Validation of self-administered single item screening questions (SISQs) for unhealthy alcohol and other drug use in primary care patients at two sites Jennifer McNeely, Charles M. Cleland, Shiela M. Strauss, Joseph J. Palamar, John Rotrosen, Marc N. Gourevitch, Richard Saitz No relevant financial relationships to disclose Objectives 1. Describe the need for a self-administered approach to substance use screening 2. Single Item Screening Questions (SISQs) for alcohol and drug use 3. Present results of a validation study in primary care 4. Discuss feasibility and application Screening for substance use in primary care • Medical providers fail to identify clinically D’Amico, Medical Care 2005 relevant substance use • Barriers to screening: o o o o o Time Workflow Knowledge/Training Discomfort Attitudes Sterling, Addict Med Clin Pract 2012 Friedmann, J Gen Int Med 2000 Friedmann, Arch Int Med 2001 Anderson, Alcohol Alcoholism 2004 McCormick, J Gen Int Med 2006 Self-administered screening is a more feasible approach Screening: Screening SISQ-alcohol and SISQ-drug + Assessment Low Risk Moderate Risk Education Monitoring Office-based counseling High Risk or Dependence Treatment Single Item Screening Questions • SISQ-alcohol How many times in the past year have you had X or more drinks in a day? (X=5 for men, and X=4 for women) • SISQ-drug How many times in the past year have you used an illegal drug or used a prescription medication for nonmedical reasons (for example, because of the experience or feeling it caused? Prior validation of SISQs • Adult primary care patients (N=286) • Single site, urban safety net medical center • Good sensitivity and specificity for detection of unhealthy use • SISQ-alcohol: Sensitivity 82%, Specificity 79% • SISQ-drug: Sensitivity 85%, Specificity 96% Smith et al., JGIM 2009 Smith et al., Arch Int Med 2010 Current Study Screening (computer) • • SUBS SISQ-alcohol, SISQ-drug Validation Measures (interviewer) • • • Timeline Follow Back SIP-A and SIP-D MINI-Plus • • REALM Demographics Second Consent Saliva drug screen Referrals Incentive Reference standard measures Timeline followback (30d) Alcohol Unhealthy use + SIP-A SIP-D + MINI-Plus MINI-Plus Intercept oral fluid screening abuse or dep test* + Disorder Drugs Unhealthy use Disorder + + + + + + * Collected at Site A only Statistical Analysis 1. Comparison of SISQs to composite reference standards 2. Examined site differences 3. Calculate sensitivity, specificity, AUC: oUnhealthy use oSubstance use disorder 4. Subgroup analyses Study Sites and Recruitment • • • • Adult primary care clinics 2 urban safety net hospitals Patients presenting for medical visits Consecutive recruitment Eligibility Criteria: • Age 21-65 • Current clinic patient • Fluent in English • No disability preventing computer use Participant Recruitment Screened: N = 2131 Eligible: N = 915 1216 were excluded Language: 679 Age: 306 Not a patient: 168 Other: 115 453 declined No time: 363 Other: 90 1 lost data Completed interview: N = 459 Site A: 265* Site B: 194 *230 (87%) Site A participants agreed to saliva test Characteristics of the 459 participants Age (years) Mean = 46, SD = 12 Range = 21-65 Sex (%) Male Female Transgender 48.4 51.2 0.4 Race/Ethnicity (%) Black/African American Hispanic White/Caucasian Other 51.8 20.2 19.1 8.6 Country of Birth (%) United States Outside of United States 64.6 35.3 Education and Health Literacy Highest Level of Education 25% 14% Less than high school High school diploma/GED 61% Health Literacy Level College degree 41% < High school 59% High school or higher Prevalence of substance use Substance Alcohol Drugs Specific drug categories Illicit drugs Marijuana Cocaine Heroin Hallucinogens Prescription drugs (non-medical use) Opioids Benzodiazepines Stimulants Past year use (MINI) N (%) 103 (22.3)a 114 (24.7)c 108 (23.4) ----21(4.6) ---- Past month use (TLFB) N (%) 89 (19.3)b 73 (15.8)c 58 (12.6) 12 (2.6) 10 (2.2) 1 5 3 2 Unhealthy use Substance + on SISQ N (%) + on Reference Alcohol 155 (34) N (%) 146 (32) Drugs 107 (23) 122 (27) Oral fluid test results: Sensitivity Specificity AUC % % (95% CI) (95% CI) (95% CI) 73.3 (65.3, 80.3) 84.7 (80.2, 88.5) 0.79 (0.75, 0.83) 71.3 (62.4, 79.1) 94.3 (91.3, 96.6) 0.83 (0.79, 0.87) 8 tested positive, all reported use on SISQ No change to results Substance use disorder Substance + on SISQ N (%) + on Reference Alcohol 155 (34) N (%) 60 (13) Drugs 107 (23) 74 (16) Sensitivity Specificity AUC % % (95% CI) (95% CI) (95% CI) 86.7 (75.4, 94.1) 74.2 (69.6, 78.4) 0.80 (0.76, 0.85) 85.1 (75.0, 92.3) 88.6 (85.0, 91.6) 0.87 (0.83, 0.91) Subgroup Analysis Subgroups anticipated to have greater difficulty with self-administered screening: • Male • Age greater than 50 • Hispanic/Latino • Primary language other than English • Born outside US • Education or health literacy lower than high school level Subgroup Analysis • No differences for SISQ-alcohol • Lower sensitivity of SISQ-drug among: Primary language other than English (p<0.01) Sensitivity Specificity English 74.3 (65.1, 82.2) 94.4 (90.7, 96.9) Non-English 46.2 (19.2, 74.9) 94.3 (87.1, 98.1) Less than high school education (p<0.01) Sensitivity Specificity High school 79.0 (66.8, 88.3) 95.2 (91.0, 97.8) < High school 63.3 (49.9, 75.4) 93.3 (88.0, 96.7) 50% Required assistance 40% Other Computer-related Comprehension/ reading 30% 20% 10% 0% Site A Site B Limitations • Safety net primary care populations • English speaking only • Tested in research context, with assurance of confidentiality Conclusions • SISQs accurately identified unhealthy substance us in primary care patients • Lower sensitivity and specificity than interviewer-administered versions • Efficiency, fidelity, and patient comfort may be advantages to self-administered approach Acknowledgements Funding: • K23 Career Development Award NIDA K23 DA030395 • NYU-HHC CTSI Pilot Grant NIH/NCATS UL1 TR000038 • The MITRE Corporation (contract from ONC and SAMHSA) Staff and others: • Seville Meli • Jacqueline German • Ritika Batajoo • Catherine Federowicz • Marshall Gillette • Charlie Jose • Emily Maple • Keshia Toussaint • Julianne Cameron • Arianne Ramautar • Derek Nelsen • Linnea Russell • Study participants