Abstract
Objectives. To examine the extent to which differences in medication for opioid use disorder (MOUD) in pregnancy and infant neonatal opioid withdrawal syndrome (NOWS) outcomes are associated with maternal race/ethnicity.
Methods. We performed a secondary analysis of a statewide quality improvement database of opioid-exposed deliveries from January 2017 to April 2019 from 24 hospitals in Massachusetts. We used multivariable mixed-effects logistic regression to model the association between maternal race/ethnicity (non-Hispanic White, non-Hispanic Black, or Hispanic) and prenatal receipt of MOUD, NOWS severity, early intervention referral, and biological parental custody at discharge.
Results. Among 1710 deliveries to women with opioid use disorder, 89.3% (n = 1527) were non-Hispanic White. In adjusted models, non-Hispanic Black women (AOR = 0.34; 95% confidence interval [CI] = 0.18, 0.66) and Hispanic women (AOR = 0.43; 95% CI = 0.27, 0.68) were less likely to receive MOUD during pregnancy compared with non-Hispanic White women. We found no statistically significant associations between maternal race/ethnicity and infant outcomes.
Conclusions. We identified significant racial/ethnic differences in MOUD prenatal receipt that persisted in adjusted models. Research should focus on the perspectives and treatment experiences of non-Hispanic Black and Hispanic women to ensure equitable care for all mother–infant dyads.
The number of individuals with opioid use disorder (OUD) and the rate of overdose deaths in the United States have increased dramatically over the last 2 decades with a more than 400% increase in opioid overdose deaths among women,1,2 but the opioid crisis has not affected all racial and ethnic groups equally. Among women, non-Hispanic White women are more likely to be prescribed opioids as treatment of chronic pain, to visit the emergency department because of opioid misuse, to have an OUD in pregnancy, and to die of an opioid overdose compared with Black and Hispanic women.3–5 In the general population, while non-Hispanic White individuals have higher rates of opioid overdose deaths,2 non-Hispanic Black individuals with OUD increasingly experience a greater burden of opioid-related mortality, particularly among middle-aged urban adults compared with non-Hispanic White people.6,7 In addition, studies have found that non-Hispanic Black and Hispanic individuals are less likely to be able to access, receive, and complete substance use treatment than are non-Hispanic White individuals, including receipt of medications for OUD (MOUD).8–10 Less is currently known about racial/ethnic differences in MOUD treatment utilization and neonatal health outcomes in the perinatal period.
The recommended treatment of OUD in pregnancy is medication with methadone or buprenorphine combined with behavioral therapy.11 After delivery, some infants exposed to opioids in utero (including MOUD) may experience withdrawal signs, known as neonatal abstinence syndrome or neonatal opioid withdrawal syndrome (NOWS).12 Given increasing concerns about racial and ethnic differences in substance use disorder treatment outcomes combined with documented disparities in key maternal health outcomes, the US Department of Health and Human Services’ Office of Women’s Health has called for increased research examining differences in key outcomes for opioid-exposed mother–infant dyads by race or ethnicity, including maternal receipt of MOUD and infant diagnosis of NOWS.13
We used a Massachusetts statewide quality improvement database of opioid-exposed deliveries to examine the extent to which differences in (1) prenatal receipt of MOUD, (2) severity of NOWS, (3) infant care characteristics including rooming in and breastfeeding, and (4) biological parental custody at discharge are associated with maternal race/ethnicity. We hypothesized that among opioid-exposed dyads in Massachusetts, there will be differences in key maternal and infant health utilization and outcomes by maternal race/ethnicity.
METHODS
We performed a secondary analysis of data collected by the Perinatal-Neonatal Quality Improvement Network of Massachusetts (PNQIN) as part of a quality improvement project. PNQIN developed a standardized data form of key process and outcome measures for tracking NOWS outcomes across the state.14 The database was launched in January 2017, and this analysis includes data collected through April 2019. Hospital participation in PNQIN and data collection is voluntary: among 47 birthing hospitals in Massachusetts, 26 are submitting data.15 We included 24 hospitals in this analysis, excluding hospitals with fewer than 10 deliveries during the study period.
Hospitals used different assessment tools for NOWS severity including transitioning to the Eat, Sleep, and Console method during the study period (n = 15) as well as the modified Finnegan Neonatal Abstinence Scoring Tool (n = 9). Hospitals in our cohort, which accounted for 61% of deliveries in the state, had higher rates of deliveries receiving public insurance and more deliveries to women with adequate prenatal care.16 A comparison of the differences between included and not-included hospitals is found in Table A (available as a supplement to the online version of this article at http://www.ajph.org). Participating hospitals have designated team members who extracted and submitted information on all mother–infant dyads in which the infant was at risk for NOWS because of in utero opioid exposure by using a secure electronic database (see Appendix, available as a supplement to the online version of this article at http://www.ajph.org). The database is maintained at Beth Israel Deaconess Medical Center (BIDMC) where the BIDMC institutional review board reviewed this project as part of the larger PNQIN quality improvement effort.
Variables
Our primary maternal outcome was any MOUD during pregnancy, defined as receipt of methadone or buprenorphine during pregnancy (abstracted from the electronic medical record based on record of buprenorphine prescription, documentation of methadone in prenatal notes, medication list at delivery, or delivery hospitalization admission notes). Infant outcomes were categorized as either NOWS severity or infant social service outcomes. NOWS severity was measured by any infant receipt of pharmacological treatment of NOWS (defined as need for any morphine, methadone, clonidine, or phenobarbital during delivery hospitalization). We measured infant social services outcomes during the infant birth hospitalization through infant referrals to early intervention (early childhood developmental services) and parental custody (defined as discharge home with biological parent). Other infant care outcomes included rooming in for at least 1 night before maternal discharge (yes or no) and receipt of any of the mother’s own breastmilk among eligible infants (yes or no).
