focus on psychiatric disorders	8	Part of study criteria
14	Brogly et al. (2018)	Cohort	2015	Northeast	Obstetric and addiction recovery clinic at an urban, safety-net hospital	113	Medical record
15	Chabarria et al. (2016)	Cohort	2011	South	Tertiary referral hospitals associated with Baylor College of Medicine	Total pregnant women = 12,069; total pregnant women reporting marijuana use = 106	Delivered at hospital site
16	Chambers et al. (2019)	Cross-sectional	2012	Pacific West	Large urban city in Pacific Southwest	93	Identified mothers of children with FASD
17	Chisolm et al. (2009)	Cross-sectional	2000	South	Johns Hopkins Center for Addiction and Pregnancy, a comprehensive program for substance-dependent pregnant women	122	Medical record
18	Chung et al. (2010)	Cohort	2000	Northeast	FQHCs and FQHC Look-Alikes	Total women = 1476	Medical record
19	Coleman-Cowger et al., (2018a, 2018b)	Cohort	2017	South	Two obstetric clinics	Total general pregnant women = 500; total co-use of marijuana and tobacco = 45	Pregnant at time of study
20	Coleman-Cowger et al., (2018a, 2018b)	Cross-sectional	2006	Coverage across ≥ 45 states	Household survey	Total pregnant women = 8,695	Self-report
21	Conner et al. (2015)	Cohort	2004	Midwest	Washington University, St. Louis Medical Center, tertiary care facility	Total general pregnant women = 8,138; Total who used marijuana = 680	Delivered at hospital site
22	Crume et al. (2018)	Cross-sectional	2014	West	N/A	Total general pregnant women = 3,207	Delivered at hospital site
23	De Genna et al. (2014)	Cohort	Not reported	Northeast	Magee-Women’s Hospital prenatal clinic	92	Identified at prenatal care
24	Desai et al. (2015)	Cohort	2000	Coverage across ≥ 45 states	Population-based	Alcohol base substance: pregnant women with a history of alcohol or non-opioid drug misuse or dependence, but no history of opioid misuse or dependence: short-term = 23,580; long-term = 4,973.Opioid base substance: pregnant women with a history of opioid misuse or dependence: short-term = 2,677; long-term = 2,139	Other (specify): Administrative data
25	Dukes et al. (2017)	Cohort	2007	Midwest	Comprehensive clinical sites	Total pregnancies from people who were reported as American Indian race = 2,021; total pregnancies from people who were reported as White race = 2,672	Medical record
26	Eggleston et al. (2009)	Case–Control	1999	South	Comprehensive perinatal drug treatment program	105	Not reported
27	Eiden et al. (2020)	Cohort	Not reported	Not reported	Large urban prenatal clinic	Total dyads = 238; total with tobacco exposure = 150/238	Identified at prenatal care
28	Forray et al. (2014)	Clinical Trial	2006	Northeast	Two inner city hospital-based reproductive health clinics	Total = 176; total who smoked in pregnancy = 122	Not reported
29	Gaalema et al. (2013)	Clinical Trial	Not reported	Northeast	Obstetric practices in Burlington, Vermont	Total participants = 115	Not reported
30	Gallagher et al. (2017)	Cohort	2006	Northeast	Records of households receiving PCE (tetrachloroethylene) exposed water	All alcohol exposed. Total prenatally exposed to tetrachloroethylene = 302; total not prenatally exposed to tetrachloroethylene = 201	Vital records
31	Garrison-Desany et al. (2020)	Cohort	1998	Northeast	Boston Medical Center (urban hospital)	Total mother-newborn pairs = 8,261; total who used stimulant drugs (cocaine, amphetamines) = 131; total who used depressant drugs (heroin, methadone, barbiturates) = 192	Delivered at hospital site
32	Gauthier et al. (2010)	Cohort	Not reported	South	Two urban large hospitals	Total women = 321; total who used alcohol = 83/321	Medical record
33	Godleski et al. (2018)	Cohort	Not reported	Northeast	Local hospital (appears to be in Buffalo, NY)	Total exposed to tobacco = 178	Not reported
34	Goldschmidt et al. (2012)	Cohort	1982	Northeast	Urban hospital-based prenatal clinic	Total women = 579; total who smoked = 306/579	Part of study criteria
35	Good et al. (2010)	Cohort	2000	West	St Joseph’s Hospital and Medical Center (697 bed, tertiary care, urban, academic, not-for-profit medical center)	Patients who used methamphetamine = 276	Medical record
36	Grant et al. (2009)	Cohort	1989	Pacific West	Hospitals	Study 1 = 7178; study 2 = 2230; study 3 = 3118	Delivered at hospital site
37	Gray et al., (2010a, 2010b)	Cohort	Not reported	South	Urban, multi-disciplinary care treatment for drug-dependent women	49	Not reported
38	Gray et al., (2010a, 2010b)	Cohort	Not reported	Northeast	Not reported	Total pregnant women = 86; total who used cannabis = 38/86	Not reported
39	Guille et al. (2020)	Clinical Trial	2017	South	4 outpatient obstetric practices—Women’s Reproductive Behavioral Health Program at Medical University of South Carolina	Total women = 98; total who received in-person treatment = 54/98; total who received telemedicine = 44/98	Medical record
40	Hand et al. (2017)	Cross-sectional	2013	Coverage across ≥ 45 states	TEDS data from private and public substance use disorder treatment facilities that receive public funding	8656	Self-report
41	Harrison and Sidebottom (2009)	Cohort	2005	Midwest	Four urban Federally Qualified Healthcare Centers	Total women = 1492; Total who used alcohol and another drug while pregnant = 40/1,492	Not reported
42	Hensley et al. (2018)	Cohort	2009	Midwest	2 large referral and tertiary care centers	74	Medical record
43	Holtrop et al. (2010)	Cross-sectional	2005	Midwest	Medicaid eligible women in Michigan’s enhanced prenatal services program	Total pregnant women = 2203; total who smoked = 566/2203	Not reported
44	Huybrechts et al. (2017)	Cohort	2000	Coverage across ≥ 45 states	Medicaid records	Total pregnancies with exposure to prescription opioids 45 days before delivery: from article text = 201,275; from article tables = 200,705	Medical record
45	Jansson et al. (2017)	Cohort	Not reported	South	Not stated but treatment center is in Baltimore	49	Medical record
46	Jarlenski et al. (2017)	Cross-sectional	2005	Coverage across ≥ 45 states	Nationally representative survey	Weighted = 205,979; unweighted = 101	Self-report
47	Kiblawi et al. (2014)	Cohort	2002	Includes states across ≥ 2 regions	Urban clinical sites	185	Delivered at hospital site
48	Kim et al. (2009)	Cross-sectional	1996	Pacific West	State-based	11,837 (unweighted)	Vital records
49	Ko et al. (2015)	Cross-sectional	2007	Coverage across ≥ 45 states	National	Total pregnant women = 4971; total who used marijuana = 265 (results stratified by past 1 month and past 12 month use)	Self-report
50	Ko et al. (2018)	Cross-sectional	2009	Includes states across ≥ 2 regions	State-based	Total = 9013; total with marijuana use = 463/9013	Vital records
51	Kozhimannil et al. (2019)	Cross-sectional	2007	Coverage across ≥ 45 states	Rural	Total = 942,798; total with OUD = 4606	Delivered at hospital site
52	Kozhimannil et al. (2017)	Cross-sectional	2005	Coverage across ≥ 45 states	National	Total pregnant women ages 12–44 = 8721; total weighted = 23,855,041. Total (weighted) with past-month opioid use = 217,106	Self-report
53	Kreitinger et al. (2016)	Cohort		West	Patients at University of New Mexico’s prenatal substance use program	70	Identified at prenatal care
54	Kreshak et al. (2016)	Cohort	2014	Pacific West	Tertiary care university health system’s urban and suburban ambulatory obstetric offices	295	Medical record
55	LaGasse et al. (2011)	Cohort	Not reported	Includes states across ≥ 2 regions	4 urban hospitals	183	Delivered at hospital site
56	Lappen et al. (2020)	Cohort	2000	Midwest	Comprehensive perinatal opioid dependency program (MetroHealth Medical Center)	Total women on MAT = 480; total on methadone = 345; total on buprenorphine = 137	Medical record
57	Lee et al. (2020)	Cohort	2016	Pacific West	Referral clinic serving patient population with public insurance	Total women = 466; total who tested positive for marijuana = 45	Medical record
58	Leszko et al. (2020)	Cohort	2013	Includes states across ≥ 2 regions	Racially diverse regions	603	Identified at prenatal care
59	Lopez et al. (2011)	Cohort	2005	Midwest	Eight maternity clinics; public and private sector prenatal clinics	Total who continued smoking = 197	Medical record
60	Lu et al. (2009)	Cohort	Not reported	Pacific West	Unspecified clinical setting	Methamphetamine group = 14	Identified mothers of living children
61	MacAfee et al. (2019)	Cross-sectional	2009	South	Statewide	Total = 3,042; Total who used illicit substances = 168	Vital records
62	Maeda et al. (2014)	Cross-sectional	2007	Coverage across ≥ 45 states	Nationally represented	60,994	Medical record
63	Mark et al. (2016)	Cohort	2009	South	Single, urban, university-based prenatal care clinic	116	Medical record
64	Mark et al. (2017)	Cross-sectional	2015	South	Outpatient Obstetrics and Gynecology Clinic at the University of Maryland Medical Center	Total = 306	Medical record
65	Massey et al. (2018)	Cohort	2003	Includes states across ≥ 2 regions	GUH: urban hospital-based obstetric clinic in the Northeastern United States; BAM BAM: obstetric clinics, health centers, and community postings in the Northeastern United States; EGDS: Mid-Atlantic, Southwestern, Midwestern, and Pacific Northwestern regions of the U.S	Total smokers in pooled sample = 608; Total = pregnant women in ECDS = 625	Medical record
66	May et al., (2020a, 2020b, 2020c)	Case–Control	2010	Midwest	24 public schools, 8 private schools	Total children with ARND = 10; total children with FASD = 31; total control children = 305	Identified mothers of children with FASD
