War disruption linked to opioid treatment discontinuation in Ukraine, cohort study finds

In the years since Russia’s full-scale invasion of Ukraine in February 2022, researchers and clinicians around the world have watched anxiously to see whether the country’s fragile health infrastructure could withstand the shock of active combat, mass displacement, and the destruction of medical facilities. Now, the first national study to quantify how war and displacement shape addiction treatment continuity offers a strikingly detailed answer. Drawing on Ukraine’s complete national registry of opioid agonist maintenance treatment, a team led by Roman Ivasiy of Yale University and Ukrainian public health institutions found that internal displacement and residence in frontline regions were by far the strongest predictors of treatment discontinuation, yet the overall treatment system proved remarkably resilient, retaining roughly 70 percent of patients and expanding enrollment by 17 percent during the first two years of the war.

Opioid agonist maintenance treatment, delivered with daily oral methadone or sublingual buprenorphine, is the cornerstone of both addiction care and HIV prevention for people who inject drugs. Ukraine carries the second-largest HIV epidemic in Europe, and among an estimated 360,000 people who inject drugs there, approximately 82 percent have opioid use disorder. When patients are forced off these medications, the consequences cascade rapidly: return to unsafe injection, elevated risk of HIV acquisition and transmission, overdose, suicide, and disengagement from antiretroviral therapy. Understanding what drives treatment interruption under wartime conditions is therefore not an abstract epidemiological exercise but a matter of immediate survival for thousands of patients.

The new study, published in The Lancet Regional Health – Europe, is the first anywhere to incorporate objective conflict exposure metrics and dynamic displacement data into longitudinal survival models of treatment retention. The research team analyzed records from SyRex, Ukraine’s national OAMT registry covering all government-operated treatment sites, identifying 17,211 adults who were actively receiving methadone or buprenorphine on February 23, 2022, the day before the invasion began. The study period ran through January 1, 2024, spanning nearly two full years of war. Because the registry records enrollment, treatment episodes, inter-site transfers, quarterly dosing histories, and take-home dosing status as mandatory fields, the dataset was complete, with no missing values for the variables analyzed, a rarity in observational research under crisis conditions.

A central methodological innovation was the way the team quantified war exposure. Rather than treating the conflict as a uniform backdrop, the researchers constructed a composite regional measure using six indicators: cumulative air raid alerts, explosions, artillery attacks, the proportion of internally displaced persons, days of partial occupation, and days of active frontline combat. Air raid and shelling data came from the Air Alarms platform, which aggregates real-time official emergency alert data, while displacement estimates were drawn from quarterly reports of the International Organization for Migration. Applying k-means clustering to these standardized indicators, validated with the gap statistic method, the team assigned each Ukrainian oblast to one of three conflict categories: low-intensity regions with limited hostilities, high-intensity regions marked by frequent missile strikes and elevated displacement, and frontline regions characterized by prolonged combat, occupation, and extensive artillery fire. Of the 17,211 patients, 26.9 percent lived in low-intensity regions, 40.7 percent in high-intensity regions, and 32.4 percent in frontline areas at the war’s outset.

To model outcomes, the researchers used a frailty-adjusted Cox proportional hazards framework, an extension of standard survival analysis that introduces an individual-level random effect to account for unobserved heterogeneity between patients. This mattered because 1,171 participants, or 6.8 percent, contributed more than one treatment episode during follow-up, and repeated episodes within the same person violate the independence assumptions of conventional Cox models. Dose and dispensing strategy were treated as time-varying covariates updated through quarterly registry snapshots, while discontinuation events and inter-regional transfers were recorded with exact dates. Treatment discontinuation was defined as missing ten or more consecutive days of medication, consistent with national reporting standards, and a sensitivity analysis using a 30-day definition confirmed the robustness of the findings.

