Abstract
Objective:
To estimate treatment and postpartum health care utilization among pregnant persons with opioid use disorder (OUD) in Vermont and Maine.
Methods:
Vermont’s and Maine’s All Payer Claims Databases were used to identify deliveries 2010—2018 that were paid for, in part, by Medicaid. OUD was identified among pregnant persons if they had any claim with an OUD-diagnosis code (ICD-9/10) or medication for addiction treatment (MAT) code during the 5 months prior to delivery event. Consistent and inconsistent MAT were compared to no MAT on the rate of hospitalizations and emergency department (ED) visits in the first 12 months’ postpartum using negative binomial regression.
Results:
From 2010 through 2018, 27,652 deliveries in Vermont and 43,480 deliveries in Maine were among persons insured by Medicaid. The prevalence of OUD among pregnant persons increased from 6.7% to 11.6% in Vermont and from 7.4% to 11.0% in Maine. Among pregnant persons with OUD in 2018, 57% had consistent MAT in Vermont and 50% had consistent MAT in Maine; approximately 32% and 27% were not in treatment in Vermont and Maine, respectively. In Maine, consistent MAT was associated with a 47% lower rate of hospitalization and 37% to 46% lower rates of ED visits when compared to those without MAT; in Vermont, those with consistent buprenorphine treatment had a 30% lower rate of ED visits.
Conclusions:
Medicaid data from Vermont and Maine suggests that medication for addiction treatment for opioid use disorder during pregnancy reduces emergency health care utilization in the first year postpartum.
Keywords: opioid, pregnancy, substance use disorder, medication for addiction treatment, rural, Maine, Vermont
Introduction
Opioid use disorder (OUD) during pregnancy is associated with higher risks of maternal and newborn morbidity and mortality,1 as well as substantial healthcare costs,2,3 and has increased sharply in recent years among rural residents in the US.4,5 Vermont and Maine had the two highest average annual increases in OUD per 1,000 delivery hospitalizations during the past decade, with each state experiencing more than a 40-fold increase since 1999.1,6 In Maine, the prevalence of maternal OUD at delivery hospitalization was 34.9 per 1,000 deliveries (nearly 3.5%) in 2018.7 The treatment for OUD during pregnancy and its effect on postpartum healthcare utilization has not been studied in these states, where more than 60% of the population lives in a rural area.8
Although methadone has historically been the primary treatment for pregnant persons with OUD, buprenorphine has also been a first-line pharmacotherapy treatment since at least 2012.9–12 Consistent OUD treatment has been found to decrease the use of illicit opioids during pregnancy and the risk of opioid overdose,11,13,14 but there is limited information on its effect on emergency health care utilization postpartum.
We sought to estimate the prevalence of medication for addiction treatment among pregnant and postpartum persons with OUD in Vermont and Maine, as well as the association between treatment type during pregnancy and emergency healthcare utilization postpartum. Results from our study can be used to inform and refine interventions designed to improve health outcomes for pregnant and postpartum persons with OUD in states with a majority rural population.
Methods
Data source
We identified delivery events in Vermont and Maine from 2010 to 2018 using data from each state’s All Payer Claims Database (APCD). These databases are repositories of health care claims data for state residents with health insurance, and are maintained by the Green Mountain Care Board and the Maine Health Data Organization. Both APCDs include eligibility, medical, dental, and pharmacy data from commercial payers, Medicare, Medicaid, and select self-funded/third party administrators.15
Claims containing substance use disorder (SUD) diagnoses have been redacted from the Maine APCD since June 2017, in response to federal regulations released by the US Substance Abuse and Mental Health Services Administration.16,17 These redactions remain in place despite rule changes that now permit the Centers for Medicare and Medicaid Services (CMS) to include SUD claims in their research identifiable files.18 These redactions led us to restrict our analysis to Medicaid-covered pregnant persons, as SUD claims data from public payers in Maine was fully recoverable while commercial payer data was not. Vermont did not perform similar redactions; however, we restricted our analysis to Medicaid-covered pregnancies in Vermont for consistency. National data suggest that approximately 80% of pregnant persons with OUD are insured by Medicaid.2 The University of Southern Maine’s and University of Vermont’s School of Medicine institutional review boards approved our study protocol.
Deliveries
Deliveries were identified using delivery-related ICD-9/10 diagnosis and procedure codes, as well as Current Procedural Terminology (CPT) codes documented on professional, inpatient, and outpatient medical claims. We used delivery code lists published by the Alliance for Innovation in Maternal Health, which included stillbirths and livebirths.19 We identified delivery episodes for each reproductive age female enrollee (13 to 49 years old). For persons who had delivery episodes 9 or fewer months apart, we excluded the subsequent delivery episode because this could have been an erroneous delivery record. For each delivery episode, we identified maternal age at time of delivery, delivery type (vaginal vs. c-section), rurality, and delivery hospital level of care. We classified rurality based on the zip code of maternal residence at the time of delivery using the 2010 United States Department of Agriculture Rural-Urban Commuting Area codes.20 Delivery hospital level of care and type were categorized using publicly available sources.21,22
Study population
There were 44,121 deliveries in Vermont and 94,823 deliveries in Maine identified. After excluding deliveries not covered by Medicaid and those with delivery dates too close together, 29,481 and 53,426 deliveries remained in the analytic cohorts, respectively (Appendix Figure 1). For analysis of OUD prevalence and treatment during pregnancy, we restricted the analysis to deliveries among persons who were enrolled in Medicaid for at least 5 months prior to delivery in order to ensure we had the full record of services used by these persons during the majority of their pregnancy (27,652 Vermont; 43,480 Maine). For analyses of postpartum outcomes (e.g. hospitalizations, emergency department [ED] visits, medication for addiction treatment [MAT] postpartum) we further restricted the analysis to deliveries among persons with OUD during pregnancy who were enrolled in Medicaid for at least 12 months following delivery without coverage gaps (2,556 Vermont; 3,666 Maine).