67	May et al., (2020a, 2020b, 2020c)	Case–Control	2007	Includes states across ≥ 2 regions	City with population of about 60,000	Total with FASD = 35; total controls = 197	Identified mothers of children with FASD
68	May et al., (2020a, 2020b, 2020c)	Case–Control	2013	South	24 elementary schools	Total children with FASD = 47; total control children = 251	Identified mothers of children with FASD
69	May et al. (2015)	Case–Control	2007	Includes states across ≥ 2 regions	City of 59,000; 17 elementary schools	Total with PFAS/FAS = 26; total who had maternal interview data on exposure = 17	Identified mothers of living children
70	Mbah et al. (2012)	Cohort	1998	South	Population-based, state-level data	Total births = 1,698,223; total births with cocaine exposure = 5,026/1,698,223	Medical record
71	Metz et al. (2017)	Cohort	2006	Includes states across ≥ 2 regions	Geographically and racially diverse population	Total marijuana use = 48	Medical record
72	Metz et al. (2018)	Cross-sectional	2005	Coverage across ≥ 45 states	US representative, population-based sample	Total = 818; total with opioids only = 281/818; total with opioid poly-drug use = 241/818; total with other illegal drugs = 296/818	Self-report
73	Minnes et al. (2012)	Cohort	1994	Midwest	Midwestern urban hospital	Total = 321; total who used cocaine = 158	Delivered at hospital site
74	Nellhaus et al. (2019)	Cohort	2016	South	MAT programs, Cabell Huntington Hospital, Marshall Health and Valley Health System	Total pregnant women on MAT = 109 (with total of 110 neonates)	Delivered at hospital site
75	O’Connor et al. (2021)	Cohort	2013	Northeast	Rural medical facility	137	Medical record
76	O’Connor et al. (2017)	Cohort	2007	Northeast	Rural clinic in family medicine residency program	191	Medical record
77	Obisesan et al. (2020)	Cross-sectional	2016	Coverage across ≥ 45 states	N/A (BRFSS)	Total pregnant women = 7,434; total who used E-cigarettes = 2.2%	Self-report
78	Oga et al. (2018)	Cross-sectional	2017	South	Two obstetric practices in urban setting	494	Identified at prenatal care
79	Oh et al. (2017)	Cross-sectional	2005	Coverage across ≥ 45 states	N/A (NSDUH)	Total pregnant women = 8240; total adolescents aged 12–17 = 529/8240; total adults aged 18–44 = 7,711/8,240	Self-report
80	Oncken et al. (2020)	Clinical Trial	2012	Northeast	Unspecified	Total smokers = 129	Medical record
81	Patrick et al. (2015)	Cohort	2009	South	State-based (Medicare)	Total dyads = 112,029; total prescribed opioids = 31,354/112,029	Identified mothers of living children
82	Patterson et al. (2012)	Cohort	1999	Northeast	Urban, emergency room at the Hospital of the University of Pennsylvania	Total pregnant women = 1521; total who smoked = 338/1521	Other (specify): Self-reported and verified using medical record
83	Qato et al. (2020)	Cross-sectional	2006	Coverage across ≥ 45 states	N/A (NSDUH)	5.1% (4.5, 5.7) of 20,744,268	Self-report
84	Ram et al. (2016)	Cohort	2005	South	CAP, a comprehensive drug and alcohol treatment program for pregnant women located at the Johns Hopkins Bayview Medical Center	118	Part of study criteria
85	Richardson et al. (2019)	Cohort	1988	Northeast	Magee-Women’s Hospital prenatal clinic	Total who used cocaine during 1st trimester = 92	Medical record
86	Roberts and Nuru-Jeter (2012)	Cohort	2001	Pacific West	N/A	Total pregnant women = 8,449; total who used alcohol and/or drugs = 508	Medical record
87	Rollins et al. (2020)	Cross-sectional	2015	Northeast	Single, low-income, urban clinic and local obstetric offices and community centers	Total pregnant women = 1,365; total who used e-cigarettes = 54; total who used conventional cigarettes = 372	Pregnant at time of study
88	Roth et al. (2020)	Cohort	2017	South	Marshall University	Total women = 176; total neonates = 177 (one mother gave birth to twins)	Medical record
89	Roussotte et al. (2011)	Cohort	Not reported	Pacific West	Clinical setting	Total participants prenatally exposed to methamphetamine = 19	Identified mothers of living children
90	Salas-Wright et al. (2016)	Cross-sectional	2002	Coverage across ≥ 45 states	Representative sample of the US civilian	810	Self-report
91	Salzwedel et al. (2015)	Cohort		South	Local residential and outpatient treatment programs; one OB clinic was noted as low-income	88	Identified mothers of living children
92	Sanlorenzo et al. (2019)	Cohort	2009	South	Facilities accepting Medicaid	Total mother-infant dyads = 822; total non-pharmacologically treated for NAS = 224/822; total pharmacologically treated for NAS = 598/822	Medical record
93	Schauberger et al. (2014)	Cohort	2013	Midwest	Single obstetrical urban-based clinic in a rural area	200	Medical record
94	Serino Ma et al. (2018)	Cohort	2004	Northeast	Prenatal clinics in New York City	Total = 146; total people who used cocaine = 35; total people who used marijuana = 38; total people who used methadone = 24	Identified at prenatal care
95	Shannon et al. (2010)	Cohort	2005	South	Medical center	Rural = 85; urban = 29	Medical record
96	Shaw et al. (2014)	Cohort	1998	Pacific West	Four urban and five rural counties	Rural = 96; urban = 677	Part of study criteria
97	Shen et al. (2020)	Cross-sectional	2009	Northeast	Hospitals	Total = 1,463,302; total with OUD = 8324	Other (specify): Algorithm based on diagnoses, procedures, and DRG codes
98	Shmulewitz and Hasin (2019)	Cross-sectional	2002	Coverage across ≥ 45 states	National	13,488	Self-report
99	Shrestha et al. (2018)	Cohort	2013	West	UNM-affiliated clinics in Albuquerque metro area, including general obstetrics and midwifery clinics and a specialized prenatal care clinic for women with SUD, during one of their first prenatal care visits	Total who used medication for opioid use disorder = 26; total who used alcohol = 22; total who used both = 27	Medical record
100	Smith et al. (2015)	Cohort	2005	Northeast	137 obstetric practices and hospital-based clinics	Total = 2748; total subsample that used opioids = 165	Medical record
101	Stewart et al. (2013)	Cohort	2006	South	Inpatient hospital	Total women = 95; total with illicit drug use at delivery = 42; total on medication for opioid use disorder (methadone, buprenorphine) = 17	Medical record
102	Stitely et al. (2010)	Cross-sectional	2009	South	8 hospitals	Total = 759; total who used drugs = 146/759	Other (specify): Collected umbilical cord
103	Stroud et al. (2020)	Cohort	Not reported	Northeast	Sample was low-income from obstetrical offices, health centers, and community postings	Total pregnant women = 111; total who used tobacco = 45; total who used tobacco + marijuana = 24	Medical record
104	Tai et al. (2017)	Cohort	2001	Northeast	Outpatient substance abuse and mental health treatment center	Total 88; total alcohol use cohort = 57/88	Medical record
105	Terplan et al. (2009)	Cross-sectional	1994	Coverage across ≥ 45 states	Substance treatment faciltiies that receive federal funding	Year 1994: Total pregnant = 18,034; total who used methamphetamine = 1457. Year 2006: Total pregnant = 22,382; total who used methamphetamine = 5312	Medical record
106	Tith et al. (2018)	Cohort	2012	Pacific West	University of Washington Medical Center L&D unit	8	Medical record
107	Towers et al. (2019)	Cohort	2014	South	Clinic at University of Tennessee Medical Center	429	Delivered at hospital site
108	Towers et al. (2020)	Cohort	2017	South	University hospital; obstetrics OUD clinic at University of Tennessee Medical Center	Total who used buprenorphine/methadone = 109; total who used naltrexone = 121	Part of study criteria
109	Wachman et al. (2011)	Cohort	2003	Northeast	Boston Medical Center—hospital	Total mother-infant dyads = 273; total exposed to methadone = 251/273; total exposed to buprenorphine = 22/273	Medical record
110	Washio et al. (2017)	Cross-sectional	1992	Coverage across ≥ 45 states	Substance use disorder treatment facilities	71,960	Medical record
111	Washio et al. (2018)	Cross-sectional	1992	Coverage across ≥ 45 states	Substance use treatment facilities	Total who used substances = 489,796; total who used cannabis = 198,886/489,796	Medical record
112	Winhusen and Lewis (2017)	Clinical Trial	2007	Includes states across ≥ 2 regions	Four outpatient substance use disorder treatment facilities for pregnant women	Total smokers = 145	Medical record
113	Witt et al. (2015)	Cohort	2001	Coverage across ≥ 45 states	Family homes	Total who used cigarettes in the last 3 months of pregnancy: unweighted = 1050; weighted = 413,816	Identified mothers of living children
114	Wood et al. (2014)	Cohort	2008	Midwest	University of Iowa Hospitals and Clinics	2,497	Medical record
115	Wouldes et al. (2013)	Case–Control	2002	Includes states across ≥ 2 regions	7 hospitals	Total = 320; total who reported using methamphetamines = 127	Medical record
116	Wright et al. (2011)	Cohort	Not reported	Pacific West	Not stated	Total = 103; total who smoked = 37; total who used street drugs = 28	Medical record
117	Wright et al. (2015)	Cohort	2007	Pacific West	Path Clinic—prenatal care and social services for women with addictions	Total methamphetamine use during pregnancy = 144	Identified at prenatal care
118	de Wit et al. (2013)	Cross-sectional	2002	Coverage across ≥ 45 states	National	Total with alcohol use disorder diagnosis = 12,081	Medical record
119	van Gelder et al. (2010)	Case–Control	1997	Includes states across ≥ 2 regions	Population-based survey	Total women = 5871; total who used illicit drugs during pregnancy = 210/5,871	Medical record