Our primary exposure was maternal race/ethnicity, defined as non-Hispanic White, non-Hispanic Black, or Hispanic, based on the patient’s self-report at the time of hospital registration. We removed women of Asian, unknown, or other race from the analysis because of their small numbers (representing 6.3% of the total sample).
We categorized maternal OUD as untreated OUD (nonprescribed opioid use including heroin, fentanyl, and nonmedical use of prescription opioids [including methadone or buprenorphine] during the pregnancy without any MOUD received), stable treated OUD (prescribed MOUD with no nonprescribed opioid use), and unstable treated OUD (concomitant prescribed MOUD use with nonprescribed opioid use) during pregnancy. Nonprescribed opioid use was determined by self-report, maternal toxicology testing, or neonatal toxicology testing. We defined any nonprescribed opioid use as the combination of untreated OUD and unstable treated OUD categories described previously. Other maternal characteristics included any use of selective serotonin reuptake inhibitor (SSRI), nicotine, alcohol, benzodiazepine, cocaine, or either amphetamine or methamphetamine, though our data set did not differentiate prescribed from unprescribed use. We defined geographic region of birth hospitalization by using the Massachusetts Executive Office of Health and Human Services Regions. Boston and Metro West regions were consolidated into 1 region (“Metro Boston”) to allow for 5 regions with relatively equal NOWS populations (see Figure A, available as a supplement to the online version of this article, for map).17 Infant birth characteristics included infant sex (defined as male or female), location of birth (inborn in the hospital managing NOWS signs or born at a different location), and term delivery (defined as infant born at ≥ 37 weeks).
Statistical Analysis
We compared maternal and infant characteristics across maternal racial/ethnic groups by using the χ2 test and Fisher exact test for categorical variables. For continuous variables, we presented medians with interquartile ranges and assessed differences by using the Wilcoxon rank-sum test. For our primary maternal outcome, we used a multivariable mixed-effects logistic regression model to assess associations of mother’s race/ethnicity on any MOUD receipt during pregnancy, adjusting for other maternal substance use as well as region of delivery hospital. We also used multivariable mixed-effects logistic regression models to measure the associations of mother’s race/ethnicity on each infant care outcome after adjusting for maternal characteristics (i.e., nonprescribed drug use and MOUD receipt), infant characteristics (i.e., sex and gestational age at delivery), and region of delivery hospital.
We used directed acyclic graphs for model development, shown in Figure B, available as a supplement to the online version of this article at http://www.ajph.org). All models included random intercepts to control for clustering by delivery hospital. We used a logistic link for binary outcomes: MOUD receipt, rooming in, any maternal breastmilk, pharmacologic treatment of NOWS, early intervention referral, and discharge home with biological parent), and measures of associations were reported in terms of odds ratios (ORs). We performed analyses by using SAS statistical software (version 9.4; SAS Institute Inc, Cary, NC). Statistical significance was achieved with a 2-sided P of less than .05. We confirmed model fit by ensuring no convergence issues in our multivariable mixed effects regression models.
Sensitivity Analyses
We conducted 3 sensitivity analyses. First, because of small sample sizes among non-Hispanic Black and Hispanic women in our cohort, we re-examined our main exposure as a dichotomous variable (non-Hispanic White compared with other) and reassessed our primary infant outcomes. Second, to assess how type of MOUD affected use by maternal race/ethnicity, we separated our primary outcome into receipt of any buprenorphine, exclusively methadone,18 or no treatment using a multivariable mixed-effects multinomial logistic regression model. Given emerging literature suggesting disparities in buprenorphine utilization by race/ethnicity, we were interested in any receipt of buprenorphine, so women who received both methadone and buprenorphine were classified as “any buprenorphine.” Finally, in a subset of hospitals where we were able to identify repeat deliveries to the same mother, we excluded repeat deliveries to assess for any effect of clustering by maternal characteristics.
RESULTS
Between January 2017 and April 2019, 1989 opioid-exposed infants born to women with OUD were cared for in 24 hospitals across Massachusetts. We excluded 65 (3.3%) women from the study sample who had no opioid exposure recorded (e.g., cocaine exposure, other psychiatric medications, and naltrexone) and 100 (5.0%) who used only prescription opioids with no diagnosis of OUD, as they would not have a clinical indication to receive treatment with MOUD. After removing women classified as Asian, other, or unknown race/ethnicity, our final study cohort comprised 1710 deliveries. In our final sample, 89.3% of women (n = 1527) were non-Hispanic White, 3.3% (n = 57) were non-Hispanic Black, and 7.4% (n = 126) were Hispanic (Figure C, available as a supplement to the online version of this article at http://www.ajph.org). The race/ethnicity of women delivering opioid-exposed infants was not distributed evenly around the state, with the largest proportion of non-Hispanic Black women delivering in the Boston Metro region.
Maternal and Infant Characteristics
Maternal and infant characteristics for the entire sample and by maternal race/ethnicity are presented in Table 1. Among all deliveries to women with OUD during pregnancy, 67.5% of women received MOUD and had no nonprescribed opioid use, 20.1% received MOUD with nonprescribed opioid use, and 12.5% had nonprescribed opioid use with no MOUD receipt. There were significant differences (P < .001) in the use of MOUD during pregnancy by race/ethnicity, with 88.9% (n = 1357) of non-Hispanic White women receiving any MOUD compared with 75.4% (n = 43) of non-Hispanic Black women and 77.0% (n = 97) of Hispanic women. There was also a significant difference in the type of MOUD prescribed (P < .001) with non-Hispanic Black and Hispanic women significantly more likely to receive methadone than buprenorphine compared with non-Hispanic White women. Non-Hispanic Black women were significantly more likely (P = .004) to use cocaine during pregnancy, with 31.6% using cocaine compared with 26.2% of Hispanic women and 18.0% of non-Hispanic White women.