Opioid use disorder
We determined OUD prevalence among pregnant persons using International Classification of Diseases, Clinical Modification (ICD-CM) diagnosis codes along with evidence of MAT. As our analysis spanned the ICD-9/10 transition (October 1, 2015), we used a code list published by researchers at the Agency for Healthcare Research and Quality for ICD-9/10 diagnosis codes related to opioid use, abuse, dependence, adverse effects, or poisoning.23 In addition, we included diagnosis codes for OUD “in remission” in order to comprehensively capture in utero opioid exposure.7 We also used treatment with MAT (see below) as evidence of an OUD diagnosis. If there was any OUD diagnosis, or MAT within the 5 months prior to delivery, a person was considered to have OUD during pregnancy.
Medication for addiction treatment
We identified persons receiving MAT using medical and prescription claims. We used a procedure code (HCPCS code H0020) on medical claims to identify methadone treatment.24,25 For buprenorphine, we used a published list of buprenorphine national drug codes,25 omitting codes for patch use, which is only approved for analgesic treatment. In addition, we searched claims for procedure codes for buprenorphine implants and injections.26 To comprehensively capture all MAT types, we also used code lists compiled by the Medicaid Outcomes Distributed Research Network to scan procedure and drug codes for the use of naltrexone,24 even though this medication is not currently recommended for use in pregnancy in the US.27 MAT type (yes/no for each methadone, buprenorphine, or naltrexone) was assessed for each of the 5 months leading up to delivery, and for each month up to 12 months’ postpartum. We categorized persons in terms of MAT during pregnancy as follows: no evidence of treatment for any of the 5 months leading up to delivery (“no treatment”); buprenorphine use at delivery and consistent MAT for each of the 5 months leading up to delivery (“consistent buprenorphine”); methadone use at delivery and consistent MAT for each of the 5 months leading up to delivery (“consistent methadone”); and all other use patterns (“inconsistent treatment”). If a person met criteria for both consistent buprenorphine and consistent methadone treatment, the person was categorized as having consistent buprenorphine treatment in order to have mutually exclusive treatment groups (this occurred fewer than 10 times in each state).
Co-occurring mental health conditions
We identified co-occurring mental health conditions by evaluating prescription claims for benzodiazepines and antidepressants in the 5 months prior to delivery. We used National Drug Code lists from a previously published analysis.25 Co-occurring mental health conditions is a recognized confounder of the association between MAT and pregnancy outcomes and prescriptions for these conditions were used as proxy measures for clinical diagnoses.28
Postpartum hospitalizations and emergency department visits
Unique hospitalizations and ED visit episodes for the first 12 months’ postpartum were identified for each person. We first captured all claims related to ED visits using the Yale definition,29 which uses bill type, facility type, revenue codes, and procedure codes. All ED and inpatient hospitalization claims data were sorted by member ID and service dates, and unique episodes were defined if the end and start service dates were more than one day apart. If claims data indicated an inpatient hospitalization and ED visit occurred during the same episode, we categorized that event as a hospitalization. Among emergency episodes, we identified those related to opioid-related drug poisoning using a subset of the OUD codes.23
Statistical analysis
We estimated the prevalence of OUD during pregnancy within Medicaid enrollees by state and delivery year. We then tabulated characteristics among deliveries to persons with and without OUD during pregnancy. Among persons with OUD, we estimated the prevalence of MAT during pregnancy by delivery year. We also estimated the proportion of persons with OUD who were using MAT by month during the 5 months leading up to delivery and through 12 months’ postpartum.
We used negative binomial regression models, with an offset term for treatment group denominator, to estimate rate ratios for hospitalizations, ED visits, and opioid overdose visits within the first 12 months’ postpartum. These models compared the rate of events among persons with consistent buprenorphine, consistent methadone, and inconsistent treatment to those with no treatment. We selected negative binomial models over Poisson models due to the high number of zero events observed. Models were adjusted for: maternal age at delivery, rurality of residence, delivery type, hospital level of care, and prescriptions for benzodiazepines and antidepressants, separately. For the postpartum analysis, we excluded persons with an interpregnancy interval <12 months in order to not capture postpartum hospitalizations related to a new delivery event.
Sensitivity analyses
To assess the accuracy of our delivery episode identification methodology, we compared the number of deliveries paid for by Medicaid by state and year with other available estimates. To ascertain how sensitive our findings were to the code lists we selected, we estimated OUD and MAT prevalence using separate code lists compiled by the Medicaid Outcomes Distributed Research Network.24
All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC) and SQL. Analyses were conducted separately for each state; at no point were data from the two states combined.
Results
We included 29,481 and 53,426 deliveries among persons enrolled in Medicaid in Vermont and Maine, respectively (Appendix Figure 1). Approximately 94% and 81% of these deliveries, respectively, were among persons enrolled in Medicaid for at least 5 months leading up to the delivery month, and, of those persons, 86% and 65% were enrolled through at least 12 months’ postpartum. Medicaid enrollment patterns differed by delivery year for Maine, with a notable decrease in pregnancy and postpartum coverage beginning in 2013 (Appendix Tables 1, 2).
From 2010 through 2018, the prevalence of OUD during pregnancy increased from 6.7% to 11.6% in Vermont, and from 7.4% to 11.0% in Maine (Figure 1). In both states, a higher proportion of persons with OUD during pregnancy (2,769 in Vermont; 4,461 in Maine) had c-sections, delivered at a hospital with a level III Neonatal Intensive Care Unit, and had benzodiazepines and antidepressants prescribed during pregnancy, as compared to persons without OUD (Table 1).
