Substance-related diagnosis type predicts the likelihood and co-occurrence of preterm and cesarean delivery

Abstract 
 Objective:
 This article aimed to evaluate whether a substance-related diagnosis (SRD; i.e., alcohol, opioids, cannabis, stimulants, nicotine) predicts the likelihood and co-occurrence of preterm (20–37 weeks’ gestation) and cesarean delivery. 
 Methods:
 This study reviewed electronic health record data on women (aged 18–44 years) who delivered a single live or stillbirth at ≥ 20 weeks of gestation from 2012 to 2019. Women with and without an SRD were matched on key demographic characteristics at a 1:1 ratio. Adjusting for covariates, odds ratios and 95% confidence intervals were calculated. 
 Results:
 Of the 19,346 deliveries, a matched cohort of 2,158 deliveries was identified. Of these, 1,079 (50%) had an SRD, 280 (13%) had a preterm delivery, 833 (39%) had a cesarean delivery, and 166 (8%) had a co-occurring preterm and cesarean delivery. An SRD was significantly associated with preterm and cesarean delivery (AOR = 1.84 [95% CI, 1.41–2.39], p-value= <0.0001; AOR = 1.51 [95% CI, 1.23–1.85], p-value= <0.0001). An alcohol-related diagnosis (AOR = 1.82 [95% CI, 1.01–3.28], p-value= 0.0471), opioid-related diagnosis (AOR = 1.94 [95% CI, 1.26–2.98], p-value= 0.0027), stimulant-related diagnosis (AOR = 1.65 [95% CI, 1.11–2.45], p-value= 0.0142), and nicotine-related diagnosis (AOR = 1.54 [95% CI, 1.05–2.26], p-value= 0.0278) were associated with co-occurring preterm and cesarean delivery. 
 Conclusions:
 Pregnant women with an SRD experienced disproportionally higher odds of preterm and cesarean delivery compared to pregnant women without an SRD. Substance-type predicts the type of delivery outcome. An SRD in pregnant women should be identified early to reduce potential harm through intervention and treatment.


Introduction
Women with a substance-related diagnosis (SRD; i.e., use, misuse, or dependence on substances) during pregnancy may be experiencing disproportionately higher odds of preterm (20-37 weeks' gestation) and cesarean delivery compared to women without an SRD during pregnancy. 1 Studies have shown that increases in preterm and cesarean delivery in the United States may be due to the changes in the population of women giving birth. Increases in maternal age, 2 preexisting chronic health conditions (e.g., hypertension), 3,4 and obesity 5 have all been identified as common predictors of preterm and cesarean delivery. Studies that have examined the relationship between SRDs and preterm and cesarean delivery found mixed results and may have been limited to small sample sizes and failure to adequately control for confounding variables such as older age and preexisting chronic health conditions. 6,7 Previous research has observed a significant relationship between substance use and preterm delivery. [8][9][10] However, the type of substance and its association with preterm delivery vary. Much less is known about the relationships between prenatal substance use, cesarean delivery, and co-occurring preterm and cesarean delivery. Increasing rates of maternal substance use, 11 the ongoing opioid epidemic, and the relatively recent legalization of cannabis in California 12 may also impact rates of preterm and cesarean delivery in this vulnerable population. As such, updated prevalence and correlates of preterm and cesarean delivery in pregnant women with an SRD is needed.
To address the current gap in the literature, a retrospective cross-sectional study of pregnant women was conducted to evaluate whether an SRD predicts the likelihood and co-occurrence of preterm and cesarean delivery in a healthcare system that provides tertiary care and is a referral system for other providers in the community in Southern California. It was hypothesized that pregnant women with an SRD would have higher prevalence of preterm and cesarean delivery compared to pregnant women without an SRD.

