Key Takeaways
- A study of 9,360 pregnancies among Tennessee Medicaid beneficiaries with opioid use disorder found that a greater number of days of dispensed medication around pregnancy and early postpartum was associated with a lower risk of overdose or death later in the first postpartum year (Kim et al., 2026).
- Between postpartum days 42 and 365, 148 first recorded outcomes of nonfatal overdose or all-cause death were observed among the 9,360 pregnancy records, representing 1.58% of the cohort (Kim et al., 2026).
- Among pregnancies with evidence of buprenorphine treatment, a modeled 132 days of dispensed medication was associated with approximately half the hazard of first nonfatal overdose or death compared with a one-day supply (Kim et al., 2026).
- The study was observational, so it cannot establish that longer treatment itself caused the lower risk or that 132 days represents a clinical treatment threshold.
- Nearly all pregnancies in the cohort involved buprenorphine, so the numerical associations primarily reflect buprenorphine-treated pregnancies in this Tennessee Medicaid population.
Introduction
Longer medication treatment for opioid use disorder during pregnancy and early postpartum was associated with a lower risk of first nonfatal overdose or all-cause death later in the postpartum year in a Tennessee Medicaid cohort.
Remaining in medication treatment for opioid use disorder during pregnancy and after delivery may be particularly important during a period when patients remain vulnerable to overdose and death.
A large study of Tennessee Medicaid beneficiaries found that a greater number of days of dispensed medication during late pregnancy and the first six weeks after delivery was associated with a lower subsequent risk of nonfatal overdose or all-cause death.
The study, published October 2 in JAMA Network Open, examined 9,360 pregnancies among 7,776 people with opioid use disorder who had at least some medication treatment. Longer treatment duration was associated with lower risk during the period from postpartum day 42 through day 365 (Kim et al., 2026).
The results add to evidence that continuity of opioid use disorder treatment may matter beyond pregnancy itself. However, because this was an observational study, continued treatment could also reflect differences in healthcare access, treatment engagement, stability, and other factors associated with overdose and mortality risk.
What the Study Examined
Kim et al. (2026) used data from Tennessee’s Medicaid program, TennCare, linked with birth and death certificates and Tennessee Hospital Discharge Data.
The study included people aged 15 to 44 who delivered a live-born infant at 20 weeks of gestation or later between 2007 and 2020. Follow-up data extended through 2021.
Everyone included in the analysis had opioid use disorder and had at least one day of buprenorphine or naltrexone dispensed during a 132-day window extending from 90 days before delivery through 41 days afterward.
Participants also had continuous TennCare enrollment during this exposure period, allowing gaps of five days or fewer.
Importantly, pharmacy dispensing records indicate that medication was supplied. They cannot establish that every dispensed dose was actually taken.
Most treatment involved buprenorphine. Of the 9,360 pregnancies, 9,085, or 97.1%, involved buprenorphine alone. Another 172 involved naltrexone alone, while 103 involved both medications (Kim et al., 2026).
The median total medication supply during the 132-day exposure window was 88 days. The median was 62 days before delivery and 30 days after delivery.
Researchers then followed participants from postpartum day 42 through day 365 for the first occurrence of either nonfatal overdose or all-cause death. Overdose diagnoses could involve opioids or nonopioid drugs.
What Researchers Found
Between postpartum days 42 and 365, 148 first recorded outcomes of nonfatal overdose or all-cause death were observed among the 9,360 pregnancy records, representing 1.58% of the cohort. The outcomes included 109 first nonfatal overdoses and 39 deaths (Kim et al., 2026).
Longer medication treatment was associated with progressively lower hazards of the combined outcome.
Among pregnancies with evidence of buprenorphine treatment, compared with a one-day medication supply, the adjusted hazard ratios were (Kim et al., 2026):
- 30 days: 0.86 (95% CI, 0.76-0.97)
- 60 days: 0.74 (95% CI, 0.58-0.94)
- 90 days: 0.64 (95% CI, 0.45-0.92)
- 132 days: 0.52 (95% CI, 0.31-0.88)
A modeled 132 days of dispensed medication supply was therefore associated with approximately 48% lower hazard of first nonfatal overdose or death compared with a one-day supply after statistical adjustment.
This was a relative, observational comparison. It does not mean that 132 days of treatment reduced every patient’s absolute risk by 48%, nor does it establish 132 days as a clinical treatment threshold. The 132-day figure represented the maximum possible medication coverage during the study’s exposure window.