Figure 1.
Opioid use disorder among pregnant persons with deliveries paid for (in part) by Medicaid ending in 2010–2018 in Vermont and Maine: All Payer Claims Databases
(Vermont n= 27,652; Maine n=43,480)
Table 1.
Characteristics of persons with deliveries paid for (in part) by Medicaid ending in 2010–2018 in Vermont and Maine, by opioid use disorder during pregnancy: Vermont and Maine All Payer Claims Databases
| Vermont | Maine | |||
|---|---|---|---|---|
| Characteristic | Opioid use disorder (n=2,769) | No opioid use disorder (n=24,883) |
Opioid use disorder (n=4,461) | No opioid use disorder (n=39,019) |
| n (%) | N (%) | n (%) | n (%) | |
| Age at delivery1 | ||||
| 15–19 | 93 (3.4) | 2018 (8.1) | 81 (1.8) | 3565 (9.1) |
| 20–24 | 765 (27.6) | 6934 (27.9) | 1014 (22.7) | 12843 (32.9) |
| 25–29 | 1153 (41.6) | 7400 (29.7) | 1772 (39.7) | 11706 (30.0) |
| 30–34 | 576 (20.8) | 5484 (22.0) | 1195 (26.8) | 7053 (18.1) |
| 35+ | 181 (6.5) | 3028 (12.2) | 399 (8.9) | 3831 (9.8) |
| Delivery mode2 | ||||
| Vaginal | 1897 (68.5) | 18089 (72.7) | 2781 (62.3) | 25488 (65.3) |
| Cesarean | 824 (29.8) | 6022 (24.2) | 1394 (31.3) | 11132 (28.5) |
| Pregnancy number | ||||
| One | 1172 (62.5) | 12157 (67.9) | 2863 (64.2) | 27772 (71.2) |
| Two | 537 (28.6) | 4737 (26.4) | 1228 (27.5) | 8935 (22.9) |
| Three or more | 166 (8.9) | 1023 (5.7) | 370 (8.3) | 439 (5.9) |
| Rural-urban maternal residence1,3 | ||||
| All metro | 524 (18.9) | 4042 (16.2) | 1512 (33.9) | 12378 (31.7) |
| Large rural | 1004 (36.3) | 7428 (29.9) | 1441 (32.3) | 13148 (33.7) |
| Small rural | 1092 (39.4) | 11223 (45.1) | 1242 (27.8) | 10734 (27.5) |
| Isolated rural | 148 (5.3) | 2182 (8.8) | 225 (5.0) | 2327 (6.0) |
| Other/out of state | Suppressed | Suppressed | 32 (0.7) | 333 (0.9) |
| Delivery hospital level of care and type | ||||
| Hospitals with level III NICU | 1063 (38.4) | 7666 (30.8) | 2052 (46.0) | 10755 (27.6) |
| Hospitals with level II specialty care | Suppressed | 26 (0.1) | 773 (17.3) | 5858 (15.0) |
| Hospitals with level 1 care, not critical access | 826 (29.8) | 6017 (24.2) | 1003 (22.5) | 15412 (39.5) |
| Hospitals with level 1 care, critical access | 695 (25.1) | 9755 (39.2) | 534 (12.0) | 5963 (15.3) |
| Unclassified4 | 184 (6.6) | 1419 (5.7) | 99 (2.2) | 1031 (2.6) |
| Any benzodiazepine prescription in pregnancy | ||||
| Yes | 129 (4.7) | 409 (1.6) | 472 (10.6) | 940 (2.4) |
| Any antidepressant prescription in pregnancy | ||||
| Yes | 450 (16.3) | 2086 (8.4) | 754 (16.9) | 3948 (10.1) |
Suppressed cells (n=<10).
Observations were missing data for the following: age at delivery (n=41), rural-urban maternal residence (n=108).
Based on CPT codes only. Delivery mode unknown for 2685/43480 deliveries in Maine; 820/27652 deliveries in Vermont.
Rural-urban residence based on rural-urban community area codes (RUCA) a census-tract based classification system. Zip codes were mapped to census-tracts.
Hospitals were not classified due to disclosure concerns because fewer than 5 births occurred at each facility per year. For Vermont births, those occurring outside of Vermont and the most frequent hospitals in NH are unclassified.
The prevalence of MAT during pregnancy showed similar patterns in Vermont and Maine, with some key differences (Figure 2). Consistent buprenorphine use decreased in Vermont from 55% to 35% and increased in Maine from 23% to 41%. Consistent methadone treatment increased in Vermont from 11% to 21% and decreased from 19% to 9% in Maine. Approximately, 32% and 27% of pregnant persons with OUD in 2018 were not in treatment in each state, and the remainder were in inconsistent treatment. For both states, fewer than 10 deliveries had evidence of naltrexone treatment during pregnancy.
Figure 2.
Medication for addiction treatment during pregnancy among persons with opioid use disorder with deliveries paid for (in part) by Medicaid ending in 2010–2018 in Vermont and Maine: All Payer Claims Databases (Vermont n= 2,769; Maine n=4,461)
The use of MAT in pregnant and postpartum persons with OUD was similar between the states but varied across the years. Approximately 10% to 30% of persons were treated with methadone per month during both perinatal periods; the year(s) with the lowest methadone treatment by month was 2010/2011 for Vermont and 2018 for Maine (Figure 3 top). For buprenorphine, approximately 35% to 70% of persons were treated each month during pregnancy, and 40% to 60% were treated each month up to 12 months’ postpartum; the year with the highest buprenorphine treatment by month was 2010 for Vermont and 2018 for Maine (Figure 3 bottom). For buprenorphine, but not for methadone, there appeared to be a drop off in treatment in the first 3 months’ postpartum.