Study participants and procedures
De-identified electronic health record (EHR) data on any woman (aged 18-44 years) who delivered a single live or stillbirth at ≥ 20 weeks of gestation were collected from a large health system in Southern California from January 1, 2012 through August 31, 2019 (7.8 years of available data). This healthcare system averages about 3000 deliveries per year. To protect adolescents under the age of 18 who are considered an especially vulnerable population, the sample was restricted to maternal aged ≥ 18 so those < 18 could not be identified due to small sample sizes.
Women with an International Classification of Diseases, 10 th edition (ICD-10) code for a single live or stillbirth at ≥ 20 weeks of gestation were used for analysis (Table A in the supplemental material). 13 Deliveries of multiple gestation were omitted due to potential differences in delivery outcomes related to more than one gestation. Data were collected from the antepartum (conception to ≤ 42 weeks) and intrapartum (labor and delivery) periods. When an individual record had more than one delivery carried to a gestational age of ≥ 20 weeks, each patient identification number (ID) and its unique delivery date represented one subject. The number of previous pregnancies for each delivery by ID number was identified by delivery codes that appeared before the most recent delivery in the dataset.
ICD-10 codes for SRDs and other mental illness diagnoses correspond with the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5), which provides a more detailed description of each diagnosis ( Table A in the  supplemental material). 13,14 An SRD, other mental illness (e.g., depression), or other preexisting health condition (e.g., cardiovascular disease) may be included in a patient chart during any outpatient visit (e.g., prenatal visit with their obstetrician, psychiatric visit), inpatient visit (e.g., hospitalization), emergency department visit (e.g., delivery), or during one of the many other types of healthcare-related visits. Each of these health-related variables is defined in detail below.
The Institutional Review Board approved the study protocol (#191588; Date of approval: October 1, 2019). 15 Data were collected from the health center's biomedical informatics team through their standardized data request process. Data were provided by staff in a secured Health Insurance Portability and Accountability Act (HIPAA)-approved Virtual Research Desktop (VRD). The VRD interface is protected by multi-factor authentication and is managed and monitored by the biomedical informatics team.

Measures
The primary outcomes of this study were preterm delivery (20-36-weeks' gestation; yes/no), cesarean delivery (yes/no), and co-occurring preterm and cesarean delivery (yes/no). Because the type of delivery includes vaginal, cesarean, and spontaneous or therapeutic abortion, the sample used to assess cesarean delivery only included those with a cesarean or vaginal delivery. As such, those with a spontaneous or therapeutic abortion (n = 43) were omitted from the analysis for the cesarean delivery cohort but not for the preterm delivery or co-occurring preterm and cesarean delivery cohort.
The primary predictor variable was preexisting and/or new SRD (yes/no) during the antepartum and intrapartum period. Any SRD ICD-10 code for alcohol, opioids, cannabis, stimulants (i.e., cocaine and methamphetamines), nicotine, and nonspecific SRDs or other (i.e., sedatives, hallucinogens, and inhalants) were included in the SRD variable (Table A in the supplemental material). SRDs that were only identified after the intrapartum period were not included in the analysis. In addition, the relationship between an alcohol-, opioid-, cannabis-, stimulant-, or nicotine-related diagnosis and the delivery outcomes were assessed individually.
Covariates included age  and race/ethnicity (Hispanic/Latina, non-Hispanic/Latina Black, non-Hispanic/Latina White, and other race/ethnicity [American Indian/Alaskan Native, Asian/Pacific Islander, and other race or mixed)]) at delivery. Patients are asked to include their race (e.g., Black, White) and ethnicity (e.g., Hispanic/Latina, African American, and Caucasian) as separate categories on intake. To manage small cell sizes, those who did not select Hispanic/Latina, non-Hispanic/Latina Black or non-Hispanic/Latina White were grouped into the "other" category. Other variables include marital status (single, divorced/separated/widowed, or married), and Body Mass Index (BMI; calculated as weight in kilograms divided by height in meters squared) at delivery. To assess and control for the impact of previous pregnancies, one or more previous pregnancy at ≥ 20 weeks and ending in a livebirth or stillbirth (yes/no) was identified. Health insurance was defined as private (e.g., commercial), public (e.g., Medicaid) and no insurance. Those in the private insurance category could also have public insurance. Those grouped in the public insurance category did not have private insurance. Table A in the supplemental material includes a full list of the ICD-10 codes used to identify serious mental illness (SMI; e.g., schizophrenia) and non-SMI (e.g., anxiety). A summary variable for preexisting health condition included cardiovascular disease, diabetes (non-gestational), anemia, kidney failure, hypertension, lupus erythematosus, epilepsy, pulmonary disease, cancer, human immunodeficiency virus, acquired immunodeficiency syndrome, hepatitis C virus, and tuberculosis (codes supplied on request).

