Only 11% of Hospitalized Patients With Alcohol Use Disorder Get Medication

Of 4,232 hospitalizations involving a diagnosis of alcohol use disorder at two hospitals between 2018 and 2023, 467 ended with a new prescription for an FDA-approved addiction medication. That is 11 percent — roughly one in nine discharges — according to a retrospective cohort study published this week in the Journal of General Internal Medicine.

The gap is not explained by which patients walked through the door. When Neela Nataraj, John T. Ratelle and colleagues modeled the odds of a new prescription, the only patient-level factor that reached significance was whether alcohol use disorder was the primary reason for admission rather than a secondary diagnosis. Almost everything else that moved the needle was a feature of how the hospital organized its care.

What the study measured

The cohort covered 2,347 patients treated at two hospitals inside a large academic medical center. Both sites lacked a dedicated inpatient addiction medicine consult service, a detail that matters for interpreting the results: the hospitals in the study are not places where addiction specialists are routinely on the floor.

The outcome was a new prescription for a medication for alcohol use disorder at the point of discharge, which excludes patients already stabilized on one before admission. The investigators used multivariable models with generalized estimating equations to handle the fact that some patients were admitted repeatedly during the six-year window.

Alcohol use disorder is common among hospitalized adults and carries elevated risk of readmission and death after discharge. The medications in question — naltrexone and acamprosate among the FDA-approved options — have been shown to improve clinical outcomes. The treatment gap the study documents sits between that evidence and routine hospital practice.

Length of stay, and who signs the discharge order

Patients admitted for three to four days were considerably more likely to leave with a prescription than those discharged after two days or less, with an odds ratio of 1.70. The association weakened as stays lengthened: 1.47 for five to seven days, and 1.37 for eight days or more, a result whose confidence interval crossed one.

Consultation by care management was associated with higher odds of prescribing, at 1.37. So was the service that discharged the patient. Relative to a hospital internal medicine non-teaching team, discharge from a general medicine teaching service carried an odds ratio of 1.41, and discharge from a family medicine service an odds ratio of 2.07 — more than double.

A primary alcohol use disorder diagnosis, meanwhile, produced an odds ratio of just 1.25. Patients whose admission was coded for alcohol as the main problem were only modestly more likely to receive medication than those admitted for something else entirely.

The organizational reading

The authors draw a practical conclusion from that pattern: if prescribing tracks service structure and consultative support more than patient characteristics, then it is modifiable. Changes to care pathways, discharge workflows and consult services are, in principle, things a hospital can decide to do. They name service structure and consultative support as the two most promising targets, and note that neither site in the study had a dedicated inpatient addiction medicine consult service.

The discharge window itself has drawn more attention in recent years. A 2024 cohort study published in JAMA Network Open, led by researchers at Massachusetts General Hospital, found that adults who received an alcohol use disorder medication at discharge after an alcohol-related hospitalization had a lower risk of returning to the hospital within 30 days. That finding, and others like it, is why the 11 percent figure lands as a gap rather than a baseline.

For patients, the days after leaving are the ones that matter most. Someone exiting a detoxification stay or a medical admission for alcohol addiction steps out of supervision with cravings intact, and a prescription filled on the way out is one of the few interventions that carries protection into that stretch.

What the study cannot establish

The design limits what can be claimed. This is a retrospective look at two hospitals in one academic system, and the results describe associations rather than cause and effect. Whether a family medicine service prescribes more because of its clinical culture, its patient mix, or the length of its typical admission cannot be separated out with these data.

Diagnosis codes carry their own distortion. Patients whose alcohol use was documented in the chart but not coded as a discharge diagnosis would not appear in the cohort at all, and coding practice varies by service — which is itself a possible partial explanation for the service-level differences the study found.

The authors also note that the outcome stops at the prescription. A written order does not guarantee a filled script, a sustained course, or a follow-up visit, and none of those downstream steps were measured. What the study counts is the moment of decision at discharge, and on that measure, eight of every nine eligible patients left without one.