The alcohol admission ends with a phone number, and that is why it fails

Design

systematic review with GRADE assessment of hospital-based trials and cohort studies, five databases, January 2000 to February 2025

Population

17 studies of 21 interventions: 12 randomised trials (n=1,917), 3 non-randomised trials (n=459), 2 cohort studies (n=9,974)

Primary outcome

linkage to outpatient follow-up after hospitalisation for alcohol use disorder

Effect

8 of 21 interventions (38%) improved linkage; randomised effect sizes ranged from an 8% absolute decrease to a 30% absolute increase; certainty low

Admissions related to alcohol use disorder are common and carry high mortality and readmission rates, and the point at which most of them fail is the transition out of hospital. This systematic review searched five databases from January 2000 to February 2025 and found 17 studies covering 21 hospital-based interventions: 12 randomised trials totalling 1,917 patients, three non-randomised trials with 459, and two cohort studies with 9,974. Risk of bias and certainty were formally assessed.

Most interventions, 85%, had several components. Behavioural elements appeared in 76% and care coordination in 76%, social support in 33% and medication management in just 14%. Eight of the 21 interventions (38%) improved linkage to outpatient follow-up. Effect sizes across the randomised trials ranged from an 8% absolute decrease to a 30% absolute increase in follow-up attendance. Certainty of evidence was low, limited by risk of bias and imprecision.

Two things in that summary are worth acting on even at low certainty. The interventions that worked generally combined behavioural work and care coordination with either medication for alcohol use disorder or social support, rather than relying on any one of these. And medication management featured in only 14% of interventions studied at all, which is a striking omission given that naltrexone and acamprosate are inexpensive, well tolerated and can be started in hospital.

The realistic change for most clinicians is small and specific: start the medication before discharge rather than recommending it, name the person and the appointment rather than the service, and involve whoever is going home with the patient. Say clearly what this evidence does not support, which is any particular packaged programme; the review's honest conclusion is that few interventions of low certainty helped, and that better trials are needed.

• Start naltrexone or acamprosate during the admission where appropriate, rather than recommending it at discharge
• Book a named appointment with a named clinician before the patient leaves, not a referral to a service
• Involve a family member or other support person in the discharge conversation with the patient's consent
• Treat behavioural input and care coordination as a pair; single-component interventions were the ones that failed
• Record that alcohol use disorder was addressed, so that the next admission does not start from nothing

The statistics, in plain English

Only 38% of interventions improved linkage to follow-up, and effect sizes ran from an 8% absolute fall to a 30% absolute rise in attendance, a spread wide enough that the average across them means little. Low GRADE certainty means the true effects could differ substantially from these estimates, driven here by risk of bias in the individual studies and by imprecision from small samples. That is a reason to prefer the components with independent evidence behind them, such as alcohol use disorder medication, over adopting any packaged intervention on the strength of this review.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.