The model-based adjusted risk difference for 132 versus one day of dispensed medication supply was -0.53 percentage points, with a 95% confidence interval from -1.07 to 0.01 percentage points (Kim et al., 2026).
Because that confidence interval narrowly included zero, the model-based absolute difference was not statistically conclusive under the study’s stated criterion.
Treatment Before and After Delivery
Treatment timing offered another view of the association.
In a secondary observational comparison, the model-estimated one-year probability of first overdose or death was 1.41% among people with medication for opioid use disorder dispensed both before and after delivery. It was 1.89% among those with treatment only before delivery and 2.58% among those with treatment only postpartum (Kim et al., 2026).
These groups may have differed in clinical, healthcare and social factors that also affected their subsequent risk.
The researchers performed several sensitivity analyses using different statistical and exposure assumptions, and the overall association between longer treatment duration and lower risk remained.
When they excluded the 172 pregnancies involving naltrexone alone and counted only buprenorphine exposure, 132 days compared with one day was associated with an adjusted hazard ratio of 0.44 (95% CI, 0.26-0.75) (Kim et al., 2026).
How It Compares With Earlier Research
The results extend previous evidence connecting longer medication treatment during pregnancy with lower overdose risk.
A 2022 JAMA Network Open study examined 2,072 deliveries among 1,999 commercially insured people with opioid use disorder. Longer use of methadone or buprenorphine during pregnancy was associated with progressively lower risk of nonfatal overdose during pregnancy (Jarlenski et al., 2022).
In that analysis, at least 10 weeks of medication treatment was associated with a 57% lower adjusted risk of nonfatal overdose compared with no observed medication treatment. The adjusted relative risk was 0.43 (95% CI, 0.19-0.94) (Jarlenski et al., 2022).
Only 20 individuals experienced an overdose, with 22 overdose events recorded overall, producing considerable statistical uncertainty around some estimates.
The newer Tennessee study addressed a different question. Rather than concentrating on overdose during pregnancy, Kim et al. (2026) examined whether cumulative medication treatment during late pregnancy and early postpartum was associated with overdose or death during the later postpartum period.
Together, the studies provide evidence that treatment duration and continuity deserve attention rather than simply whether someone received medication at any point. Neither study, however, can establish that each additional week or month of medication directly causes a particular reduction in overdose risk.
What the Results May Mean
Buprenorphine and methadone are recommended medication treatments for opioid use disorder during pregnancy, and clinical guidance emphasizes continuity of treatment after delivery (Centers for Disease Control and Prevention, 2026; American College of Obstetricians and Gynecologists, 2017).
The new study does not determine whether patients should receive a particular number of days of treatment, nor does it establish 132 days as an optimal treatment duration.
Instead, a greater number of days of dispensed medication during the study’s late-pregnancy and early-postpartum exposure window was associated with lower subsequent risk of first nonfatal overdose or all-cause death in this cohort (Kim et al., 2026).
Continued treatment may also be a marker of treatment engagement, continuity of healthcare, greater stability or other factors that the researchers could not fully measure.
The postpartum period deserves particular attention because delivery does not end the risks associated with opioid use disorder. Changes in medical care, insurance coverage, childcare responsibilities and other circumstances can complicate continuity of treatment.
Importantly, risk remained even among patients with comparatively continuous treatment.
Among those with medication dispensed both before and after delivery, the model-estimated probability of overdose or death during follow-up was still 1.41% (Kim et al., 2026).
The findings therefore should not be interpreted as showing that medication eliminates postpartum overdose risk. Instead, they add evidence supporting efforts to avoid unnecessary disruptions in evidence-based opioid use disorder care as patients transition from pregnancy into postpartum care.
People who are pregnant or postpartum should not stop, reduce or change prescribed medication for opioid use disorder based on this study. Medication changes should be planned with the treating clinician because abrupt discontinuation can increase withdrawal and relapse risks (Centers for Disease Control and Prevention, 2025; American College of Obstetricians and Gynecologists, 2017).
Limitations
The study cannot establish cause and effect.
People who remain in medication treatment longer may differ in important ways from those who discontinue earlier. They may have more stable access to healthcare, greater treatment engagement or other characteristics that independently affect overdose and mortality risk.
The researchers adjusted for numerous measured factors, but residual confounding remains possible.
Because only 148 combined outcomes occurred, the investigators had limited statistical degrees of freedom for covariate adjustment, and residual confounding remains possible (Kim et al., 2026).