TABLE 1—
Racial/Ethnic Differences in Opioid-Exposed Mother–Infant Dyad Characteristics: Massachusetts Perinatal-Neonatal Quality Improvement Network Database, 2017–2019
| All Participants (n = 1710), No. (%) or Median (IQR) | Non-Hispanic White, (n = 1527), No. (%) or Median (IQR) | Non-Hispanic Black (n = 57), No. (%) or Median (IQR) | Hispanic (n = 126), No. (%) or Median (IQR) | P | |
| Maternal exposures | |||||
| Maternal opioid exposurea | < .001 | ||||
| Untreated OUD | 213 (12.5) | 170 (11.1) | 14 (24.6) | 29 (23.0) | |
| Stable treated OUD | 1154 (67.5) | 1055 (69.1) | 29 (50.9) | 70 (55.6) | |
| Unstable treated OUD | 343 (20.1) | 302 (19.8) | 14 (24.6) | 27 (21.4) | |
| Any maternal nonprescribed opioid use | 556 (32.5) | 472 (30.9) | 28 (49.1) | 56 (44.4) | < .001 |
| Any MOUD during pregnancy | 1497 (87.5) | 1357 (88.9) | 43 (75.4) | 97 (77.0) | < .001 |
| Type of MOUD prescribed | < .001 | ||||
| Methadone only | 699 (40.9) | 616 (40.3) | 21 (36.8) | 62 (49.2) | |
| Buprenorphine only | 784 (45.9) | 727 (47.6) | 22 (38.6) | 35 (27.8) | |
| Methadone and buprenorphineb | 14 (0.8) | 14 (0.9) | 0 (0) | 0 (0) | |
| No MOUD | 213 (12.5) | 170 (11.1) | 14 (24.6) | 29 (23.0) | |
| Other maternal exposures | |||||
| Alcohol | 53 (3.1) | 48 (3.1) | 0 (0) | 5 (4.0) | .34 |
| Benzodiazepine | 252 (14.7) | 228 (14.9) | 6 (10.5) | 18 (14.3) | .65 |
| Cocaine | 326 (19.1) | 275 (18.0) | 18 (31.6) | 33 (26.2) | .004 |
| Marijuana | 462 (27.0) | 397 (26.0) | 21 (36.8) | 44 (34.9) | .023 |
| Nicotine | 884 (51.7) | 779 (51.0) | 37 (64.9) | 68 (54.0) | .1 |
| Amphetamines | 134 (7.8) | 120 (7.9) | 5 (8.8) | 9 (7.1) | .93 |
| SSRI | 165 (9.6) | 150 (9.8) | 8 (14.0) | 7 (5.6) | .16 |
| Geography of delivery hospitalization | |||||
| Region | < .001 | ||||
| West | 430 (25.2) | 383 (25.1) | 13 (22.8) | 34 (27.0) | |
| Central | 311 (18.2) | 269 (17.6) | 8 (14.0) | 34 (27).0 | |
| Northeast | 301 (17.6) | 282 (18.5) | 1 (1.8) | 18 (14.3) | |
| Metro Boston | 333 (19.5) | 296 (19.4) | 22 (38.6) | 15 (11.9) | |
| Southeast | 335 (19.6) | 297 (19.5) | 13 (22.8) | 25 (19.8) | |
| Infant characteristics | |||||
| Gestation age, wk | 39 (37–39) | 39 (37–39) | 39 (38–39) | 38 (36–39) | .044 |
| Male sex | 874 (51.1) | 775 (50.8) | 33 (57.9) | 66 (52.4) | .55 |
| Term (≥ 37 wk) | 1404 (82.1) | 1262 (82.7) | 50 (87.7) | 92 (73.0) | .013 |
| Inborn at hospital | 1624 (95.0) | 1451 (95.0) | 54 (94.7) | 119 (94.4) | .99 |
| NOWS severity and engagement outcomes | |||||
| Length of stay, d | 12 (6–20) | 12 (6–20) | 14 (8–22) | 13 (7–20) | .28 |
| Received pharmacological therapy | 845 (49.6) | 748 (49.2) | 33 (57.9) | 64 (51.2) | .41 |
| Received skin to skin | 1219 (71.6) | 1092 (71.8) | 42 (73.7) | 85 (67.5) | .54 |
| Roomed in for ≥ 1 night | 1424 (76.2) | 1170 (76.9) | 42 (73.7) | 87 (69.1) | .13 |
| Any maternal breastmilk during hospitalization | 856 (76.8) | 773 (76.1) | 24 (80.0) | 59 (85.5) | .18 |
| Social services outcomes | |||||
| Early intervention referral | 1338 (78.7) | 1192 (78.6) | 45 (78.9) | 101 (80.2) | .92 |
| Discharge to biological parent | 1259 (74.1) | 1136 (74.8) | 37 (64.9) | 86 (68.8) | .09 |
Note. IQR = interquartile range; MOUD = medications used to treat opioid use disorder; NOWS = neonatal opioid withdrawal syndrome; OUD = opioid use disorder; SSRI = selective serotonin reuptake inhibitor.
Untreated OUD = nonprescribed opioid use including heroin, fentanyl, and nonmedical use of prescription opioids (including methadone or buprenorphine) during the pregnancy without any MOUD received; stable treated OUD = prescribed MOUD with no nonprescribed opioid use during pregnancy; and unstable treated OUD = concomitant prescribed MOUD use with nonprescribed opioid use during pregnancy.
Evidence of receipt of both prescribed methadone and buprenorphine at some point during pregnancy, not at the same time.