Statistical analysis
Descriptive statistics were used to identify the number and type of SRD during the antepartum and intrapartum periods. Using propensity score matching, pregnant women with and without an SRD were matched 1:1 (50% with an SRD and 50% without an SRD) on known predictors of preterm and/or cesarean delivery (i.e., age at delivery, BMI at delivery, ≥ 1 previous pregnancy at ≥ 20 16 For example, for every woman with a preexisting health condition and an SRD there is also a woman with a preexisting health condition and no SRD. Due to the reliance on ICD-10 codes, these data may represent an inconsistent distribution of ≥ 1 previous pregnancy compared to the rate in other studies (e.g., 92% vs. 73%). 19 The analysis was repeated without matching on this variable, and again for primigravida women. Standardized mean differences were used to examine the balance of covariate distribution between the groups.
Unadjusted and adjusted analyses were conducted in the unmatched and matched cohorts using analysis of variance (ANOVA) for continuous data and Chi-square (χ 2 ) tests for categorical data. To determine the effect/magnitude of the associations, unadjusted odds ratios (ORs) were calculated and reported. Two-sided tests with 95% confidence intervals (CIs) that cross one, indicating that there was no significant difference, and p-values ≥ 0.05 were used to determine whether a covariate would be included in the final adjusted regression models.
Regression analyses were performed and reported separately for (1) preterm delivery, (2) cesarean delivery, and (3) co-occurring preterm and cesarean delivery. The removal of the cases that resulted in spontaneous or elective abortion (n = 43) from the cesarean cohort created small differences in sample sizes for the preterm matched cohort (n = 2,158) and the cesarean matched cohort (n = 2,154). Due to this small difference in sample size, Figure 1 only represents the results for preterm and cesarean delivery in the preterm matched cohort.
Multivariable logistic regression models were conducted to determine the variables that were associated with having the three delivery outcomes compared to those without the three delivery outcomes. Standardized betas (B), standard errors (SE [B]), adjusted odds ratios (AOR), and the respective CIs and p-values were reported. Only variables significant in the unadjusted (bivariate) regression models (Tables 1 and 2) were included in the adjusted (multivariable) regression models (Table 3) for each outcome variable (preterm delivery, cesarean delivery, co-occurring preterm and cesarean delivery). As such, the variables included for the preterm delivery cohort (left side of this table) differ from the variables included in the cesarean delivery cohort (right side of this table).
The same analyses mentioned previously were then conducted by substance type. Only the results from the individual multivariable regressions by substance type, not the covariates (e.g., age, BMI), were reported. All analyses were conducted with SAS 9.4 (SAS Institute, Cary, North Carolina).

Sample characteristics in the unmatched cohort
There were 19,350 deliveries with an ICD-10 code for a single delivery at ≥ 20 weeks' gestation ( Figure 1). Four individuals were diagnosed with an SRD after delivery and were removed from the analysis. Of the 19,346 deliveries in the unmatched cohort, preterm and cesarean deliveries were reported in 2159 (11%) and 5847 (30%) deliveries respectively ( Table B in the supplemental material). Of these, 1133 (6%) had a co-occurring preterm cesarean delivery. An SRD was reported in 1113 (6%) in the preterm cohort and 1111 (6%) in the cesarean cohort. When grouped by SRD type, the most common SRD included nicotine (16%), cannabis (16%), stimulants (14%), opioids (10%), and alcohol (5%). In those with an SRD, 185 (17%) had a preterm delivery, 487 (44%) had a cesarean delivery, and 111 (10%) had a co-occurring preterm and cesarean delivery.

Sample characteristics in the matched cohort
In the preterm (n = 2158) and cesarean delivery (n = 2154) matched cohorts, 280 (13%) preterm deliveries and 833 (39%) cesarean deliveries were reported (Tables 1 and 2). Of these, 166 (8%) had a co-occurring preterm cesarean delivery. Due to matching on SRD, an SRD was reported in 1,079 (50%) in the preterm cohort and 1,077 (50%) in the cesarean cohort. Figure 2 represents the distribution of delivery outcomes by SRD type. In those with an SRD, 171 (16%) had a preterm delivery, 477 (44%) had a cesarean delivery, and 105 (10%) had a co-occurring preterm and cesarean delivery (Figures 1 and 2). Women with an SRD and co-occurring preterm and cesarean delivery accounted for 2 of the 4 spontaneous abortions.

Prevalence and correlates of preterm delivery
In the matched adjusted analysis, an SRD (Table 4).