^aSample size categories defined by each respective study

Study Characteristics

Tables 1 and 2 describe study characteristics of 119 included articles: 7 case–control studies, 7 clinical trials, 76 cohort studies, and 29 cross-sectional studies.

Table 2.

Summary characteristics of studies that reported polysubstance prevalence estimates among pregnant people, US, 2009-2020^a (N = 119)

Characteristic	No. of articles	%
Study design
Case–control	7	5.9
Clinical trial	7	5.9
Cohort	76	63.9
Cohort, prospective	43	36
Cohort, retrospective	32	26.9
Cohort, prospective and retrospective	1	0.8
Cross-sectional	29	24.4
Administrative dataset	8	6.7
Survey	13	10.9
Year of publication
2009–2012	33	27.7
2013–2016	28	23.5
2017–2020	58	48.7
First year of study time period
1980–1989	4	3.4
1990–1999	13	10.9
2000–2009	57	47.9
2010–2017	27	22.7
Not reported	18	15.1
Race/ethnicity (n = 100)b
American Indian or Alaska Native	19	19
Black or African American	81	81
Hispanic, LatinX, or Spanish origin	60	60
White	86	86
Other racec	67	67
Asian or Pacific Islander	19	28.4
Regiond
Coverage across 45 or more states	20	16.8
Includes states across two or more regionse	13	10.9
Midwest	13	10.9
Northeast	24	20.2
Pacific West	13	10.9
South	30	25.2
Westf	5	4.2
Not reported	1	0.8
Sampling method
Not population-based (Non-probability sampling)	87	73.1
Population-based	32	26.9
Study Objectiveg
Polysubstance use	46	38.7
Single substance use	73	61.3