Maternal and Infant Outcomes
Differences in MOUD receipt during pregnancy are described in Table 2. In both our unadjusted and adjusted models, we found that non-Hispanic Black women and Hispanic women were significantly less likely to receive any MOUD during pregnancy compared with non-Hispanic White women. In our final model, after we adjusted for the fixed effects of region and other maternal exposures, as well as for the random effects of delivery hospital, non-Hispanic Black women (adjusted odds ratio [AOR] = 0.34; 95% confidence interval [CI] = 0.18, 0.66) and Hispanic women (AOR = 0.43; 95% CI = 0.27, 0.68) were significantly less likely to receive any MOUD during pregnancy compared with non-Hispanic White women. Receipt of MOUD did not vary significantly by region when we controlled for race/ethnicity and other maternal substance use. In our final adjusted model, women who used cocaine during pregnancy were less likely to receive any MOUD (AOR = 0.47; 95% CI = 0.33, 0.66), but no other maternal substance use was associated with MOUD receipt.
TABLE 2—
Unadjusted and Adjusted Odds of Use of Any Medication for Opioid Use Disorder in Pregnancy: Massachusetts Perinatal-Neonatal Quality Improvement Network Database, 2017–2019
| Any MOUD in Pregnancy | |||
| No. (%) | OR (95% CI) | AORa (95% CI) | |
| Race/ethnicity | |||
| Non-Hispanic White | 1527 (89.3) | 1 (Ref) | 1 (Ref) |
| Non-Hispanic Black | 57 (3.3) | 0.34 (0.18, 0.65) | 0.34 (0.18, 0.66) |
| Hispanic | 126 (7.4) | 0.40 (0.25, 0.63) | 0.43 (0.27, 0.68) |
| Region | |||
| Metro Boston | 333 (19.5) | 1 (Ref) | 1 (Ref) |
| Central | 311 (18.2) | 0.59 (0.23, 1.49) | 0.59 (0.22, 1.55) |
| Northeast | 301 (17.6) | 0.81 (0.37, 1.81) | 0.79 (0.34, 1.81) |
| Southeast | 335 (19.6) | 0.76 (0.33, 1.75) | 0.71 (0.30, 1.70) |
| West | 430 (25.2) | 0.90 (0.39, 2.05) | 0.84 (0.36, 1.99) |
| Other maternal substance use | |||
| Cocaine | 326 (19.1) | 0.43 (0.31, 0.60) | 0.47 (0.33, 0.66) |
| Alcohol | 53 (3.1) | 0.43 (0.22, 0.83) | 0.55 (0.27, 1.10) |
| Benzodiazepines | 252 (14.7) | 0.99 (0.65, 1.49) | 1.00 (0.65, 1.54) |
| Nicotine | 884 (51.7) | 1.17 (0.87, 1.59) | 1.28 (0.93, 1.75) |
| Amphetamines or methamphetamine | 134 (7.8) | 1.20 (0.67, 2.14) | 1.19 (0.66, 2.17) |
| Medication to treat depression: SSRI | 165 (9.7) | 1.77 (0.98, 3.20) | 1.76 (0.96, 3.42) |
Note. AOR = adjusted odds ratio; CI = confidence interval; MOUD = medications used to treat opioid use disorder; OR = odds ratio; SSRI = selective serotonin reuptake inhibitor medication.
Adjusted for all covariates listed with random intercepts included to control for clustering by delivery hospital.
Adjusted infant NOWS and social service outcomes are presented in Table 3. After we adjusted for infant gestational age, sex, any illicit drug use in mother, maternal receipt of MOUD, region, and random effects of delivery hospital, there was no significant effect of race/ethnicity on any of the infant care characteristics measured, including pharmaceutical treatment of NOWS, rooming in, receipt of maternal breastmilk during hospitalization (among eligible infants), early intervention referral, or discharge home to a biological parent. Male versus female sex (AOR = 1.35; 95% CI = 1.04, 1.73), maternal nonprescribed opioid use (AOR = 1.51; 95% CI = 1.01, 2.25), benzodiazepine use (AOR = 1.73; 95% CI = 1.15, 2.60), SSRI use (AOR = 1.61; 95% CI = 1.06, 2.45), and MOUD use compared with no treatment (AOR = 3.66; 95% CI = 1.77, 7.58) were significantly associated with the need for infant pharmaceutical treatment of NOWS. Rooming in for at least 1 night (AOR = 0.36; 95% CI = 0.24, 0.54) was associated with a reduction in NOWS severity. Maternal nonprescribed opioid use was negatively correlated with rooming in (AOR = 0.49; 95% CI = 0.34, 0.70) and discharge home with a biological parent (AOR = 0.22; 95% CI = 0.16, 0.30). Infants born to women who received MOUD during pregnancy, compared with those who did not, were more likely be referred to early intervention (AOR = 2.27; 95% CI = 1.62, 4.82). A comparison of both maternal and infant outcomes is shown in Figure 1.