Medication exposure was determined from pharmacy dispensing records. Dispensing a medication does not guarantee that it was taken as prescribed, and treatment obtained outside the available pharmacy claims could have been missed.
Medication exposure may also have been underestimated during prolonged hospitalizations.
The study did not include methadone in its exposure analysis. TennCare did not begin covering methadone for opioid use disorder until 2020, the final year of the delivery cohort, and the researchers observed no TennCare-covered methadone claims for opioid use disorder during the study’s exposure window (Kim et al., 2026).
Nearly all pregnancies in the cohort involved buprenorphine. The numerical associations therefore primarily reflect buprenorphine-treated pregnancies in this Tennessee Medicaid population and should not be assumed to apply equally to methadone, naltrexone or other treatment settings.
Events occurring outside Tennessee could also have been missed if participants moved out of state.
The study population further limits generalizability. Participants were Tennessee Medicaid beneficiaries, and more than 95% of pregnancies in the cohort were among non-Hispanic White individuals.
Treatment practices and policies also changed during the long study period, which covered deliveries from 2007 through 2020.
The cohort was restricted to live births, meaning the findings do not necessarily apply to pregnancies ending in early pregnancy loss or other outcomes outside the study’s eligibility criteria.
Finally, the study included only people who had at least one day of medication for opioid use disorder dispensed during the exposure window. It therefore does not directly answer whether receiving medication is associated with better outcomes than receiving no medication at all.
Final Thoughts
Among Tennessee Medicaid beneficiaries with opioid use disorder who had medication treatment around childbirth, a greater number of days of dispensed treatment was associated with a lower risk of first nonfatal overdose or all-cause death during the remainder of the first postpartum year (Kim et al., 2026).
The association was substantial at longer treatment durations and persisted across several sensitivity analyses.
But this was an observational study. It cannot show that extending treatment by a particular number of days will itself produce a specific reduction in overdose or mortality.
The more defensible conclusion is that sustained medication treatment around pregnancy and early postpartum was associated with lower subsequent risk in this cohort. Continued treatment may also be a marker of treatment engagement, continuity of care and other factors that the study could not fully measure.
For healthcare systems, the results add observational evidence supporting efforts to maintain continuity of evidence-based opioid use disorder care as patients move from prenatal care through delivery and into longer-term postpartum care. The study does not establish that extending treatment by a specific number of days will prevent a specific number of overdoses or deaths.
Funding and Disclosures
The study by Kim et al. (2026) was funded by a grant from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, part of the National Institutes of Health.
The authors reported that the funder had no role in the study’s design or conduct, data collection, management, analysis or interpretation, manuscript preparation or review, or the decision to submit the work for publication.
Several authors disclosed relationships outside or during the submitted work. Sarah Osmundson reported NIH grants during the study. Carlos Grijalva reported NIH grants during the study as well as personal fees from Merck and GlaxoSmithKline, contracts from the Centers for Disease Control and Prevention and Syneos Health, and other NIH and Agency for Healthcare Research and Quality grants.
Amelie Pham reported a Novocuff grant and compensation from Vanderbilt University Medical Center for serving on a data safety monitoring board for an NIH-funded study. Stephen Patrick reported NICHD funding during the study, National Institute on Drug Abuse grants, and personal fees for expert consulting.
No other disclosures were reported.
References
Kim, A., Wiese, A. D., Osmundson, S., You, C., Grijalva, C. G., Adgent, M. A., Sundermann, A. C., Spieker, A. J., Pham, A., Patrick, S. W., & Leech, A. A. (2026). Duration of medication use for opioid use disorder and risk of postpartum overdose and death. JAMA Network Open, 9(10), e2637214. https://doi.org/10.1001/jamanetworkopen.2026.37214
Jarlenski, M., Chen, Q., Gao, A., Rothenberger, S. D., & Krans, E. E. (2022). Association of duration of methadone or buprenorphine use during pregnancy with risk of nonfatal drug overdose among pregnant persons with opioid use disorder in the US. JAMA Network Open, 5(4), e227964. https://doi.org/10.1001/jamanetworkopen.2022.7964
American College of Obstetricians and Gynecologists. (2017). Opioid use and opioid use disorder in pregnancy. Committee Opinion No. 711. https://www.acog.org/clinical/clinical-guidance/committee-opinion/articles/2017/08/opioid-use-and-opioid-use-disorder-in-pregnancy
Centers for Disease Control and Prevention. (2026). Treatment of opioid use disorder before, during, and after pregnancy. https://www.cdc.gov/opioid-use-during-pregnancy/treatment/index.html