Figure 3.
Medication for addiction treatment during pregnancy and postpartum, by month, among persons with opioid use disorder with deliveries paid for (in part) by Medicaid ending in 2010–2018 in Vermont and Maine: All Payer Claims Databases (Vermont n= 2,556; Maine n=3,666)
Pregnant persons with OUD who had consistent treatment with buprenorphine or methadone had approximately 47% lower adjusted rates of hospitalizations in Maine compared to persons not in treatment, but no significantly lower rate was observed in Vermont (Table 2). Inconsistent treatment was not significantly associated with lower rates of hospitalizations for either state.
Table 2.
Hospitalization and emergency department visit rate ratios by medication for addiction treatment during pregnancy among persons with opioid use disorder with deliveries paid for (in part) by Medicaid ending in 2010–2018 in Vermont and Maine: All Payer Claims Databases (Vermont n= 2,506; Maine n=3,460)
| Outcome | State | Treatment category | N | At least one event, % | Rate per 100 deliveries | Unadjusted rate ratio | 95% CI | Adjusted rate ratio1 | 95% CI |
|---|---|---|---|---|---|---|---|---|---|
| Hospitalization | Maine | ||||||||
| Consistent buprenorphine | 1362 | 5.9 | 7 | 0.48 | 0.33, 0.72 | 0.53 | 0.35, 0.79 | ||
| Consistent methadone | 648 | 4.8 | 8 | 0.53 | 0.33, 0.84 | 0.53 | 0.33, 0.84 | ||
| Inconsistent treatment | 693 | 8.8 | 12 | 0.80 | 0.52, 1.22 | 0.88 | 0.58, 1.34 | ||
| No treatment | 757 | 7.3 | 14 | Referent | Referent | ||||
| Vermont | |||||||||
| Consistent buprenorphine | 1077 | 9.2 | 11 | 0.80 | 0.56, 1.15 | 0.78 | 0.54, 1.13 | ||
| Consistent methadone | 398 | 9.3 | 13 | 0.92 | 0.59, 1.42 | 0.98 | 0.63, 1.52 | ||
| Inconsistent treatment | 410 | 8.5 | 10 | 0.75 | 0.48, 1.18 | 0.72 | 0.46, 1.14 | ||
| No treatment | 621 | 9.7 | 13 | Referent | Referent | ||||
| Emergency department visit | Maine | ||||||||
| Consistent buprenorphine | 1362 | 45.7 | 90 | 0.51 | 0.44, 0.59 | 0.54 | 0.47, 0.63 | ||
| Consistent methadone | 648 | 49.9 | 108 | 0.59 | 0.50, 0.70 | 0.63 | 0.53, 0.75 | ||
| Inconsistent treatment | 693 | 48.9 | 121 | 0.67 | 0.57, 0.79 | 0.70 | 0.60, 0.82 | ||
| No treatment | 757 | 55.0 | 175 | Referent | Referent | ||||
| Vermont | |||||||||
| Consistent buprenorphine | 1077 | 45.9 | 79 | 0.67 | 0.58, 0.78 | 0.70 | 0.60, 0.81 | ||
| Consistent methadone | 398 | 48.7 | 89 | 0.77 | 0.64, 0.92 | 0.84 | 0.70, 1.00 | ||
| Inconsistent treatment | 410 | 52.2 | 101 | 0.87 | 0.73, 1.04 | 0.91 | 0.76, 1.08 | ||
| No treatment | 621 | 54.4 | 117 | Referent | Referent |
Models were adjusted for age at time of delivery, rurality of residence, hospital level of care, benzodiazepine prescription during pregnancy, antidepressant prescription during pregnancy, and delivery type.
For ED visits in Maine, pregnant persons with OUD who had consistent MAT and those in inconsistent treatment had approximately 30% to 46% lower adjusted rates when compared to pregnant persons not in treatment (Table 2). In Vermont, consistent MAT with buprenorphine was associated with a 30% lower adjusted rate compared to persons not in treatment; consistent MAT with methadone was associated with a 16% lower rate, but this finding was not statistically significant.
The number of claims-identifiable opioid overdose visits in each state was less than 10 per treatment group, even when pooling across all years. Therefore, we were unable to show overdose visit rates by treatment group.
Our sensitivity analyses using alternative code lists resulted in nearly the same prevalence estimates for OUD, buprenorphine treatment, and prescription for benzodiazepines as compared to our primary analysis. Overall estimates of the number of deliveries in Vermont and Maine were somewhat higher than published reports, but these discrepancies can be explained by our use of a more expansive definition of Medicaid coverage; different code lists and delivery episode algorithm; and less stringent residency requirements (Appendix Table 3).
Discussion
From 2010 to 2018, the prevalence of Medicaid-enrolled pregnant persons with OUD increased from 6.7% to 11.6% in Vermont and from 7.4% to 11.0% in Maine. While a greater percentage of persons in Vermont received buprenorphine MAT during all study years, this shifted from predominantly buprenorphine in earlier years to more persons receiving methadone MAT in later years. Maine saw an increase in buprenorphine MAT and a decrease in methadone MAT during the study period. In 2018, approximately one half of pregnant persons with OUD had consistent MAT in each state, one third were not on MAT, and the remainder were in inconsistent treatment. In Maine, our findings suggest consistent MAT reduced the rate of hospitalization by half, and reduced the rate of ED visits by more than one third when compared to persons not in treatment; inconsistent MAT reduced the rate of ED visits by around one third. In Vermont, consistent buprenorphine MAT reduced ED visits by around one third when compared to persons not in treatment. The implications of reduced hospitalizations and ED visits postpartum is that there are fewer events requiring this level of emergency care, indicating more stable health status postpartum.