TABLE 3—
NOWS Severity, Birth Hospitalization, and Social Services Outcomes: Massachusetts Perinatal-Neonatal Quality Improvement Network Database, 2017–2019
| NOWS Severity and Birth Hospitalization Outcomes, AORa (95% CI) | Social Services Outcomes, AORa (95% CI) | ||||
| Pharmaceutical Treatment of NOWS | Rooming In | Any Maternal Breastmilk During Hospitalization (Eligible Only) | Early Intervention Referral | Discharged in Biological Parent’s Custody | |
| Race/ethnicity | |||||
| Non-Hispanic White (Ref) | 1 | 1 | 1 | 1 | 1 |
| Non-Hispanic Black | 1.25 (0.57, 2.73) | 0.74 (0.35, 1.56) | 1.05 (0.41, 2.71) | 0.86 (0.36, 2.08) | 1.02 (0.53, 1.97) |
| Hispanic | 1.06 (0.62, 1.81) | 1.27 (0.77, 2.12) | 2.07 (1.01, 4.25) | 1.40 (0.78, 2.49) | 1.24 (0.76, 2.03) |
| Term (≥ 37 wk) vs preterm | 1.62 (1.05, 2.48) | 13.47 (9.60, 18.90) | 1.07 (0.69, 1.64) | 0.53 (0.34, 0.80) | 2.29 (1.68, 3.13) |
| Male vs female sex | 1.35 (1.04, 1.73) | 0.97 (0.73, 1.28) | 1.05 (0.79, 1.41) | 1.36 (1.01, 1.82) | 0.95 (0.73, 1.23) |
| Maternal nonprescribed opioid use vs no nonprescribed use | 1.51 (1.01, 2.25) | 0.49 (0.34, 0.70) | 0.72 (0.45, 1.13) | 2.09 (1.34, 3.25) | 0.22 (0.16, 0.30) |
| Maternal MOUD vs no treatment | 3.66 (1.77, 7.58) | 1.17 (0.74, 1.85) | 0.96 (0.46, 2.01) | 2.79 (1.62, 4.82) | 1.31 (0.89, 1.93) |
| Maternal cocaine use vs none | 1.11 (0.69, 1.79) | 0.62 (0.43, 0.88) | 1.08 (0.63, 1.87) | 0.89 (0.59, 1.34) | 0.20 (0.15, 0.26) |
| Maternal benzodiazepine use vs none | 1.73 (1.15, 2.60) | 1.06 (0.70, 1.59) | 0.67 (0.43, 1.06) | 1.40 (0.91, 2.13) | 0.48 (0.34, 0.67) |
| Maternal SSRI use vs none | 1.61 (1.06, 2.45) | 1.08 (0.67, 1.75) | 1.55 (0.91, 2.62) | 1.65 (1.00, 2.73) | 1.15 (0.73, 1.81) |
| Region | |||||
| Metro Boston (Ref) | 1 | 1 | 1 | 1 | 1 |
| Central | 1.24 (0.38, 4.03) | 0.59 (0.08, 4.62) | 0.53 (0.15, 1.91) | 0.10 (0.01, 1.75) | 1.03 (0.47, 2.24) |
| Northeast | 1.83 (0.70, 4.81) | 0.78 (0.14, 4.23) | 0.47 (0.16, 1.33) | 4.48 (0.50, 39.99) | 1.53 (0.79, 2.96) |
| Southeast | 1.89 (0.66, 5.44) | 2.29 (0.34, 15.44) | 0.47 (0.15, 1.50) | 1.46 (0.11, 19.98) | 1.36 (0.67, 2.74) |
| West | 0.78 (0.27, 2.15) | 1.05 (0.17, 6.51) | 1.41 (0.45, 4.44) | 0.57 (0.05, 7.01) | 1.44 (0.73, 2.86) |
| Rooming in ≥ 1 night | 0.36 (0.24, 0.54) | . . . | . . . | . . . | . . . |
| Any maternal breastmilk during hospitalization | 0.77 (0.57, 1.05) | . . . | . . . | . . . | . . . |
Note. AOR = adjusted odds ratio; CI = confidence interval; MOUD = medications used to treat opioid use disorder; NOWS = neonatal opioid withdrawal syndrome; SSRI = selective serotonin reuptake inhibitor medication.
Adjusted for all covariates listed with random intercepts included to control for clustering by delivery hospital.
FIGURE 1—
Adjusted Odds of Maternal Medication Receipt and Infant Outcomes by Maternal Race/Ethnicity: Massachusetts Perinatal-Neonatal Quality Improvement Network Database, 2017–2019
Note. AOR = adjusted odds ratio; CI = confidence interval; MOUD = medication for opioid use disorder; NH = non-Hispanic; PT = pharmacological treatment. All referents are non-Hispanic Whites.
Sensitivity Analysis
In our first sensitivity analysis, with our main exposure defined as a dichotomous variable (non-Hispanic White compared with non-White women), we again found no statistically significant findings in our primary infant outcomes (Table B, available as a supplement to the online version of this article at http://www.ajph.org). When comparing how medication type affected prenatal MOUD use, we found that non-Hispanic Black and Hispanic women were less likely to receive buprenorphine compared with exclusive methadone, compared with non-Hispanic White women (Table C, available as a supplement to the online version of this article at http://www.ajph.org). There were no changes in our main findings of differences by race/ethnicity when we removed a subset of repeat deliveries across the study period (Table D).
DISCUSSION
In a cohort of opioid-exposed infants in Massachusetts, non-Hispanic Black and Hispanic women were significantly less likely to receive MOUD for treatment of OUD during pregnancy compared with non-Hispanic White women in both unadjusted and adjusted models controlling for geographic region, delivery hospital, and maternal nonprescribed drug use during pregnancy. However, our data show no statistically significant maternal racial/ethnic differences in infant outcomes including severity of NOWS, receipt of any maternal breastmilk, rooming in, referrals to early intervention, or discharge home with biological parent after we controlled for infant factors such as gestational age at delivery and infant sex, and maternal factors during pregnancy such as smoking, MOUD receipt, cocaine use, benzodiazepine use, or nonprescribed opioid use during pregnancy.