^aSearch strategy was conducted up to June 3, 2020

^bRace/ethnicity categories are not mutually exclusive. The following article IDs did not provide any prevalence estimates by race and ethnicity: 46, 59, 71, 82, 84, 88, 89, 96, 98, 100, 108, 109, 112, 115, 120, 124, 139, 141, 142

^cData for Asian or Pacific Islander were identified from the data reported in the Other race category. Other race also included categories reported for multiracial, racial minority, Middle Eastern, non-white or Hispanic, non-Hispanic, not African American, and unspecified (which could have included the other listed race categories but were reported as Other or unspecified race/ethnicity by those studies)

^dStudy locations were grouped into regions based on the U.S. Census Bureau: https://www2.census.gov/geo/pdfs/maps-data/maps/reference/us_regdiv.pdf

^eThis category includes the following combinations of regions: Midwest, Northeast, Pacific West, South (n = 1); Midwest, Northeast, Pacific West, South, West (n = 1); Midwest, Northeast, South (n = 3); Midwest, Pacific West, South (n = 3); Midwest, South, West (n = 1); Northeast, Pacific West (n = 1); Northeast, South, West (n = 1); Midwest, West (n = 2)

^fWest does not include states categorized as Pacific West

^gStudy objective was determined based on a review of each article title

Articles published from 2017 to 2020 contributed to 49% of the total articles, though almost half (48%) of these included study time periods that started between 2000 and 2009. Among articles that provided any prevalence estimates by race and ethnicity (n = 100), American Indian/Alaska Native (AI/AN) and Hispanic were the two categories reported by the lowest number of articles. Study locations were most commonly reported in the following regions: South (28%), Northeast (22%), and West (15%). About 11% of total articles included states across two or more regions, and about 17% of total articles represented coverage across 45 or more states. Most articles for the base substances of “alcohol” (20%) and “opioids” (38%) were from studies that represented 45 or more states of the US, while most other base substance categories were based in the South (Table 5). About 27% of total articles used a population-based sampling method. Based on a review of the articles’ titles, about 39% of total articles had a research objective related to the use of multiple substances. Therefore, 61% of articles likely would not have been identified if the search strategy were limited to include only articles with research objectives specific to the use of multiple substances. Moreover, these studies contributed to 19 of 27 studies with more recent study time periods starting in 2010–2017 (data not shown). Of note, about 18% of all articles involved substance use treatment settings (data not shown).

Polysubstance Prevalence Estimates

Table 3 presents a summary of the number of prevalence estimates reported for each polysubstance combination, with base substance categories in columns and secondary substances in rows.

Table 3.

Number of prevalence estimates^a and article IDs^b for polysubstance combinations^c reported among studies^d of pregnant people, 2009–2020 (N = 119)

Secondary substancee	Base substancef
	Alcohol	Amphetamine/methamphetamine	Cannabis	Cocaine	Medications used for opioid use disorder	Opioid	Tobacco/nicotine	Otherg	No base substanceh
Total articles	15	9	16	10	22	16	26	22	43
Unspecified polysubstance use	2 Article IDs: [104; 110]		1 [94]		4 [4; 8; 45; 94]	8 [40; 42; 46; 72]	2 [29; 78]	8 [6; 90; 94; 101; 116]	2 [54; 93]
Alcohol		8 [11; 35; 47; 55; 60; 89; 115; 117]	13 [21; 22; 49; 50; 57; 63; 98; 102; 111; 119]	6 [6; 70; 73; 91; 98; 119]	9 [4; 10; 17; 39; 53; 95; 108; 102]	17 [24; 40; 44; 46; 51; 52; 62; 72; 97; 100; 102; 107]	20 [9; 17; 28; 34; 36; 43; 48; 58; 59; 78; 79; 87; 98; 103; 112; 113]	17 [9; 26; 31; 44; 48; 61; 72; 77; 87; 90; 98; 102; 103; 119]
Amphetamines/methamphetamines			2 [57; 111]		9 [4; 10; 39; 53; 74; 108]		5 [5; 78; 112; 116]	3 [19; 26; 90]
Barbiturates			1 [57]		2 [4; 74]
Benzodiazepines	1 [102]		2 [57; 102]		13 [4; 14; 39; 45; 53; 74; 75; 95; 102; 108; 109]	11 [40; 42; 44; 92; 102; 107]	3 [29; 78; 112]
Cannabis	9 [16; 30; 66; 67; 68; 69; 102; 110]	6 [11; 35; 47; 55; 115; 117]		4 [6; 73; 91; 94]	13 [4; 10; 39; 45; 53; 74; 75; 76; 94; 95; 102; 108]	13 [40; 46; 52; 72; 92; 97; 102; 107]	16 [5; 9; 12; 27; 28; 29; 33; 34; 48; 65; 78; 82; 112; 116]	7 [26; 48; 72; 77; 90; 102]
Cocaine	5 [16; 66; 67; 68; 110]	4 [11; 35; 115; 117]	3 [57; 94; 111]		18 [4; 14; 17; 37; 39; 45; 53; 56; 74; 94; 95; 108]	8 [40; 46; 92; 97]	8 [5; 17; 28; 29; 78; 82; 112; 116]
Opioid	4 [16; 102; 110]	3 [35; 117]	5 [57; 94; 102; 111]	4 [6; 91; 94]			7 [5; 17; 29; 78; 80; 112]	5 [19; 26; 102]
Tobacco/nicotine	16 [2; 16; 24; 30; 32; 58; 66; 67; 68; 79; 104; 118]	5 [35; 47; 55; 115; 117]	14 [1; 12; 15; 21; 22; 38; 49; 50; 57; 63; 64; 71; 119]	5 [6; 71; 73; 91; 119]	20 [4; 10; 13; 14; 17; 37; 39; 45; 53; 56; 74; 75; 84; 95; 99; 108; 109]	19 [24; 42; 44; 46; 51; 52; 62; 72; 81; 88; 92; 97; 100; 107]		18 [7; 9; 31; 44; 61; 72; 77; 86; 87; 93; 96; 101; 116; 119]
Other	6 [16; 68; 104; 110]	3 [35]	4 [21; 49; 71; 111]	4 [3; 6; 23; 85]	18 [4; 8; 10; 14; 39; 53; 56; 74; 75; 76; 84; 101; 108; 109]	17 [46; 51; 62; 92; 97; 100; 107]	7 [5; 9; 28; 59; 77; 112]	6 [26; 90]
Alcohol + tobacco			1 [119]		1 [17]				4 [25; 58; 83; 119]
Alcohol + cannabis									6 [66; 68; 83; 90]
Alcohol + other							1 [5]		2 [41; 86]
Amphetamine + cannabis									1 [114]
Cocaine + cannabis									1 [114]
Cocaine + tobacco					1 [17]				1 [83]
Cannabis + opioid									2 [54; 114]
Cannabis + tobacco									6 [12; 15; 19; 20; 65; 83]
Cannabis + other					1 [76]		1 [5]		2 [18; 114]
Opioid + tobacco					1 [17]				1 [83]
Opioid + other					2 [106]				4 [83; 114]
Tobacco + other	1 [104]								3 [77; 104; 114]
Other combination									5 [83; 114]