Our study estimates are higher than what has previously been reported on the prevalence of OUD among Medicaid-enrolled persons, but are generally consistent with trends in OUD treatment during pregnancy in other states. We report higher prevalence of OUD during pregnancy (7% to 11%) than an analysis of Medicaid-enrolled persons in three states, which found that 2.2% had OUD in the year before and after delivery in 2013–2016.30 Similarly, an analysis of Medicaid-enrolled pregnant persons in Pennsylvania found that 3.5% had an OUD diagnosis in 2009–2015.31 Regarding trends in treatment, in Pennsylvania, the use of buprenorphine during pregnancy increased from 15% to 30% during the same time period, while the use of methadone decreased from 35% to 25%.32 In Massachusetts, 64% of pregnant persons with OUD (publicly and privately insured) in 2012–2014 had evidence of pharmacotherapy in the year before delivery.14 These treatment prevalence estimates are generally consistent with our finding that the majority of persons with OUD in Maine and Vermont are now receiving MAT during pregnancy. Maternal treatment information is not currently captured in state-wide data for newborns with neonatal abstinence syndrome in Maine and Vermont; therefore, our findings can’t be compared to maternal treatment trends observed among newborns in these states.
In terms of the rates of hospitalizations postpartum and the effect of OUD treatment on reducing these rates, our study findings also generally align with prior work. The number of hospitalizations we observed (6% to 10% with at least 1 hospitalization) are higher than the risks for 1-year rehospitalization reported for postpartum persons in the general obstetric population (approximately 3%),33,34 but not as high as all-cause 90-day rehospitalization risk among persons with non-fatal opioid overdose in the US (24%).35 In the study of Medicaid-enrolled persons in Pennsylvania, persons who initiated treatment early in pregnancy, and remained adherent, had 20% to 55% lower risk of hospitalizations during pregnancy than persons who initiated buprenorphine later in pregnancy and those with declining adherence or early discontinuation.32 In Massachusetts, MAT during pregnancy and postpartum reduced opioid-related overdose event rates (deaths, ED visits, hospitalizations, and ambulance incidents) during the first year postpartum by 19% to 88% (depending on the postpartum time period).14 These studies didn’t present results for overall hospitalization rates in the first year postpartum, so could not be directly compared to our findings, but generally support a protective effect of MAT on hospitalizations.
Our observed ED visit rates are also in line with previous work. We report rates between 79 and 175 per 100 deliveries (46% to 55% with at least 1 ED visit), with a protective effect of consistent MAT, particularly buprenorphine. Previous studies have reported the proportion of patients with OUD who had at least one ED visit was approximately 40% for patients age 18–35 (over a 12 month period),36 14% for the general obstetric population (within the first 2 years postpartum),33 and 21% for female adults in the US (over a 12 month period),37 which are all lower than estimates we report. Consistent with our study findings, the Pennsylvania study found persons who initiated treatment early in pregnancy, and remained adherent, had the lowest risk of all-cause ED visits/hospitalizations between 85 days and 1 year post-delivery compared to other treatment patterns (27% vs. 29%-37%).32
We found claims-identifiable opioid overdose visits in each state were fewer than 10 per treatment group (<1%). This is less than would be expected in a 12 month follow-up for non-pregnant persons with OUD using claim-based data (1.1% to 6.4%),38 and could be due to lower rates of these events in postpartum persons, or due to lower capture of overdose events in claims-based data in Maine and Vermont.
The differences between Vermont and Maine in terms of Medicaid coverage during pregnancy and postpartum can be explained by policy differences. While both states currently have Medicaid eligibility income limits for pregnant persons of around 213% of the federal poverty level (FPL),39 Maine’s eligibility cuts to the Medicaid program in 2014 likely resulted in the decreased coverage after 2 months’ postpartum that we observed beginning with 2013 deliveries. Federal rules require Medicaid to cover the first 6 weeks’ postpartum, but income limits apply after that point and Maine reduced these income limits from 200% to 105% FPL in January 2014.39 On the other hand, Vermont expanded Medicaid under the Affordable Care Act in 2014, resulting in increased in coverage postpartum.
In terms of MAT trends, Vermont had a higher prevalence of consistent buprenorphine treatment in the early years of our analysis (2010–2012), which could be due to prioritizing provision of MAT to pregnant persons during these years.40 In Maine, use of buprenorphine among pregnant persons with OUD was lower than in Vermont from 2010 through 2013, but increased to Vermont’s level by 2014 as more programs began to offer integrated prenatal and substance use treatment. The increase of methadone use in Vermont during the study period may be due to the implementation of its Hub and Spoke MAT model, begun in 2012, which integrated care at over 75 local spokes (providing both methadone and buprenorphine treatment) with 9 regional hubs with expertise in treating OUD.40
Our study has several limitations. Primarily, our analysis had to rely on Medicaid claims data aggregated at the delivery level because SUD-related claims have been redacted from the Maine APCD since June 2017, in response to federal regulations.16 This redaction prohibited use of commercial claims in our analysis, and meant that we could not use claim line level data for Maine, e.g. we could not examine and report on the distribution of diagnosis codes to characterize the ED visits or hospitalizations. In addition, we only examined pharmacotherapy treatment of OUD during pregnancy, and not any other forms of treatment, like behavioral health treatment. We were also unable to link infant claims with delivery claims, precluding the analysis of the effect of treatment for OUD during pregnancy on infant health outcomes and health care use.30 By limiting our analytic data set to persons with 5 months of Medicaid coverage during pregnancy, persons presenting late to care, as is common for persons with OUD, were not included in our analysis. Additionally, residual confounding could have accounted for our finding of a protective effect of treatment on emergency health care utilization rates. The relatively small sample size from Vermont could have reduced our power to detect effects, particularly for consistent methadone treatment and ED visits, for which the effect estimate was protective but not statistically significant.