Compared with all deliveries in Massachusetts in 2017 (the most recently available internal data from the Massachusetts Department of Health), where 57% of deliveries were to non-Hispanic White mothers, 10% to non-Hispanic Black mothers, 19% to Hispanic mothers, and 9% to non-Hispanic Asian mothers,16 we found that the overwhelming majority of births of opioid-exposed infants were to non-Hispanic White women. This is consistent with existing literature that has identified that non-Hispanic White pregnant women are more likely to have an OUD than women of other races/ethnicities.4,5
There should be no treatment differences attributable to health status in pregnancy, as all pregnant women with OUD are recommended to initiate MOUD.11 Yet, similar to previous studies, our analysis identified racial and ethnic disparities in prenatal MOUD treatment.4,19,20 We further identified the persistence of disparities after we adjusted for other key clinical metrics including nonprescribed maternal opioid or cocaine use. Our results are consistent with national trends in disparities in MOUD receipt and retention in the general population.8–10 In addition, the disparities we identified in receipt of MOUD during pregnancy mirror longstanding racial/ethnic inequities in access to prenatal care, as well as maternal morbidity and mortality.21
Racial differences in pain treatment and opioid prescription availability has contributed to non-Hispanic White individuals in the general population being disproportionately affected by OUD6,22 and may also contribute to racial/ethnic differences in MOUD use. Factors contributing to non-White individuals being less likely to receive MOUD may include socioeconomic differences, a lack of physicians trained to prescribe buprenorphine in communities of color, fear and stigma around the use of methadone and buprenorphine, and bias, discrimination, and racism within the substance use treatment system.23–25 In Massachusetts, pregnant women receive priority status in opioid use treatment programs and are all eligible for Medicaid, which provides greater treatment access than for the general population, making the disparities we identified even more striking.
Although we were unable to assess for patient preferences in receiving treatment of OUD, on the basis of previous research, we hypothesized that women may feel ambivalent about MOUD because of the stigmatization of substance use in pregnancy.26 In addition, punitive child welfare reporting policies and criminalization of drug use in pregnancy, which historically disproportionately affect families of color, may be a powerful motivator for women to try to hide drug use rather than access treatment.27 Further research is indicated to explore individual-level factors (e.g., lack of engagement in prenatal care, views on MOUD) and system-level factors (e.g., lack of screening for OUD in pregnancy, poor retention in care for non-Hispanic Black and Hispanic women, stigma toward medications to treat OUD) contributing to these inequities.
Notably, despite these differences in maternal treatment use, we did not see any statistical differences in our assessment of infant outcomes, including measures of NOWS severity, family engagement, and discharge and follow-up. Although a recent study found that non-Hispanic Black infants with NOWS were less likely to require pharmacological therapy compared with non-Hispanic White infants with NOWS,28 our study, with more than 5 times the number of dyads, found no differences in need for pharmacological therapy in both the unadjusted model and when we adjusted for factors known to affect NOWS severity such as term delivery, birth weight, maternal polysubstance use, concomitant maternal benzodiazepine use, SSRI use, methadone use, maternal smoking, and male sex of infant.29,30 In addition, we saw no difference among racial/ethnic groups in infant treatment outcomes and social service outcomes.
These findings warrant further confirmation across larger data sources; to date, none of the national studies assessing prevalence of NOWS or reports from other statewide collaboratives have reported information by maternal or infant race/ethnicity.31–33 If these findings are confirmed in other data sources, this could suggest that the health care community has been more successful in addressing or preventing race/ethnicity–based differences in care in the inpatient postbirth setting as compared with the outpatient prenatal setting. It is also possible that hospitals that reported data to our neonatal quality improvement collaborative may not be a representative sample as many have worked for years to standardize and improve newborn care for opioid-exposed infants.
We found that prenatal MOUD use was associated with need for infant pharmacological treatment. This was expected, as chronic in utero opioid exposure with methadone or buprenorphine can cause neonatal opioid withdrawal symptoms. This has been described in smaller analyses using clinical data sets, but is not commonly reported in larger studies looking at factors associated with NOWS severity using administrative data, as current diagnosis claims codes for NOWS do not allow for differentiating if NOWS results from prescribed or nonprescribed sources.28,32,34,35 Given that MOUD utilization is on the expected causal pathway for NOWS, our finding of no differences in infant outcomes despite stark differences in maternal prenatal treatment suggests that other environmental and genetic factors may play a larger role on NOWS severity.
Limitations
There are several important additional limitations in this study. First, because our data set was initially designed to focus on infant outcomes, we did not have complete demographic information for women including education, socioeconomic status, insurance type, language, and other potential confounders that could contribute to the differences across racial/ethnic groups. We were, however, able to adjust for several key clinical maternal coexposures that are associated with NOWS severity including nonprescribed substance use and receipt of SSRIs.
Second, there was a relatively low number of opioid-exposed deliveries by non-Hispanic Black and Hispanic women, and these deliveries were clustered in the Boston Metro region. Despite this, our effects were large enough to be significant in the adjusted models.
Third, we excluded a small number of women of Asian, other, or unknown race/ethnicity. Removing these women may have introduced an omitted-variable bias.
Fourth, because each site had its own maternal testing policies, there is a potential for selection bias in terms of which mothers were identified by toxicology testing and which infants were included as at risk for in utero exposure.
Fifth, dyads were entered into the data set by infant delivery, and, thus, twin deliveries and multiple gestations to the same mother during the study period were counted as 2 or more distinct entries in this analysis. In our sensitivity analysis excluding a subset of repeat deliveries, we did not, however, find differences in our main findings.
Finally, the data set we used accounted for 61% of total births in the state16 with hospitals with the highest rates of NOWS routinely reporting into this data set. Yet it is unknown how many opioid-exposed deliveries occurred at nonparticipating hospitals that were not captured or how many opioid-exposed infants were not identified at participating hospitals. Hence, additional investigation to validate our findings of disparities in MOUD treatment is needed.
Despite these limitations, our study had several strengths, including being a contemporary sample with a large sample size of more than 1700 opioid-exposed deliveries across Massachusetts with clinical data on maternal prenatal exposures.
