Frequently Asked Questions

The New Hampshire Hospital Screening and Referral Algorithm (NHHSRA) showed high accuracy in this referred inpatient sample, with 96.6% sensitivity, 93.6% specificity, 98.5% positive predictive value, 86% negative predictive value, 95.5% overall accuracy, and an F1 score of 97.5%. ROC analysis found an AUC of 0.82 (95% CI, 0.78–0.85), which the authors interpreted as good discriminative ability. Among 927 patients, the algorithm identified 730 true positives and 160 true negatives, with 11 false positives and 26 false negatives.

The NHHSRA uses 5 objective data points available at admission rather than relying mainly on patient self-report. The article states that these include recent positive urine toxicology results, history of substance use disorder diagnoses, legal mandates for treatment, and current pharmacologic substance use treatment regimens. It is structured as a yes/no decision tree that classifies a patient as either needing substance use intervention or not needing intervention.

The study analyzed 927 psychiatric inpatients with serious mental illness admitted to New Hampshire Hospital between January 2023 and February 2025 after exclusions. Patients were admitted under involuntary status and had schizophrenia spectrum disorders, bipolar disorders, or severe major depression; those older than 65 years were excluded because the algorithm had been developed and validated for ages 18–65 years. The initial referred sample included 1,045 patients, and the final analytic sample excluded 38 patients older than 65 years and 80 with non-SUD classifications such as gambling or pain management.

The most common primary substance use disorders in the 927-patient sample were opioids (37.9%), alcohol (23.9%), cannabis (15.1%), methamphetamine (13.7%), and cocaine (2.7%). The first 5 substance categories accounted for 93% of the sample. These data describe the referred inpatient population evaluated in this study rather than all psychiatric inpatients more broadly.

The most frequent principal psychiatric diagnoses were bipolar disorder (34.0%), schizophrenia spectrum disorders (29.7%), and major depressive disorder (15.2%). The article also notes less frequent diagnoses including borderline personality disorder, unspecified mood disorder, and posttraumatic stress disorder. The main diagnostic categories analyzed together accounted for 84% of the sample.

Yes. Compared with alcohol use disorder, opioid use disorder was associated with substantially higher adjusted odds of receiving a recommendation for substance use intervention from the addiction team (aOR = 6.45; CI, 3.11–14.71; P < .001). Opioid use disorder was also associated with higher adjusted odds of screening positive on the NHHSRA (aOR = 5.21; 95% CI, 2.75–10.48; P < .001).

In this study, cannabis use disorder was associated with lower adjusted odds of both clinician-recommended substance use intervention and a positive NHHSRA screen compared with alcohol use disorder. For addiction-team recommendation, the adjusted odds ratio was 0.23 (CI, 0.14–0.39; P < .001). For NHHSRA screening, the adjusted odds ratio was 0.22 (95% CI, 0.13–0.37; P < .001).

Yes. In the clinician-based referral model, patients with borderline personality disorder had lower adjusted odds of receiving a recommendation for substance use intervention than patients with depression (aOR = 0.36; CI, 0.13–0.99; P = .045). The discussion notes that this difference was seen for clinician-driven referral but not for NHHSRA screening positivity.

Yes. The sample had a mean age of 36.48 years (SD = 11.0), and 66.9% were younger than 40 years. Patients with cannabis use disorder were younger on average than those without it (mean 31.3 years, SD 12.5 vs mean 37.4 years, SD 10.4; P < .001), while the discussion and conclusion note that older patients were more likely to have alcohol use disorder. Females were slightly older overall (P = .005), and fewer females had cannabis or opioid use disorder as the primary SUD.

The gold standard was the determination of a board-certified addiction psychiatrist who directly evaluated each referred patient. The assessment included a clinical interview, electronic health record and toxicology review, Prescription Drug Monitoring Program data, and collateral information such as family input. A patient was coded as needing intervention if the addiction psychiatrist determined that pharmacologic relapse-prevention treatment or motivational or relapse-prevention psychotherapy would be clinically meaningful.

The article suggests the NHHSRA may be useful because it relies on objective admission data rather than self-report, which can be limited in acutely decompensated inpatients with serious mental illness. The authors state that the tool provides a standardized, reproducible, and scalable way to identify patients who may benefit from substance use interventions while reducing variability from individual clinical judgment. They further argue that this approach may support more equitable access to addiction services in psychiatric settings.

The authors identified several limitations.

• Less common substance and psychiatric categories were collapsed into broader groups, which may have masked important differences.
• Both NHHSRA performance and the gold standard assessment may depend on local resource availability, staff expertise, and patient history, which could differ across settings.
• Variables such as substance use severity, substance use history, and socioeconomic status were outside the study scope.
• Using mutually exclusive categories for primary substance and psychiatric diagnosis does not fully capture the high prevalence of comorbidity and polysubstance use in this population.