Conclusions
In summary, Medicaid data from Vermont and Maine suggest that the use of MAT during pregnancy reduces emergency health care utilization in the first year postpartum. Efforts should be made to identify barriers and expand access to MAT for pregnant persons with OUD in these two states with majority rural populations.
Supplementary Material
Appendix Figure 1. Flow chart showing total number of deliveries paid for (in part) by Medicaid ending in 2010–2018 in Vermont and Maine, according to enrollment status in Medicaid during pregnancy and the first year postpartum
Acknowledgments:
We thank Kevin Rogers, AA, Senior Business Systems Analyst; and Kristin Battis, MPH, Research Associate, at Human Services Research Institute for performing analyses of the Maine All Payer Claims Data according to our data request specifications. We also thank Karynlee Harrington, BS, Executive Director of the Maine Health Data Organization, for her assistance facilitating this data request; the Maine Health Data Organization is responsible for the State of Maine’s All Payer Claims Data. This work was made possible by the establishment of the MODRN. The authors appreciate the contributions of the MODRN Founding Members, the Academy Health State University Partnership Learning Network, the Medicaid Medical Directors Network, and NIDA 1 R01 DA048029-01.
Funding: The research reported here was supported by grant U54 GM115516 from the National Institutes of Health for the Northern New England Clinical and Translational Research network. KAA is supported by a faculty development award from the Maine Economic Improvement Fund.
Appendix Table 1:
The total number of deliveries paid for (in part) by Medicaid ending in 2010–2018 in Maine by year, according to enrollment status in Medicaid during pregnancy and the first year postpartum
| Among those Medicaid enrolled -5 months through delivery | ||||||
|---|---|---|---|---|---|---|
| Delivery Year | Not Medicaid enrolled -5 months through delivery | Medicaid enrolled -5 months through delivery | Not enrolled in Medicaid through first 2 months’ postpartum | Enrolled through first 2 months, but not first 3 months’ postpartum | Other patterns | Enrolled through first 12 months’ postpartum |
| 2010 | 13.1 | 86.9 | 2.2 | 4.9 | 10.1 | 82.7 |
| 2011 | 9.5 | 90.5 | 2.1 | 5.3 | 12.4 | 80.2 |
| 2012 | 11.0 | 89.0 | 2.7 | 6.5 | 15.9 | 74.9 |
| 2013 | 15.8 | 84.2 | 4.4 | 11.2 | 16.4 | 68.0 |
| 2014 | 19.3 | 80.7 | 4.6 | 27.4 | 15.1 | 53.0 |
| 2015 | 26.8 | 73.2 | 4.6 | 27.0 | 14.6 | 53.9 |
| 2016 | 26.0 | 74.0 | 4.7 | 27.8 | 13.8 | 53.8 |
| 2017 | 25.8 | 74.2 | 4.0 | 35.0 | 13.6 | 47.4 |
| 2018 | 25.3 | 74.8 | 4.2 | 34.4 | 13.0 | 48.5 |
| Total | 9,946 | 43,480 | 1,549 | 7,830 | 6,012 | 28,089 |
Appendix Table 2:
The total number of deliveries paid for (in part) by Medicaid ending in 2010–2018 in Vermont by year, according to enrollment status in Medicaid during pregnancy and the first year postpartum
| Among those Medicaid enrolled -5 months through delivery | ||||||
|---|---|---|---|---|---|---|
| Delivery Year | Not Medicaid enrolled -5 months through delivery | Medicaid enrolled -5 months through delivery | Not enrolled in Medicaid through first 2 months’ postpartum | Enrolled through first 2 months, but not first 3 months’ postpartum | Other patterns | Enrolled through first 12 months’ postpartum |
| 2010 | 7.92 | 92.1 | 1.1 | 3.9 | 12.9 | 82.1 |
| 2011 | 6.6 | 93.5 | 0.9 | 4.2 | 11.7 | 83.1 |
| 2012 | 5.6 | 94.4 | 0.9 | 4.7 | 12.9 | 82.4 |
| 2013 | 6.0 | 94.0 | 1.2 | 3.7 | 9.6 | 85.7 |
| 2014 | 6.3 | 93.7 | 0.7 | 1.2 | 3.6 | 94.6 |
| 2015 | 4.8 | 95.2 | 0.7 | 0.7 | 10.5 | 88.2 |
| 2016 | 4.5 | 95.5 | 1.5 | 1.5 | 10.9 | 86.2 |
| 2017 | 7.3 | 92.7 | 0.9 | 0.9 | 11.7 | 86.6 |
| 2018 | 6.9 | 93.1 | 1.5 | 1.3 | 15.8 | 81.5 |
| Total | 1,829 | 27,652 | 279 | 699 | 2,977 | 23,697 |
Appendix Table 3.