Public Health Implications
Using pooled data collected as part of a statewide quality improvement collaborative, we identified significant racial/ethnic disparities in receipt of any maternal MOUD during pregnancy that remained after we controlled for region, delivery hospital, and maternal characteristics. To ensure equitable care and treatment of all women suffering from OUD, additional research should focus on understanding the perspectives and treatment experiences of minority women and the role that partners and families play in influencing treatment decisions. Statewide collaboratives focused on improving care for families affected by substance misuse can play an important role in identifying hospital and regional variations in care and developing standards and clinical benchmarks to address inequities in treatment receipt. Improved care for all women with OUD must include early screening during pregnancy, linkage to treatment programs, equitable access to medication treatment, and increased supports to retain women in treatment throughout pregnancy and after giving birth.
ACKNOWLEDGMENTS
The Perinatal-Neonatal Quality Improvement Network of Massachusetts thanks the Massachusetts Health Policy Commission for their funding support of the Massachusetts statewide quality improvement initiative to improve care of mothers with opioid use disorder and infants at risk for neonatal abstinence syndrome.
We thank Moira Nolan, BA, for her editorial review of an earlier version of this article.
CONFLICTS OF INTEREST
The authors have no conflicts of interest to disclose.
HUMAN PARTICIPANT PROTECTION
Beth Israel Deaconess Medical Center institutional review board reviewed this project as part of the larger quality improvement effort of the Perinatal-Neonatal Quality Improvement Network of Massachusetts and did not require a separate institutional review board review.
Footnotes
See also Terplan, p. 1728.
REFERENCES
- 1.Centers for Disease Control and Prevention. Prescription painkiller overdoses. September 4, 2018. Available at: https://www.cdc.gov/vitalsigns/prescriptionpainkilleroverdoses/index.html. Accessed June 14, 2019.
- 2.Rudd RA, Aleshire N, Zibbell JE, Gladden RM. Increases in drug and opioid overdose deaths—United States, 2000–2014. MMWR Morb Mortal Wkly Rep. 2016;64(50-51):1378–1382. doi: 10.15585/mmwr.mm6450a3. [DOI] [PubMed] [Google Scholar]
- 3.Centers for Disease Control and Prevention. Vital signs: overdoses of prescription opioid pain relievers and other drugs among women—United States, 1999–2010. MMWR Morb Mortal Wkly Rep. 2013;62(26):537–542. [PMC free article] [PubMed] [Google Scholar]
- 4.Krans EE, Kim JY, James AE, Kelley D, Jarlenski MP. Medication-assisted treatment use among pregnant women with opioid use disorder. Obstet Gynecol. 2019;133(5):943–951. doi: 10.1097/AOG.0000000000003231. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Schiff DM, Nielsen T, Terplan M et al. Fatal and nonfatal overdose among pregnant and postpartum women in Massachusetts. Obstet Gynecol. 2018;132(2):466–474. doi: 10.1097/AOG.0000000000002734. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Alexander MJ, Kiang MV, Barbieri M. Trends in Black and White opioid mortality in the United States, 1979–2015 [erratum in Epidemiology. 2019;30(2):e13]. Epidemiology. 2018;29(5):707–715. [DOI] [PMC free article] [PubMed]
- 7.Lippold KM. Racial/ethnic and age group differences in opioid and synthetic opioid–involved overdose deaths among adults aged ≥18 years in metropolitan areas—United States, 2015–2017. MMWR Morb Mortal Wkly Rep. 2019;68(43):967–973. doi: 10.15585/mmwr.mm6843a3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Guerrero EG, Marsh JC, Duan L, Oh C, Perron B, Lee B. Disparities in completion of substance abuse treatment between and within racial and ethnic groups. Health Serv Res. 2013;48(4):1450–1467. doi: 10.1111/1475-6773.12031. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lagisetty PA, Ross R, Bohnert A, Clay M, Maust DT. Buprenorphine treatment divide by race/ethnicity and payment. JAMA Psychiatry. 2019 doi: 10.1001/jamapsychiatry.2019.0876. Epub ahead of print. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Stahler GJ, Mennis J. Treatment outcome disparities for opioid users: are there racial and ethnic differences in treatment completion across large US metropolitan areas? Drug Alcohol Depend. 2018;190:170–178. doi: 10.1016/j.drugalcdep.2018.06.006. [DOI] [PubMed] [Google Scholar]
- 11.American College of Obstetricians and Gynecologists’ Committee on Obstetric Practice, American Society of Addiction Medicine. Opioid use and opioid use disorder in pregnancy. 2017. Available at: https://www.acog.org/Clinical-Guidance-and-Publications/Committee-Opinions/Committee-on-Obstetric-Practice/Opioid-Use-and-Opioid-Use-Disorder-in-Pregnancy. Accessed May 11, 2019.
- 12.Center for Substance Abuse Treatment. Medication-Assisted Treatment for Opioid Addiction in Opioid Treatment Programs. Rockville, MD: Substance Abuse and Mental Health Services Administration; 2005. Medication-assisted treatment for opioid addiction during pregnancy. [PubMed] [Google Scholar]
- 13.Office of Womens Health. Final report: opioid use, misuse, and overdose in women. US Department of Health and Human Services. 2017:87. Available at: https://www.womenshealth.gov/files/documents/final-report-opioid-508.pdf. Accessed May 14, 2019.
- 14.Wachman EM, Houghton M, Melvin P et al. A quality improvement initiative to implement the Eat, Sleep, Console Neonatal Opioid Withdrawal Syndrome Care Tool in Massachusetts’ PNQIN Collaborative. J Perinatol. 2020 doi: 10.1038/s41372-020-0733-y. Epub ahead of print. [DOI] [PubMed] [Google Scholar]
- 15.Perinatal-Neonatal Quality Improvement Network of Massachusetts. About us. 2018. Available at: https://www.pnqinma.org. Accessed May 14, 2019.
