We conducted a prospective, quasi-experimental evaluation of an inpatient addiction consultation service, with 30-day post-discharge outcomes as the primary outcome measures. Because the program was implemented on a subset of medical floors, and not all eligible patients on implementation floors received inpatient consults, we were able to conduct a controlled evaluation of the program.
The study was conducted in an urban academic medical center with approximately 48,000 inpatient admissions annually. In November 2014, all inpatients began being screened for substance use by the admitting nurse using the Alcohol Use Disorders Identification Test–Consumption (AUDIT-C) and the National Institute on Drug Abuse (NIDA) single-question screening for drug use.15
16 The AUDIT-C, which comprises three questions from the Alcohol Use Disorders Identification Test (AUDIT), has demonstrated validity in identifying heavy drinking and alcohol use disorder. An AUDIT-C score of 8 or higher is 92% specific for alcohol use disorder and 99% specific for heavy drinking and/or alcohol use disorder.15 The single-question drug screen asks “How many times in the past year have you used an illegal drug or used a prescription medication for non-medical reasons?” A response of at least one time is considered positive. This screening question is 100% sensitive and 73.5% specific for the detection of drug use disorder, with test characteristics similar to the longer 10-item Drug Abuse Screening Test (DAST-10).16
Participants included adults 18 years or older admitted during the study period who had or were at high risk for drug or alcohol use disorder as determined by an AUDIT-C score of 8 or higher, had a positive single-item drug screen, or who were identified by the primary nurse as having a substance use disorder. Exclusion criteria were 1) pregnancy; 2) inability to be interviewed due to acute illness, somnolence, or cognitive impairment; 3) and screening solely for marijuana use.
On 14 study floors, eligible participants were asked by a unit nurse for permission to be approached by a research assistant. The research assistant, available only on weekdays during business hours, completed further screening to confirm eligibility. Participants provided oral informed consent and received a $25 gift card at each follow-up assessment.
Twelve of the 14 study floors had access to an addiction consult team (ACT). The remaining two floors did not. The limited access to ACT was due to a phased rollout of this new service. The two floors without access to ACT were floors with a mixed population of medical and surgical or respiratory acute care patients; non-medical patients on these floors were not included in the study. Floor assignment for patients was independent of this study and was based on bed availability at the time of admission. The decision to consult ACT was made by the primary clinical team independently of this study.
ACT is a multidisciplinary team consisting of a psychiatrist, a rotating group of internists with addiction expertise (all of whom had been granted waivers to prescribe buprenorphine, some board-certified in addiction medicine), advanced practice nurses (one full- and one part-time), three clinical social workers, a clinical pharmacist, recovery coach, and resource specialist. ACT provides each patient with a diagnosis and a longitudinal treatment plan, which is initiated in the hospital. The treatment plan includes an assessment of the ideal level of care, appropriateness of pharmacotherapy initiation, psychosocial needs, harm reduction needs, and patient readiness and preferences. Social workers provide bedside motivational enhancement therapy. Pharmacotherapy, when indicated, is initiated during hospitalization. Recovery coaches, who can provide peer support during hospitalization, are available in select outpatient practices and can longitudinally follow patients who receive primary care there. These coaches offer a range of services; examples include attending appointments, providing motivational support, helping with legal or social service needs, and connecting patients to other community supports. The social workers and resource specialist identify community-based treatment resources and make every effort to link patients directly to care following discharge. Patients without a concrete discharge plan are offered care through a low-threshold post-discharge clinic which provides transitional addiction treatment including pharmacotherapy.
Participants in the control group received standard of care for SUD prior to the establishment of ACT. Standard of care included access to a general psychiatry consult liaison team and floor social work. Hospital management of SUD generally included withdrawal treatment and referral to outpatient addiction care. Pharmacotherapy for SUD was rarely initiated in the hospital.
Participants were assessed upon enrollment and at 30 and 90 days following discharge. Baseline assessments were conducted in person. Follow-up assessments were completed by phone, in person, or through the mail, depending on participant preference. The research assistant made a maximum of 15 attempts at contact per participant. Participants were asked about substance use and a number of secondary measures, including healthcare utilization.
Substance use severity was assessed using the questions from the Addiction Severity Index (ASI) which make up the alcohol and drug composite scores. The ASI is a standardized instrument for evaluating the severity of problems for patients with SUD.17 The composite score for each problem area of the ASI was derived from sets of items within the ASI, and may be a better indicator of overall problem severity and change in problem status.18 The time period for composite scores is the prior 30 days. The items are combined using a formula to ensure equal weighting of each variable in the composite score. In addition, participants were asked to report the number of days within in the past 30 days that they had used alcohol or drugs.
Self-efficacy was assessed by a question demonstrated to be predictive of relapse.19 Participants were also asked about motivation for abstinence, lifetime and past-30-day history of overdose, treatment engagement, quality of life, and tobacco use and quit attempts. (See online Appendix for full survey).
The primary outcomes of this study were change in ASI composite score for alcohol and drug use and self-reported abstinence at 30 days post-discharge compared to baseline. This combination of self-reported abstinence and ASI composite score has been used effectively in measuring the impact of care interventions.20
Secondary outcomes included change in self-efficacy, motivation for abstinence, self-reported overdose, treatment engagement, mutual help attendance, self-reported health care utilization, and quality of life.
Although the primary endpoint of the study was 30 days, participants were followed and outcomes were collected for up to 90 days.
Continuous variables were summarized using the mean with standard deviation, and categorical variables were summarized using frequency with percentage. Separate analyses were conducted for those with 30-day follow-up and those with 90-day follow-up and changes from baseline were compared between the two study groups. For primary outcomes, regression models were used to compare outcomes, adjusting for differences in baseline characteristics. Using an intention-to-treat analysis, all enrolled patients were included and analyzed according to their original treatment groups. In order to include these participants in the analysis, we needed to impute outcomes for participants who were lost to follow-up. We conservatively assumed that participants who were lost to follow-up had no change from baseline severity. For binary secondary outcomes, the differences in change from baseline were compared by testing the interaction between group and time in the logistic regression models. All analyses were conducted using SAS version 9.4 software (SAS Institute Inc., Cary, NC). A two-sided p value of 0.05 or less was considered statistically significant.
The study was designed to enroll 200 intervention patients and 200 control patients in order to detect a mean difference of 0.28 in ASI drug use composite score, assuming a standard deviation of 0.1 with 80% power.20
The Partners Healthcare institutional review board approved the study.