Total number of deliveries each calendar year, by state, and paid for in part by Medicaid funds: Vermont and Maine All Payer Claims Database vs. other reports, 2010–2018
| Vermont | Maine | |||
|---|---|---|---|---|
| Calendar year | APCD1 | Medicaid reports2 | APCD1 | Medicaid reports3 |
| 2010 | 3,473 | 2,806 | 6,588 | Not reported |
| 2011 | 3,519 | 2,729 | 6,564 | Not reported |
| 2012 | 3,482 | 2,735 | 6,469 | Not reported |
| 2013 | 3,529 | 2,727 | 6,276 | 5,764 |
| 2014 | 3,590 | 2,812 | 5,932 | 5,427 |
| 2015 | 3,407 | 2,877 | 5,669 | 5,327 |
| 2016 | 2,887 | 2,668 | 5,367 | 5,073 |
| 2017 | 2,879 | 2,629 | 5,313 | 4,959 |
| 2018 | 2,715 | 2,492 | 5,248 | |
| Total | 29,481 | 24,475 | 53,428 |
APCD= All Payer Claims Database
APCD data for deliveries was defined as: unique member IDs and delivery episodes (identified using current procedure terminology [CPT] codes, ICD-9 procedure and diagnosis codes, and ICD-10 procedure and diagnosis codes with service dates no more than 60 days apart) for females between the age of 13 and 49. If the difference between first service dates for a member was between 0 and 9 months (inclusive), the subsequent episode was excluded from analysis. Each delivery had at least one Medicaid claim.
For Vermont: Medicaid reported deliveries from Medicaid institutional claims. The Vermont APCD includes persons who are residents, but not citizens of Vermont, in addition to those receiving supplemental Medicaid coverage, while institutional claims are limited to only citizens of Vermont.
For Maine: Medicaid reported deliveries from “Before the First Tooth” report, which tallied all deliveries among Medicaid members by year for 2013–2017, identified as having both ICD diagnosis codes and CPT codes for either the same claim number or same service date.
Footnotes
Conflicts of interest: None.
Disclosures: All analyses, conclusions, and recommendations drawn from Vermont Health Care Uniform Reporting and Evaluation System (VHCURES) are solely those of the authors and not necessarily those of the Green Mountain Care Board.
References
- 1.Maeda A, Bateman BT, Clancy CR, Creanga AA, Leffert LR. Opioid abuse and dependence during pregnancy: temporal trends and obstetrical outcomes. Anesthesiology. 2014;121(6):1158–1165. [DOI] [PubMed] [Google Scholar]
- 2.Winkelman TNA, Villapiano N, Kozhimannil KB, Davis MM, Patrick SW. Incidence and Costs of Neonatal Abstinence Syndrome Among Infants With Medicaid: 2004–2014. Pediatrics. 2018;141(4). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Patrick SW, Davis MM, Lehman CU, Cooper WO. Increasing incidence and geographic distribution of neonatal abstinence syndrome: United States 2009 to 2012. J Perinatol. 2015;35:650.+. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kozhimannil KB, Chantarat T, Ecklund AM, Henning-Smith C, Jones C. Maternal Opioid Use Disorder and Neonatal Abstinence Syndrome Among Rural US Residents, 2007–2014. The Journal of rural health : official journal of the American Rural Health Association and the National Rural Health Care Association. 2018. [Google Scholar]
- 5.Villapiano NL, Winkelman TN, Kozhimannil KB, Davis MM, Patrick SW. Rural and Urban Differences in Neonatal Abstinence Syndrome and Maternal Opioid Use, 2004 to 2013. JAMA pediatrics. 2017;171(2):194–196. [DOI] [PubMed] [Google Scholar]
- 6.Haight SC, Ko JY, Tong VT, Bohm MK, Callaghan WM. Opioid Use Disorder Documented at Delivery Hospitalization - United States, 1999–2014. MMWR Morbidity and mortality weekly report. 2018;67(31):845–849. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Gabrielson SMB, Carwile JL, O’Connor AB, Ahrens KA. Maternal opioid use disorder at delivery hospitalization in a rural state: Maine, 2009–2018. Public Health. 2020;181:171–179. [DOI] [PubMed] [Google Scholar]
- 8.Growth in urban population outpaces rest of nation, Census Bureau reports. https://www.census.gov/newsroom/releases/archives/2010_census/cb12-50.html. Published 2012. Accessed January 31, 2019.
- 9.Kraus ML, Alford DP, Kotz MM, et al. Statement of the American Society Of Addiction Medicine Consensus Panel on the use of buprenorphine in office-based treatment of opioid addiction. J Addict Med. 2011;5(4):254–263. [DOI] [PubMed] [Google Scholar]
- 10.Saia KA, Schiff D, Wachman EM, et al. Caring for Pregnant Women with Opioid Use Disorder in the USA: Expanding and Improving Treatment. Curr Obstet Gynecol Rep. 2016;5(3):257–263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.ACOG Committee Opinion No. 711: Opioid Use and Opioid Use Disorder in Pregnancy. Obstet Gynecol. 2017;130(2):e81–e94. [DOI] [PubMed] [Google Scholar]
- 12.The ASAM National Practice Guideline for the Treatment of Opioid Use Disorder: 2020 Focused Update. J Addict Med. 2020;14(2S Suppl 1):1–91. [DOI] [PubMed] [Google Scholar]
- 13.ACOG Committee Opinion No. 524: Opioid abuse, dependence, and addiction in pregnancy. Obstet Gynecol. 2012;119(5):1070–1076. [DOI] [PubMed] [Google Scholar]
- 14.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] [PMC free article] [PubMed] [Google Scholar]
- 15.Maine Health Data Organization Governing Statute. In. 22, Chapter 1683. Vol 90 590. [Google Scholar]
- 16.Maine Health Organization Release Notes, 2020. Impact of SAMHSA 42 CFR Part 2 on MHDO Historical and Future Claims Data. https://mhdo.maine.gov/. Published 2020. Accessed. [Google Scholar]
- 17.Electronic Code of Federal Regulations. PART 2—CONFIDENTIALITY OF SUBSTANCE USE DISORDER PATIENT RECORDS. https://www.ecfr.gov/cgi-bin/text-idx?rgn=div5;node=42%3A1.0.1.1.2. Accessed. [Google Scholar]
- 18.Research Data Assistance Center. Redaction of Substance Abuse Claims
- 19.Alliance for Innovation in Maternal Health. Severe Maternal Morbidity codes list v6-27-2020. Birth Admission Codes. https://safehealthcareforeverywoman.org/aim-data/. Accessed. [Google Scholar]
- 20.Rural-Urban Commuting Area Codes - Data.gov. https://catalog.data.gov/dataset/rural-urban-commuting-area-codes. Accessed March 25, 2019.