- 16.Massachusetts Department of Public Health. Massachusetts births 2016. 2018:58. Available at: https://www.mass.gov/files/documents/2018/06/01/birth-report-2016.pdf. Accessed June 19, 2019.
- 17.Massachusetts Department of Public Health, MA Executive Office of Health and Human Services Regions. Massachusetts environmental public health tracking. 2019. Available at: https://matracking.ehs.state.ma.us/eohhs_regions/eohhs_regions.html. Accessed June 19, 2019.
- 18.Krans EE, Bogen D, Richardson G, Park SY, Dunn SL, Day N. Factors associated with buprenorphine versus methadone use in pregnancy. Subst Abus. 2016;37(4):550–557. doi: 10.1080/08897077.2016.1146649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Short VL, Hand DJ, MacAfee L, Abatemarco DJ, Terplan M. Trends and disparities in receipt of pharmacotherapy among pregnant women in publically funded treatment programs for opioid use disorder in the United States. J Subst Abuse Treat. 2018;89:67–74. doi: 10.1016/j.jsat.2018.04.003. [DOI] [PubMed] [Google Scholar]
- 20.Schiff DM, Nielsen T, Hoeppner BB et al. Assessment of racial and ethnic disparities in the use of medication to treat opioid use disorder among pregnant women in Massachusetts. JAMA Netw Open. 2020;3(5):e205734. doi: 10.1001/jamanetworkopen.2020.5734. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Bryant AS, Worjoloh A, Caughey AB, Washington AE. Racial/ethnic disparities in obstetric outcomes and care: prevalence and determinants. Am J Obstet Gynecol. 2010;202(4):335–343. doi: 10.1016/j.ajog.2009.10.864. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hoffman KM, Trawalter S, Axt JR, Oliver MN. Racial bias in pain assessment and treatment recommendations, and false beliefs about biological differences between Blacks and Whites. Proc Natl Acad Sci U S A. 2016;113(16):4296–4301. doi: 10.1073/pnas.1516047113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Hansen HB, Siegel CE, Case BG, Bertollo DN, DiRocco D, Galanter M. Variation in use of buprenorphine and methadone treatment by racial, ethnic, and income characteristics of residential social areas in New York City. J Behav Health Serv Res. 2013;40(3):367–377. doi: 10.1007/s11414-013-9341-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hansen H, Netherland J. Is the prescription opioid epidemic a White problem? Am J Public Health. 2016;106(12):2127–2129. doi: 10.2105/AJPH.2016.303483. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Santoro TN, Santoro JD. Racial bias in the US opioid epidemic: a review of the history of systemic bias and implications for care. e3733. Cureus. 2018;10(12) doi: 10.7759/cureus.3733. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ostrach B, Leiner C. “I didn’t want to be on Suboxone at first . . . ”—ambivalence in perinatal substance use treatment. J Addict Med. 2019;13(4):264–271. doi: 10.1097/ADM.0000000000000491. [DOI] [PubMed] [Google Scholar]
- 27.Stone R. Pregnant women and substance use: fear, stigma, and barriers to care. Health Justice. 2015;3(1):2. doi: 10.1186/s40352-015-0015-5. [DOI] [Google Scholar]
- 28.Parikh A, Gopalakrishnan M, Azeem A, Booth A, El-Metwally D. Racial association and pharmacotherapy in neonatal opioid withdrawal syndrome. J Perinatol. 2019;39(10):1370–1376. doi: 10.1038/s41372-019-0440-8. [DOI] [PubMed] [Google Scholar]
- 29.Charles MK, Cooper WO, Jansson LM, Dudley J, Slaughter JC, Patrick SW. Male sex associated with increased risk of neonatal abstinence syndrome. Hosp Pediatr. 2017;7(6):328–334. doi: 10.1542/hpeds.2016-0218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.McQueen K, Murphy-Oikonen J. Neonatal abstinence syndrome. N Engl J Med. 2016;375(25):2468–2479. doi: 10.1056/NEJMra1600879. [DOI] [PubMed] [Google Scholar]
- 31.Patrick SW, Davis MM, Lehman CU, Cooper WO. Increasing incidence and geographic distribution of neonatal abstinence syndrome: United States 2009–2012. J Perinatol. 2015;35(8):650–655. doi: 10.1038/jp.2015.36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Tolia VN, Patrick SW, Bennett MM et al. Increasing incidence of the neonatal abstinence syndrome in US neonatal ICUs. N Engl J Med. 2015;372(22):2118–2126. doi: 10.1056/NEJMsa1500439. [DOI] [PubMed] [Google Scholar]
- 33.Walsh MC, Crowley M, Wexelblatt S et al. Ohio Perinatal Quality Collaborative improves care of neonatal narcotic abstinence syndrome. Pediatrics. 2018;141(4):e20170900. doi: 10.1542/peds.2017-0900. [DOI] [PubMed] [Google Scholar]
- 34.Patrick SW, Schumacher RE, Benneyworth BD, Krans EE, McAllister JM, Davis MM. Neonatal abstinence syndrome and associated health care expenditures: United States, 2000–2009. JAMA. 2012;307(18):1934–1940. doi: 10.1001/jama.2012.3951. [DOI] [PubMed] [Google Scholar]
- 35.Wachman EM, Hayes MJ, Brown MS et al. Association of OPRM1 and COMT single-nucleotide polymorphisms with hospital length of stay and treatment of neonatal abstinence syndrome. JAMA. 2013;309(17):1821–1827. doi: 10.1001/jama.2013.3411. [DOI] [PMC free article] [PubMed] [Google Scholar]