Bolded values denote combinations with the highest number of estimates reported for a given column (i.e., base substance category)

Other entries denote combinations that do not contain the highest number of estimates reported for a given column (i.e., base substance category) or, if blank, data not reported

^aMultiple estimates may have been reported by a single article. For articles reporting multiple estimates with different subgroups, overall estimates were included in total count of estimates if available. If overall estimates were not reported, reported estimates could have been provided for any of the following subgroups: ‘Other’ substance or polysubstance subgroups (n = 12); specific drug type (opioid n = 7; medications for opioid use disorder (MOUD) n = 2; amphetamine n = 2; cocaine n = 1); substance use pattern (n = 5); exposure definition (e.g., substance use disorder (SUD) diagnosis, toxicology results or specimen type, unrelated environmental exposure, mental health condition diagnosis) (n = 5); study-specific subgroups (n = 2); rural vs urban area (n = 2); region (n = 1); age (n = 1); race (n = 1); trimester (n = 1); neonatal abstinence syndrome (NAS) treatment subgroups (n = 1)

^bArticle IDs correspond to the list of articles described in Table 1

^cThe following rows for select polysubstance combinations were omitted from the table for brevity since there were no prevalence estimates reported by any study included in the review: Alcohol + Opioid; Alcohol + Cocaine; Alcohol + Benzodiazepines; Alcohol + Barbiturates; Tobacco + Benzodiazepines; Cannabis + Benzodiazepines; Cannabis + Barbiturates; Opioid + Cocaine; Opioid + Benzodiazepines; Opioid + Barbiturates; Cocaine + Benzodiazepines; Cocaine + Barbiturates; Cocaine + Other

^dSuperscripts refer to the Article IDs listed in Table 5

^eThe secondary substance was the other substance reported by the study population and served as the numerator of polysubstance prevalence calculations

^fThe base substance was the primary substance reported that defined the study population and served as the denominator for polysubstance prevalence calculations

^gThe “Other” base substance category represents studies assessing polysubstance use with unspecified substances or substances other than the existing base substance categories. This category included the use of any unspecified substance (e.g., “other illicit drug use”; “any drug use”); hallucinogens (unspecified); stimulants (unspecified); tranquilizers or sedatives (unspecified); gabapentin; phencyclidine; smokeless tobacco or e-cigarettes; and any diagnosis of “other” substance use disorder

^hData from the “No base substance” base substance category included estimates from studies that reported polysubstance use among pregnant people without respect to a base substance

Figure 2 shows bubble plots of prevalence estimates for each base substance category with secondary substances represented in the X axis.

Fig. 2.

Bubble plots¹ representing the distribution of polysubstance estimates reported among studies of pregnant people, by base substance^2,3 U.S. 2009–2020 (N = 119)

“Alcohol” and “tobacco/nicotine” were the most commonly reported secondary substances (Table 3). In the “no base substance” category, “alcohol + cannabis” and “cannabis + tobacco” had the greatest number of contributing prevalence estimates (Table 3). Moreover, regardless of the base or secondary substance, prevalence estimates for polysubstance combinations with “alcohol” (range 1–99% for population-based studies; 1–84% for nonpopulation-based studies), “cannabis” (range 3–60% for population-based studies; 7–95% for nonpopulation-based studies), and “tobacco/nicotine” (range 2–82% for population-based studies; 1–95% for nonpopulation-based studies) had the highest prevalence estimates but with wide ranges (Fig. 2a, b, and d). Additionally, one population-based estimate was reported for the three-drug combination with “alcohol + tobacco” in the “cannabis” base substance category (48%) (Fig. 2b). Prevalence estimates for polysubstance combinations with “alcohol,” “cannabis,” and “tobacco/nicotine” as secondary substances were often population-based, with exception of the base substance categories of “amphetamine/methamphetamine” and “MOUD,” which did not include any population-based estimates (Fig. 2).

Prevalence estimates for a given combination of substances can vary depending on the base substance for a particular study (Fig. 2). For example, though “alcohol,” “cannabis,” and “tobacco/nicotine” had the three highest prevalence estimates for the base substance of non-MOUD “opioids,” non-MOUD “opioids” as a secondary substance was not among the highest reported estimates across “alcohol,” “cannabis,” and “tobacco/nicotine” base substance categories. Also, in the “alcohol” base substance category, “opioids” were not among the highest prevalence estimates and had a relatively narrower range of prevalence estimates (range 2–7%) (Fig. 2a). In contrast, “alcohol” had the highest prevalence estimate as a secondary substance for the “opioid” base substance category, but with a much wider range of prevalence estimates (range 1–99%) (Fig. 2c).

Substance use during pregnancy was ascertained using various approaches across studies and by base substance (Table 6). The base substance categories that most often used self-report as a single method to ascertain exposure were “alcohol” (55%), “tobacco/nicotine” (51%),” cannabis” (43%), “cocaine” (35%), and “methamphetamine” (35%), with other categories more often relying on medical records or biochemical validation. The base substance categories that most often used biochemical validation and self-report for two methods of ascertaining exposure were “methamphetamine” (19%), “amphetamine” (15%), “cocaine” (15%), and “cannabis” (13%). The base substance categories that most commonly used three methods for exposure ascertainment (i.e., biochemical validation, medical record, and self-report) were “methamphetamine” (15.4%), “amphetamine” (10%), “cocaine” (9.3%), and “cannabis” (9.3%).

Discussion

Prevalence estimates for polysubstance combinations that included alcohol, cannabis, or tobacco/nicotine were the highest among studies of pregnant people. Possible explanations for common reporting of alcohol, cannabis, and tobacco use may include the relative ease of attainment of licit substances (i.e., alcohol and tobacco), and the increase in cannabis legalization across the nation (Encyclopaedia Britannica, 2020). The routine practice of documenting the use of licit substances, such as alcohol (Centers for Disease Control & Prevention, 2014), through substance use screening during prenatal visits in the US might also contribute to the high number of reports in the literature by self-report and/or medical record compared to other substances included in this review.

These findings are consistent with recent results from NSDUH, which estimated that about 38% of pregnant individuals who reported current drinking also reported current use of at least one other substance, primarily tobacco or cannabis (England et al., 2020). Another recently published study of NIS data found that among pregnant women at delivery who were diagnosed with cocaine, amphetamine, alcohol, or opioid use disorders, the most common secondary substances used were tobacco and cannabis (Jarlenski & Krans, 2021). Overall, this review reflects similar trends of previous studies and included data from studies using various methods of exposure ascertainment, such as diagnoses at in medical records and self-report (Table 6).

Polysubstance use has had an increasing role in estimates of overdose mortality (Ciccarone, 2021). From 2000 to 2020, deaths involving methamphetamine combined with opioids were most commonly among the AI/AN racial group, and death rates in 2020 impacted multiple regions across the US with state-level variation (The National Institute for Health Care Management (NIHCM) Foundation, 2022). The current review demonstrated that there are a limited number of articles reporting estimates for racial groups and regions that may be disproportionately impacted by polysubstance use and overdose. Given that overdose is one of the leading causes of pregnancy-associated death in the US (Campbell et al., 2021), there is a need for more data on how polysubstance use affects morbidity and mortality in the pregnant population.