- 21.Perinatal Health; For Providers - Division of Disease Prevention - Maine CDC: DHHS Maine. https://www.maine.gov/dhhs/mecdc/population-health/mch/perinatal/providers.html. Accessed April 22, 2019.
- 22.Critical Access Hospital Locations - Flex Monitoring Team. https://www.flexmonitoring.org/data/critical-access-hospital-locations/. Accessed April 22, 2019.
- 23.Heslin KC, Owens PL, Karaca Z, Barrett ML, Moore BJ, Elixhauser A. Trends in Opioid-related Inpatient Stays Shifted After the US Transitioned to ICD-10-CM Diagnosis Coding in 2015. Medical care. 2017;55(11):918–923. [DOI] [PubMed] [Google Scholar]
- 24.Medicaid Outcomes Distributed Research Network, the Academy Health State University Partnership Learning Network. https://www.academyhealth.org/MODRN. Accessed.
- 25.Brogly SB, Saia KE, Werler MM, Regan E, Hernandez-Diaz S. Prenatal Treatment and Outcomes of Women With Opioid Use Disorder. Obstet Gynecol. 2018;132(4):916–922. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.National Drug Codes HCPCS Crosswalk. https://ndclist.com/ndc-hcpcs-crosswalk. Accessed.
- 27.Wachman EM, Saia K, Miller M, et al. Naltrexone Treatment for Pregnant Women With Opioid Use Disorder Compared With Matched Buprenorphine Control Subjects. Clin Ther. 2019;41(9):1681–1689. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Brogly SB, Saia K, Hernandez-Diaz S, Werler M, Sebastiani P. The comparative safety of buprenorphine versus methadone in pregnancy-what about confounding? Addiction. 2016;111(12):2130–2131. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Venkatesh AK, Mei H, Kocher KE, et al. Identification of Emergency Department Visits in Medicare Administrative Claims: Approaches and Implications. Acad Emerg Med. 2017;24(4):422–431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Clemans-Cope L, Lynch V, Howell E, et al. Pregnant women with opioid use disorder and their infants in three state Medicaid programs in 2013–2016. Drug Alcohol Depend. 2019;195:156–163. [DOI] [PubMed] [Google Scholar]
- 31.Krans EE, Kim JY, James AE 3rd, Kelley D, Jarlenski MP. Medication-Assisted Treatment Use Among Pregnant Women With Opioid Use Disorder. Obstet Gynecol. 2019;133(5):943–951. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Lo-Ciganic WH, Donohue JM, Kim JY, et al. Adherence trajectories of buprenorphine therapy among pregnant women in a large state Medicaid program in the United States. Pharmacoepidemiol Drug Saf. 2019;28(1):80–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Bryant A, Blake-Lamb T, Hatoum I, Kotelchuck M. Women’s Use of Health Care in the First 2 Years Postpartum: Occurrence and Correlates. Maternal and child health journal. 2016;20(Suppl 1):81–91. [DOI] [PubMed] [Google Scholar]
- 34.Declercq E, Stern JE, Luke B, et al. Maternal Postpartum Hospitalization Following Assisted Reproductive Technology Births. Epidemiology. 2015;26(6):e64–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Peterson C, Liu Y, Xu L, Nataraj J, Zhang K, Mikosz CA. U.S. National 90-Day Readmissions After Opioid Overdose Discharge. American Journal of Preventive Medicine. 2019;56(6):875–881. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Choi S, Biello KB, Razzi AR, Drainoni M-L. Age differences in emergency department utilization and repeat visits among patients with opioid use disorder at an urban safety-net hospital: A focus on young adults. Drug and Alcohol Dependency. 2019:14–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Health, United Stats 2018 Chartbook. Emergency department visits within the past 12 months among adults aged 18 and over, by selected characteristics: United States, selected years 1997–2017. https://www.cdc.gov/nchs/hus/index.htm. Published 2019. Accessed September 30, 2020.
- 38.Wakeman SE, Larochelle MR, Ameli O, et al. Comparative Effectiveness of Different Treatment Pathways for Opioid Use Disorder. JAMA Netw Open. 2020;3(2):e1920622. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Kaiser Program on Medicaid and the Uninsured National Survey, with the Center on Budget and Policy Priorities, 2003–2009; and with the Georgetown University Center for Children and Families, 2011–2020. Available at:. https://www.kff.org/medicaid/report/annual-updates-on-eligibility-rules-enrollment-and-renewal-procedures-and-cost-sharing-practices-in-medicaid-and-chip/. Accessed September 30, 2020.
- 40.Brooklyn JR, Sigmon SC. Vermont Hub-and-Spoke Model of Care for Opioid Use Disorder: Development, Implementation, and Impact. J Addict Med. 2017;11(4):286–292. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Appendix Figure 1. Flow chart showing total number of deliveries paid for (in part) by Medicaid ending in 2010–2018 in Vermont and Maine, according to enrollment status in Medicaid during pregnancy and the first year postpartum