The results are sobering. During follow-up, 5,040 participants, or 29.3 percent, experienced at least one treatment discontinuation, and 921 patients, or 5.4 percent, died. Two sharp spikes in discontinuation occurred in March and July of 2022 within frontline regions, coinciding with intensified Russian offensives in the Donbas in the east and Kherson in the south. Displacement status combined with regional war intensity was the dominant predictor of dropping out. Compared with patients who remained locally in low-intensity regions, those living in high-intensity regions had a 74 percent higher risk of discontinuation, and those in frontline regions had roughly three times the risk. Displacement amplified the hazard dramatically: patients displaced into low-intensity regions had a 3.55-fold increased risk, those displaced into high-intensity regions a 5.90-fold increase, and those displaced into frontline regions an 8.26-fold increase compared with stable residents of low-intensity areas. Only 519 patients, about 3 percent, relocated internally at least once during the study, a fraction that the authors acknowledge likely underestimates true population mobility because some displaced patients may have simply stopped treatment without formally transferring.

Yet the study also documents what protected patients. Receiving an optimally dosed regimen, defined as at least 90 milligrams per day of methadone or at least 16 milligrams per day of buprenorphine, reduced the hazard of discontinuation by 26 percent. Unsupervised, take-home dosing reduced it by 29 percent, an effect that held across nearly every regional stratum, with hazard ratios as low as 0.48 in low-intensity regions. Patients who had already been on treatment for more than 180 days before the invasion were 35 percent less likely to discontinue than those newly enrolled. These findings carry direct policy weight, because Ukraine entered the war having already expanded take-home dosing during the COVID-19 pandemic, and emergency measures enacted after the invasion allowed up to 30 days of take-home medication and simplified cross-regional patient transfers. At baseline, 84.7 percent of patients were already receiving unsupervised dosing, a fact that almost certainly buffered the system against the worst disruptions.

The analysis also revealed disparities that had not appeared in pre-war Ukrainian studies. Women were 20 percent more likely than men to discontinue treatment, a pattern consistent across all three conflict strata and one the authors attribute plausibly to wartime dynamics: women were more likely to relocate internationally for caregiving and child safety, while facing gender-specific barriers including stigma, trauma, childcare burdens, and heightened vulnerability to violence during displacement. Men, by contrast, had an added incentive to remain engaged, since patients on OAMT were exempt from military conscription under Order No. 402 of Ukraine’s Ministry of Health. HIV status was not independently associated with discontinuation overall, though it modestly predicted dropout in frontline regions. Structural economics mattered as well: living in a wealthier region, measured by pre-war gross regional domestic product per capita, was strongly protective, while buprenorphine patients in high-intensity regions fared worse than methadone patients, possibly reflecting reliance on privately financed, out-of-pocket buprenorphine that becomes unstable during displacement and economic disruption.

The broader significance of the findings extends well beyond Ukraine. Previous modeling work has established that expanding OAMT is one of the most cost-effective HIV prevention strategies in Eastern Europe and Central Asia, and registry analyses show that patients on OAMT engage more fully at every stage of the HIV care cascade, from testing to antiretroviral initiation and retention. Molecular epidemiology has even shown that the 2014 war in eastern Ukraine physically moved HIV lineages across the country as displaced people who inject drugs carried the virus through their sexual and injecting networks. Interrupting addiction treatment at scale, in other words, risks not only individual overdose deaths but a measurable acceleration of the national epidemic. The authors frame their results using the Big Events framework, warning that the current war could unleash a broader HIV epidemic if harm reduction and OAMT programs are not protected and scaled up.

Ukraine’s relative success, the researchers argue, was not accidental. It was built on years of patient-led advocacy, sustained funding from PEPFAR and the Global Fund, national policy reforms, and NIH-supported implementation initiatives such as NIATx that improved clinic efficiency and quality of care before the invasion. When war came, regions redistributed medication supplies, public and private clinics cooperated, and flexible transfer procedures kept patients connected to care even as transportation corridors and supply chains collapsed. In occupied territories, where OAMT and syringe services were shut down entirely, the consequences were severe, echoing earlier documented harms after methadone cessation in Crimea. The study’s conclusion is ultimately a prescription for preparedness: protocols for rapidly expanding unsupervised dosing, pre-positioned medication stocks, redundant supply chains, simplified transfer mechanisms, and cross-jurisdiction coordination should be treated as standard emergency planning rather than exceptional policy waivers, transforming addiction treatment from a vulnerable point of failure into a stabilizing platform for both individual health and population-level HIV prevention during humanitarian crises.