This review could not be used to assess some patterns in polysubstance use. Studies often assessed “any” exposure during pregnancy without detailed information on timing, dose, or frequency, and drug use was often reported by drug class (e.g., “stimulants” rather than amphetamine or methamphetamine). Moreover, race and ethnicity were not uniformly captured across studies. These constraints limited the authors” ability (1) to assess the trajectory of polysubstance use over the course of pregnancy, (2) to differentiate between the use of more than one substance simultaneously vs concurrently (within the same time period but not simultaneously) during pregnancy, (3) to assess polysubstance use more granularly for use of multiple substance within the same drug class, and (4) to analyze prevalence data on polysubstance use during pregnancy by clearly defined categories for race and ethnicity. To better ascertain patterns of polysubstance use and prevalence estimates for specific polysubstance groups, researchers are encouraged in future studies to clearly define the base substance categories and periods of exposure during pregnancy and to delineate polysubstance use by race/ethnicity when possible.

Of note, most studies included in this systematic review were based on non-probability sampling and cannot be generalized to the broader population of pregnant people (Non-Probability Sampling, 2013). Approximately 73% of studies used non-probability sampling, which resulted in a heterogenous group of studies with prevalence estimates among subgroups that differed widely by base substance and inclusion and exclusion criteria. Methods to ascertain exposure also varied, with most studies relying on a single measure of exposure, which typically results in underestimating prevalence. For example, measuring substance use by self-report would likely underestimate the true prevalence if pregnant people were reluctant to disclose use due to concern about stigma or legal implications; similarly, biochemical validation as a single method could result in underestimations of prevalence (e.g., urine testing may only detect drug use within four days of specimen collection). Reliance on diagnosis codes in medical records to define exposure may result in the over-representation of individuals with more advanced disease. These study limitations hinder the ability to assess the true prevalence for specific combinations of polysubstance use during pregnancy in the US. Only 1–19% of studies used more than one method to establish exposure (Table 6). Combining methods, such as self-report and serial biochemical testing, can result in greater sensitivity and more accurate estimates of overall disease burden.

A meta-analysis was not performed since the purpose of the review was not to provide a single summary estimate for the US, but rather to summarize polysubstance prevalence estimates across various geographic locations and polysubstance subgroups. Additionally, many differences existed between studies (i.e., study design, location, setting, time period, sampling and participant selection, method for determining pregnancy, and exposure), and not all studies reported confidence intervals or standard errors. Future reviews that aim to provide a single summary estimate for the prevalence of polysubstance use during pregnancy in the US are encouraged to carefully review the current literature as models and approaches to meta-analyses of prevalence are continuing to evolve (Munn et al., 2015).

Strengths and Limitations

This review on the prevalence of polysubstance use during pregnancy included many studies with research objectives specific to both single-substance and polysubstance use, and multiple prevalence estimates could have been derived from the same article or data source. Although the methods of this review were labor intensive, the comprehensive approach led to a more complete ascertainment of relevant articles and polysubstance prevalence estimates. Another strength of this review’s methodology relates to the exclusion criteria for studies with overlapping study populations and study years, which prevented overreporting of duplicative results from different articles using the same datasets. Overall, this review strengthens the existing evidence on national prevalence estimates of polysubstance use during pregnancy by describing a heterogenous compilation of studies with varying study catchment areas and time points, methods of exposure ascertainment, and definitions of exposure and timing of exposure during pregnancy. Additionally, as shown in Fig. 2, this review revealed gaps in the literature by demonstrating the variations in prevalence estimates reported due to differences in sampling methods and sample sizes, which can help to identify where more or better data are needed.

This review was subject to at least five limitations. First, this review was only based on substances included in the 2019 NSDUH and therefore did not include other emerging substances (Musial et al., 2022; Substance Abuse and Mental Health Services Administration, 2020a). However, based on recent studies (England et al., 2020; Qato et al., 2020), substances included in the search criteria likely captured the most common substances used in the past decade. Second, the authors did not include articles published beyond June 2020 due to the low likelihood of the addition of one year of data leading to a major change in conclusions (Bashir et al., 2018). Third, almost a fifth of articles in the current review involved substance use treatment settings, which may reflect higher substance use prevalence estimates than the general pregnant population, though results were not able to be reported by study setting type. Fourth, an interrater reliability (IRR) test was not conducted for this review. However, details including the number of reviewers at each phase and methods to resolve disagreements were fully described. To facilitate reliability and consistency in coding decisions, future systematic reviews should consider conducting IRR testing at appropriate points of the review process (Belur et al., 2021). Fifth, this review was unable to determine the intention of use for prescribed medications. However, describing the prevalence of specific polysubstance combinations used during pregnancy is still essential to inform guidance on the most common polysubstance groups of concern.

Conclusions

Polysubstance use during pregnancy is common, especially in combinations including alcohol, cannabis, and/or tobacco/nicotine. Current expert opinions and recommendations for the management of polysubstance use during pregnancy include screening for use of multiple substances, provision of patient education and behavioral counseling using person-centered language and current medical terminology, and treatment of each substance use disorder simultaneously, when indicated (Smid & Terplan, 2022). Given the steady or increased rates of overdose mortality involving CNS depressants as well as psychostimulants (e.g., methamphetamine and cocaine) (Bruzelius & Martins, 2022; The National Institute for Health Care Management (NIHCM) Foundation, 2022), it is important to understand the most common substance combinations and patterns in timing of exposure that may occur during pregnancy to inform future work on prevention and treatment. Future research could help researchers and clinicians better describe (1) factors that influence polysubstance use patterns during pregnancy; (2) associations between specific combinations of polysubstance use and adverse maternal, infant, and child outcomes; and (3) what timepoints during pregnancy represent periods of vulnerability for maternal and infant health outcomes and should be prioritized in areas of clinical management and research.

Significance.

What is already known on this subject?

Polysubstance use has potential to increase the risk or severity of adverse maternal and infant outcomes compared with single-substance use. Current approaches to estimate prevalence for pregnant individuals are limited to definitions using self-report or what is available in medical records, which often are not comprehensive.

What this study adds?

Polysubstance combinations with alcohol, marijuana, and/or tobacco/nicotine were common during pregnancy. Among included articles, there was variation in study design, time period, region, sampling and participant selection, substances assessed, and method of exposure ascertainment. Future research to assess polysubstance use during pregnancy could help better describe patterns and ultimately help mitigate its effects on maternal and infant outcomes.

Appendix

See Tables 4, 5 and 6.

Table 4.