Prevalence and correlates of co-occurring preterm and cesarean delivery
In the matched adjusted analysis, an SRD (AOR = 1.81 [95% CI, 1.30-2.52], p-value = 0.0004) and preexisting health condition (AOR = 2.42 [95% CI, 1.72-3.41], p-value = <0.0001) were associated with co-occurring preterm and cesarean delivery (data not shown in tables).   . *results show that opioid-, stimulant-, and any substance-related diagnoses were significantly associated with preterm delivery. stimulant-, nicotine-, and any substance-related diagnoses were significantly associated with cesarean delivery. alcohol-, opioid-, stimulant-, nicotine-, and any substance-related diagnoses were significantly associated with co-occurring preterm and cesarean delivery. a cannabis-related diagnosis was not significantly associated with preterm or cesarean delivery. the sample size and percentage for each substance by those with and without preterm and cesarean delivery can be found in table 4.  of this table). a other race/ethnicity includes american indian/alaskan native (n = 15), asian/Pacific islander (n = 141), and other race or mixed race/ethnicity (n = 659). * P ≤ 0.05. **P ≤ 0.01. *** P ≤ 0.001.
In the matched adjusted analysis for co-occurring preterm and cesarean delivery by substance type, alcohol-related diagnosis ( , p-value = 0.0278) were associated with co-occurring preterm and cesarean delivery (Figure 2 and Table 4).

Discussion
In a large matched pregnancy cohort at a tertiary care center from 2012 to 2019, women with a preterm delivery were more likely to have an SRD, an opioid-or stimulant-related diagnosis, lower BMI, and a preexisting health condition. Women with a cesarean delivery were more likely to have an SRD, a stimulant-or nicotine-related diagnosis, older age, a higher BMI, and preexisting health condition. Finally, women with a co-occurring preterm and cesarean delivery were more likely to have an SRD, an alcohol-, opioid, stimulant, or nicotine-related diagnosis, and a preexisting health condition. As such, only a cannabis-related diagnosis was not found to be significantly associated with preterm or cesarean delivery.
Findings from this study substantiate the previous findings of an increased risk for preterm delivery, [8][9][10] but not cesarean delivery, 10,20 in pregnant women who use opioids. This may be due to the regional variability associated with Southern California. Variance in these findings may also be related to differences in identifying pregnant women with an ICD-10 code for an opioid-related diagnosis (i.e., use, misuse, or dependence) compared to using a clinical assessment that requires a formal diagnosis per the DSM-5 to identify an opioid use disorder (OUD), which is a type of opioid-related diagnosis. Differences in these findings may also be related to the unmeasured and unknown impact of opioid agonist therapy (e.g., methadone) on delivery outcomes. Interestingly, an opioid-related diagnosis remained significantly associated with co-occurring preterm and cesarean delivery. This indicates that the strength of the relationship between opioid use and preterm delivery may be driving this observed association.
One study found that preterm delivery was more common in methadone-exposed deliveries (25%) compared to buprenorphine-exposed deliveries (14). 21 This suggests that there may be maternal physiological changes related to different agonist therapies. Additional research on how untreated OUDs and treated OUDs with medications for opioid use disorder (e.g., methadone, buprenorphine) during pregnancy impacts delivery outcomes is needed to determine the safety and utility of these essential medications.
A similar association between stimulant use and preterm delivery has been observed in previous studies. 22,23 A systematic review and meta-analysis of 31 studies found that cocaine use was associated with an increased risk for preterm delivery. 24 Cocaine use in pregnant women has also been found to lead to severe hypertension, hyperreflexia, proteinuria, edema, and seizures, which are outcomes that may present as preeclampsia. 25 In some older studies, an increased risk of placental abruption, 26 uterine rupture, 27 miscarriage, 28 and stillbirth 29 have been identified in women who use cocaine during pregnancy. 23 Currently, there is limited research on the relationship between substance use and cesarean delivery. In this study, stimulant-and nicotine-related diagnoses were the strongest predictors of cesarean delivery. This finding is supported by a previous study which found that prenatal methamphetamine use was significantly associated with preterm delivery, cesarean delivery, and maternal intensive care unit admission. 30 A significant relationship between nicotine use during the perinatal period and preterm or cesarean delivery has also been observed in the literature. 31,32 In this current study, nicotine was significantly associated with cesarean delivery, but not preterm delivery. However, a nicotine-related diagnosis remained significantly associated with co-occurring preterm and cesarean delivery, indicating that the strength of the relationship between nicotine use and cesarean delivery may be driving this observed association. An alcohol-related diagnosis was not found to be significantly associated with preterm or cesarean delivery individually. However, a significant association with an alcohol-related diagnosis was observed in women with co-occurring preterm and cesarean delivery. Associations between alcohol use and preterm delivery has been observed in numerous studies, 33,34 indicating that the strength of the relationship between alcohol use and preterm delivery may be driving this observed association. Further research on how alcohol use impacts delivery outcomes such as cesarean delivery is warranted.
Finally, a cannabis-related diagnosis was not found to be significantly associated with preterm delivery, cesarean delivery, or co-occurring preterm and cesarean delivery. This finding differs from other studies which have observed an association between cannabis use and preterm delivery. 21 One study in France found that women who used cannabis demonstrated higher rates of preterm delivery and spontaneous preterm delivery. 35 In addition, the association was observed in those who used cannabis once a month or more often, and especially in those who also used tobacco.
To our knowledge, no study has investigated the relationship between a cannabis-related diagnosis and cesarean delivery. As such, additional research on the relationship between a prenatal cannabis use and delivery outcomes is needed.
The findings in this study reinforce the need to identify SRDs in pregnant women early to minimize potential harm through intervention and treatment. Our data show high prevalence and risk of adverse delivery outcomes in those with an SRD, which supports a rational for robust SRD screening measures in all clinical settings. Because pregnant women with an SRD may be engaging with the health system in different capacities (pre-natal visit vs. emergency department), questions regarding substance use should be posed often. Screening through questions related to substance use should be posed sensitively and should include questions specifically related to specific substances such as alcohol, opioids, cannabis, stimulants, and nicotine. 36,37 Future studies should investigate the biological and environmental impact substance use, polysubstance use, and medication treatments (e.g., methadone) have on delivery outcomes, and differentiate between substance exposure, lifestyle factors, and the potential benefit of treatment on perinatal outcomes such as preterm and cesarean delivery. Screening and monitoring interventions should be implemented and tested in all types of clinical encounters including prenatal, primary care, and psychiatry visits to prevent substance use during pregnancy and subsequent adverse delivery outcomes. Investigating how stigma impacts these same delivery outcomes in these settings could also lead to increased recognition, appropriate diagnoses, and engagement in treatment.

