AlcoTail - Implementation of Tailored Interventions

Patients included in the study before the implementation of the intervention protocols will be assigned to the control group. These patients will receive acute medical care according to current procedures. Patients included in the study after implementation of intervention protocols will be assigned to the case group. In addition to acute medical care, these patients will be routinely screened for harmful alcohol consumption using Alcohol Use Disorder Identification Test (AUDIT-C). Patients scoring < 4 points (men) / 3 points (women) will not receive any intervention. Patients scoring 4-6 points (men) / 3-5 points (women) will be informed about the possible harmful effects of their consumption and will be advised to reduce their consumption. From patients scoring > 6 points (men) >5 points (women) a blood sample will be collected and analyzed for relevant biomarkers. They will be further evaluated using Prediction of Alcohol Withdrawal Severity Scale (PAWSS). Patients at risk of developing delirium tremens will be treated according to Clinical Institute Withdrawal Assessment (CIWA-Ar) procedures, and referred to specialized alcohol addiction treatment. Patients in the case group that has prescription(s) for psychoactive medicinal drugs will also be routinely evaluated by pharmacological assessment of drug concentration in serum samples. If there is a discrepancy between prescribed dose and serum concentration, or medicinal drugs other than those being prescribed are found, the patient will be advised to use medicines according to prescription or have their prescribed dosage altered. If necessary, the patient will be referred to specialized drug addiction treatment.

The patients will be recruited from 3 hospitals: Oslo University Hospital, Lovisenberg Diakonale Hospital (Oslo) and St Olav Hospital (Trondheim). The recruitment period will be approximately 4 months for the control group and 4 months for the case group, and estimations based on previous admission numbers for the 3 hospitals results in about 2500 patients in each group.

For the study, research personnel will approach the eligible patients once they have been admitted to various medical wards. The personnel will give the patients written and verbal information about the study, and the patients who agree to be part of the study will sign a written consent. Baseline data will be recorded for all patients included in the control and case groups. These data include:

• Alcohol use past 12 months, Alcohol Use Disorder Identification Test (AUDIT-4)
• Alcohol use last 24 hours,
• Stages-of-change questionnaire on alcohol, psychoactive medicinal and/or illicit drug use past 12 months,
• Non-prescribed use of psychoactive medicinal drugs past 12 months
• Drug Use Disorder Identification Test (DUDIT)
• Stages-of-change questionnaire on psychoactive medicinal drugs or illicit drugs
• Symptoms Checklist (SCL-5)
• Left-over blood samples used for routine diagnostic testing will be collected and analyzed for ethanol, phosphatidylethanol (PEth; biomarker for alcohol use last 3 weeks), diazepam, N-desmethyldiazepam, clonazepam, oxazepam, nitrazepam, alprazolam, zopiclone, zolpidem, morphine, codeine, oxycodone, buprenorphine, methadone, tramadol, amphetamine, methamphetamine, ecstasy (MDMA), THC, cocaine and benzoylecgonine.
• Patients in the case group will be invited to submit a new blood sample for PEth analysis 2 months after inclusion.
• After 12 months, baseline data will be coupled to the following data from the patient journal: date and time of admission, days and hours spent at bed-post and intensive care, number of previous hospital admissions, ICD-10 diagnoses at discharge, level of care, AUDIT-C score, ethanol and PEth concentration in blood, psychoactive medicine concentration in serum, CIWA-Ar, Glasgow Coma Scale (GCS) score, first triage at hospital, medicinal curve at admission, marital status, employment status, care status for children under 18 years of age. Data will also be coupled to various registry data; The Norwegian Patient Registry: number of admissions in the specialist health care service 5 years prior to admission, number of treatments with main diagnosis from ICD-10 the within 5 years prior to admission, number of admissions to interdisciplinary specialist treatment within 5 years after admission, number of admissions within 5 years after admission; The Social Security Database: performance status at time of admission and 1 year after; The Norway Control and Payment of Health Reimbursement (KUHR) Database and the Norwegian Registry for Primary Health Care: patients use of institutions and other primary health care services such as home nursing and nursing homes 5 years prior to and 5 years after admission, overview of drug use diagnoses; the Norwegian Prescription Registry: medication prescribed 5 years prior to and 5 years after admission; and the Cause of Death Registry: cause of death within 5 years after admission.

The database from baseline (questionnaires and blood sample analysis), patient journal data and the various registry data collected from each included patient will be treated de-identified. The questionnaires will be in an online-format ("Nettskjema"), and the data will be submitted directly in a secure database provided by the University of Oslo, called Services for Sensitive Data (short form "TSD" in Norwegian).

In the final database, missing values in variables will result in exclusion of that patient in statistical analysis (when the variable with missing value is part of a specific analysis). Statistical analyses will include:

For the alcohol intervention:

• To control for baseline measures, multiple models for repeated measures will be used to evaluate anticipated differences between the intervention and the control groups. For the continuous outcomes (number of admissions and GP visits) linear mixed models will be fitted, for binary outcomes models with log link or Poisson link function will be used.

For the psychoactive medicinal drug intervention:

• Multiple mixed effects models for repeated measures will be fitted. The three different institutions will be entered as random effects, while selected covariates as patients' age and gender will be modeled as fixed effects. Continuous outcomes (serum concentrations) will be analyzed using linear mixed models, while changes in prescriptions with the outcome being binary will be modeled using mixed models with log or Poission link function. The results will be expressed as effect sizes (for linear models) or relative risks (RR) for models of binary outcomes.