Search strategy for prenatal polysubstance use among databases from conception to June 3, 2020

Database	Strategy	Total number of citations
All databases		8736
MEDLINE (Ovid) 1946-	(Exp substance-related disorders/ep OR (Polysubstance OR poly-substance OR polydrug OR poly-drug OR polyconsumption OR poly-consumption OR (multiple ADJ2 drug*) OR (multiple ADJ2 substance*) OR co-abuse OR ((substance abuse* OR substance use* OR drug abuse* OR drug use*) ADJ3 (multiple OR co-use OR concomitant)) OR (smoking ADJ2 drinking) OR (tobacco ADJ5 alcohol) OR (marijuana ADJ5 alcohol) OR (tobacco ADJ5 marijuana) OR (cannabis ADJ5 alcohol) OR (tobacco ADJ5 cannabis) OR (cigarette* ADJ5 alcohol) OR (cigarette* ADJ5 marijuana) OR (cigarette* ADJ5 cannabis) OR (alcohol ADJ5 substance*) OR (alcohol ADJ5 drug*) OR (drinking ADJ5 substance*) OR (drinking ADJ5 drug*) OR (opioid* ADJ5 smoking) OR (opioid* ADJ5 drinking) OR (opioid* ADJ5 alcoholyt) OR (opioid* ADJ5 tobacco) OR (opioid* ADJ5 cigarettes) OR (opioid* ADJ5 marijuana) OR (opioid* ADJ5 cannabis)).ti,ab OR ( ((Polysubstance OR polysubstance OR polydrug OR poly-drug OR polyconsumption OR poly-consumption OR (multiple ADJ2 drug*) OR (multiple ADJ2 substance*) OR co-abuse OR ((substance abuse* OR substance use* OR drug abuse* OR drug use*) ADJ3 (multiple OR co-use OR concomitant))) AND (stimulants OR Ritalin OR Adderall OR crystal meth OR methamphetamine OR amphetamine* OR cocaine OR crack OR hallucinogen* OR psychedelic* OR psychotomimetic* OR LSD OR ketamine OR inhalant* OR sniff* OR hash OR hashish)) OR ((drinking OR smoking OR alcohol OR tobacco OR cigarette*) AND (stimulants OR Ritalin OR Adderall OR crystal meth OR methamphetamine OR amphetamine* OR cocaine OR crack OR hallucinogen* OR psychedelic* OR psychotomimetic* OR LSD OR ketamine OR inhalant* OR sniff* OR hash OR hashish)) ) AND (pregnancy/AND (risk behavior OR risk taking OR health behavior)) OR Pregnant OR during pregnancy OR in pregnancy Limit English; (202006* OR 202007* OR 202008* OR 202009* OR 202010* OR 202011* OR 202012* OR 2021*).dt	3149
Embase (Ovid) 1988-	Exp drug dependence/ep OR (Polysubstance OR poly-substance OR polydrug OR poly-drug OR polyconsumption OR poly-consumption OR (multiple ADJ2 drug*) OR (multiple ADJ2 substance*) OR ((substance abuse* OR substance use* OR drug abuse* OR drug use*) ADJ3 (multiple OR co-use OR concomitant OR co-abuse)) OR (smoking ADJ2 drinking) OR (tobacco ADJ5 alcohol) OR (marijuana ADJ5 alcohol) OR (tobacco ADJ5 marijuana) OR (cannabis ADJ5 alcohol) OR (tobacco ADJ5 cannabis) OR (cigarette* ADJ5 alcohol) OR (cigarette* ADJ5 marijuana) OR (cigarette* ADJ5 cannabis) OR (alcohol ADJ5 substance*) OR (alcohol ADJ5 drug*) OR (drinking ADJ5 substance*) OR (drinking ADJ5 drug*) OR (opioid* ADJ5 smoking) OR (opioid* ADJ5 drinking) OR (opioid* ADJ5 alcohol) OR (opioid* ADJ5 tobacco) OR (opioid* ADJ5 cigarettes) OR (opioid* ADJ5 marijuana) OR (opioid* ADJ5 cannabis)).ti,ab OR ( ((Polysubstance OR poly-substance OR polydrug OR poly-drug OR polyconsumption OR poly-consumption OR (multiple ADJ2 drug*) OR (multiple ADJ2 substance*) OR co-abuse OR ((substance abuse* OR substance use* OR drug abuse* OR drug use*) ADJ3 (multiple OR co-use OR concomitant))) AND (stimulants OR Ritalin OR Adderall OR crystal meth OR methamphetamine OR amphetamine* OR cocaine OR crack OR hallucinogen* OR psychedelic* OR psychotomimetic* OR LSD OR ketamine OR inhalant* OR sniff* OR hash OR hashish)) OR ((drinking OR smoking OR alcohol OR tobacco OR cigarette*) AND (stimulants OR Ritalin OR Adderall OR crystal meth OR methamphetamine OR amphetamine* OR cocaine OR crack OR hallucinogen* OR psychedelic* OR psychotomimetic* OR LSD OR ketamine OR inhalant* OR sniff* OR hash OR hashish)) ) AND (pregnancy/AND (risk behavior OR risk taking OR health behavior)) OR Pregnant OR during pregnancy OR in pregnancy Limit English; not pubmed/medline; (202006* OR 202007* OR 202008* OR 202009* OR 202010* OR 202011* OR 202012* OR 2021*).dc	2905
PsycINFO (Ovid) 1987-	Exp drug abuse/OR (Polysubstance OR poly-substance OR polydrug OR poly-drug OR polyconsumption OR polyconsumption OR (multiple ADJ2 drug*) OR (multiple ADJ2 substance*) OR ((substance abuse* OR substance use* OR drug abuse* OR drug use*) ADJ3 (multiple OR co-use OR concomitant)) OR (smoking ADJ2 drinking) OR (tobacco ADJ5 alcohol) OR (marijuana ADJ5 alcohol) OR (tobacco ADJ5 marijuana) OR (cannabis ADJ5 alcohol) OR (tobacco ADJ5 cannabis) OR (cigarette* ADJ5 alcohol) OR (cigarette* ADJ5 marijuana) OR (cigarette* ADJ5 cannabis) OR (alcohol ADJ5 substance*) OR (alcohol ADJ5 drug*) OR (drinking ADJ5 substance*) OR (drinking ADJ5 drug*) OR (opioid* ADJ5 smoking) OR (opioid* ADJ5 drinking) OR (opioid* ADJ5 alcohol) OR (opioid* ADJ5 tobacco) OR (opioid* ADJ5 cigarettes) OR (opioid* ADJ5 marijuana) OR (opioid* ADJ5 cannabis)).ti,ab OR ( ((Polysubstance OR poly-substance OR polydrug OR poly-drug OR polyconsumption OR poly-consumption OR (multiple ADJ2 drug*) OR (multiple ADJ2 substance*) OR co-abuse OR ((substance abuse* OR substance use* OR drug abuse* OR drug use*) ADJ3 (multiple OR co-use OR concomitant))) AND (stimulants OR Ritalin OR Adderall OR crystal meth OR methamphetamine OR amphetamine* OR cocaine OR crack OR hallucinogen* OR psychedelic* OR psychotomimetic* OR LSD OR ketamine OR inhalant* OR sniff* OR hash OR hashish)) OR ((drinking OR smoking OR alcohol OR tobacco OR cigarette*) AND (stimulants OR Ritalin OR Adderall OR crystal meth OR methamphetamine OR amphetamine* OR cocaine OR crack OR hallucinogen* OR psychedelic* OR psychotomimetic* OR LSD OR ketamine OR inhalant* OR sniff* OR hash OR hashish)) ) AND (pregnancy/AND (risk behavior OR risk taking OR health behavior)) OR Pregnant OR during pregnancy OR in pregnancy Limit English; (202006* OR 202007* OR 202008* OR 202009* OR 202010* OR 202011* OR 202012* OR 2021*).up	2150
CINAHL (EBSCO)	(MH “substance-related disorders + ”/EP) OR (Polysubstance OR poly-substance OR polydrug OR poly-drug OR polyconsumption OR poly-consumption OR (multiple N2 drug*) OR (multiple N2 substance*) OR ((“substance abuse*” OR “substance use*” OR “drug abuse*” OR “drug use*”) N3 (multiple OR co-use OR concomitant)) OR (smoking N2 drinking) OR (tobacco N5 alcohol) OR (marijuana N5 alcohol) OR (tobacco N5 marijuana) OR (cannabis N5 alcohol) OR (tobacco N5 cannabis) OR (cigarette* N5 alcohol) OR (cigarette* N5 marijuana) OR (cigarette* N5 cannabis) OR (alcohol N5 substance*) OR (alcohol N5 drug*) OR (drinking N5 substance*) OR (drinking N5 drug*) OR (opioid* N5 smoking) OR (opioid* N5 drinking) OR (opioid* N5 alcohol) OR (opioid* N5 tobacco) OR (opioid* N5 cigarettes) OR (opioid* N5 marijuana) OR (opioid* N5 cannabis)) OR ( ((Polysubstance OR poly-substance OR polydrug OR poly-drug OR polyconsumption OR polyconsumption OR (multiple N2 drug*) OR (multiple N2 substance*) OR co-abuse OR ((substance abuse* OR substance use* OR drug abuse* OR drug use*) N3 (multiple OR co-use OR concomitant))) AND (stimulants OR Ritalin OR Adderall OR crystal meth OR methamphetamine OR amphetamine* OR cocaine OR crack OR hallucinogen* OR psychedelic* OR psychotomimetic* OR LSD OR ketamine OR inhalant* OR sniff* OR hash OR hashish)) OR ((drinking OR smoking OR alcohol OR tobacco OR cigarette*) AND (stimulants OR Ritalin OR Adderall OR crystal meth OR methamphetamine OR amphetamine* OR cocaine OR crack OR hallucinogen* OR psychedelic* OR psychotomimetic* OR LSD OR ketamine OR inhalant* OR sniff* OR hash OR hashish)) ) AND ((MH pregnancy) AND (“risk behavior” OR “risk taking” OR “health behavior”)) OR Pregnant OR “during pregnancy” OR “in pregnancy” Limit English; exclude Medline records	532