Strengths and limitations
The research presented in this study used robust methodology in a large sample to evaluate previous potential negative outcomes and expand previous findings to address the relationships between SRDs and delivery outcomes. This study is strengthened by the large sample size over 7.5 years and the use of propensity score matching to control for confounding in the unstructured EHR data. By matching on key baseline characteristics, a greater portion of potential bias was eliminated when estimating the effects of an SRD on delivery outcomes. 38 This study is limited by the reliance on ICD-10 codes for health-related diagnoses which can lead to misclassification bias, unmeasured confounding (e.g., SRD treatment), changes in eligibility over time, and missing data. 39 These data represent an inconsistent distribution of ≥ 1 previous pregnancy compared to the rate in other studies. 21 To address this concern, the analysis was repeated without matching on this variable, and again for primigravida women. The relationship between an SRD, the other covariates (e.g., age), and the outcome variables remained significant in both analyses, confirming that our decision to match on ≥ 1 previous pregnancy did not impact the final results. It may be difficult to generalize these results to other care settings due to the race/ethnicity distribution (e.g., low proportion of patients who report Black race or Hispanic/ Latina ethnicity) in one healthcare system in Southern California.

Conclusion
The findings from this study reveal that pregnant women with an SRD are experiencing disproportionally higher odds of preterm and cesarean delivery compared to pregnant women without an SRD in a large matched pregnancy cohort. Substance type predicted the type of delivery outcome. Opioid-and stimulant-related diagnoses were significantly associated with preterm delivery while stimulant-and nicotine-related diagnoses were significantly associated with cesarean delivery. An alcohol-, opioid-stimulant-, and nicotine-related diagnosis were more likely to be observed in women with a co-occurring preterm cesarean delivery. A cannabis-related diagnosis was not significantly associated with preterm or cesarean delivery. These findings indicate that an SRD in pregnant women should be identified early to reduce potential harm through intervention and treatment.

Disclosure statement
No potential conflict of interest was reported by the author(s).