CINAHL Cumulative Index of Nursing and Allied Health

Table 5.

Number of prenatal polysubstance use-related articles reporting information on region^a, by base substance category, 2009–2020

Region	Total	Alcohol (n = 15)	Cannabis (n = 17)	Medications used for opioid use disorder (n = 23)	Opioid (n = 16)	Tobacco/nicotine (n = 26)	Amphetamine/methamphetamine (n = 8)	Cocaine (n = 10)	Other (n = 24)	No base substanceb (n = 20)
Coverage across 45 or more states	20	4 (26.7%)	3 (17.7%)		8 (50%)	4 (15.4%)		1 (10%)	5 (20.1%)	4 (20%)
Includes states across two or more regionsc	13	3 (20%)	3 (17.7%)	1 (4.4%)		3 (11.5%)	3 (37.5%)	2 (20%)	2 (8.3%)	3 (15%)
Midwest	13	2 (13.3%)	2 (11.8%)	1 (4.4%)	1 (6.3%)	3 (11.5%)		1 (10%)	1 (4.2%)	6 (30%)
Northeast	24	2 (13.3%)	2 (11.8%)	6 (26.1%)	2 (12.5%)	8 (30.8%)		4 (40%)	5 (20.1%)	2 (10%)
South	30	3 (20%)	5 (29.4%)	11 (47.8%)	5 (31.3%)	4 (15.4%)	1 (12.5%)	2 (20%)	6 (25%)	3 (15%)
Westd	5		1 (5.9%)	3 (13%)			1 (12.5%)
Pacific West	13	1 (6.7%)	1 (5.9%)	1 (4.4%)		3 (11.5%)	3 (37.5%)		5 (20.1%)	2 (10%)
Not reported	1					1 (3.9%)

^aStudy locations were grouped into regions based on the U.S. Census Bureau: https://www2.census.gov/geo/pdfs/maps-data/maps/reference/us_regdiv.pdf

^bThe “no base substance” category included estimates from studies that reported polysubstance use among pregnant people without respect to a base substance

^cThis category includes the following combinations of regions: Midwest, Northeast, Pacific West, South (n = 1); Midwest, Northeast, Pacific West, South, West (n = 1); Midwest, Northeast, South (n=3); Midwest, Pacific West, South (n = 3); Midwest, South, West (n=1); Northeast, Pacific West (n = 1); Northeast, South, West (n = 1); Midwest, West (n = 2)

^dWest does not include states categorized as Pacific West

Table 6.

Number of prenatal polysubstance use-related articles describing the exposure ascertainment method, by base substance category^a 2009–2020

Exposure ascertainment method	Base substance
	Alcohol (n = 80)		Cannabis (n = 75)		MOUD (n = 24)		Opioids (n = 53)		Tobacco/nicotine (n = 96)		Amphetamine (n = 20)		Methamphetamineb (n = 26)		Cocaine (n = 54)		Otherc (n = 57)
	N	%	N	%	N	%	N	%	N	%	N	%	N	%	N	%	N	%
Single method
Biochemical validationd	4	5	16	21.3	4	16.7	14	26.4	4	4.2	9	45	3	11.5	14	25.9	15	26.3
Birth certificate	1	1.3							2	2.1
Medical recorde	11	13.8	3	4	7	29.2	10	18.9	14	14.6	1	5	3	11.5	4	7.4	7	12.3
Part of study criteria/protocol					6	25
Self-report	44	55	32	42.7	1	4.2	14	26.4	49	51	2	10	9	34.6	19	35.2	24	42.1
Two methods
Biochemical validation + medical record	2	2.5	4	5.3	3	12.5	4	7.6	1	1	3	15	2	7.7	4	7.4	4	7
Birth certificate + medical record									2	2.1
Biochemical validation + part of study criteria/protocol					1	4.2	1	1.9
Biochemical validation + self-report	5	6.3	10	13.3	2	8.3	6	11.3	11	11.5	3	15	5	19.2	8	14.8	2	3.5
Medical record + self-report	5	6.3	1	1.3					4	4.2
Self-report + other (physical features of FASD)	1	1.3
Three methods
Biochemical validation + medical record + self-report	3	3.8	7	9.3			4	7.6	4	4.2	2	10	4	15.4	5	9.3	4	7
Not reported	4	5	2	2.7					5	5.2							1	1.8

FASD Fetal alcohol spectrum disorders, MOUD Medications for opioid use disorder

^aThe base substance was the primary substance reported that defined the study population and served as the denominator for polysubstance prevalence calculations

^bFor the purposes of this table, methamphetamine and amphetamine were assessed separately

^cThe “other” base substance category represents studies assessing polysubstance use with unspecified substances or substances other than the existing base substance categories. This category included the use of any unspecified substance (e.g., “other illicit drug use”; “any drug use”); hallucinogens (unspecified); stimulants (unspecified); tranquilizers or sedatives (unspecified); gabapentin; phencyclidine; smokeless tobacco or e-cigarettes; and any diagnosis of “other” substance use disorder

^dBiochemical validation may include any toxicology screening or testing of mother or newborn

^eMedical record may include the review of the medical record for relevant diagnoses (e.g., substance use disorders, infant or childhood outcomes), prescriptions, or documented substance use

Data Availability

Not applicable.