Computerized Screener and Opioid Assessment for Patients with Pain-Revised (SOAPP-R)

Evaluations

• Shortening the Screener and Opioid Assessment for Patients with Pain-Revised (SOAPP-R): A proof-of-principle study for customized computer-based testing

  Finkelman MD, Kulich RJ, Zacharoff KL, et al. Pain Medicine. 2015. 16: 2344-2356. doi: 10.1111/pme.12864

  Summary: Researchers evaluated the sensitivity and specificity of the SOAPP-R compared to the Aberrant Drug Behavior Index (ADBI) using curtailment (i.e. ending the assessment once a respondent is certain to score positive or negative for high-risk AMRB) and stochastic curtailment (i.e. ending the assessment once the respondent is sufficiently likely to score positive or negative, based on a predetermined criterion) with cut-off criterions of 99%, 95%, and 90%. Researchers used data from 428 patients recruited from pain clinics who had completed the paper-based SOAPP-R and the ADBI. Respondents who scored 19 or above on the SOAPP-R were considered positive for AMRB. Probabilities of assessment outcomes were determined using a statistical model trained on the full data set. SOAPP-R using curtailment matched the full screener in terms of sensitivity (0.745) and specificity (0.671) compared to the ADBI, but reduced the average assessment length to 17.7 items (80.6% stopped early). Stochastic curtailment with a criterion of 99% also matched the full screener, but reduced the average assessment length to 14.1 items (86.4% stopped early). Stochastic curtailment with a criterion of 95% had a sensitivity of 0.724 and a specificity of 0.671 for predicting the ADBI and reduced the average assessment length to 10.8 items (100% stopped early). Stochastic curtailment with a criterion of 90% had a sensitivity of 0.710 and a specificity of 0.668 for predicting the ADBI and reduced the average assessment length to 8.3 items (100% stopped early).

  Take Away: A computer-based SOAPP-R can use stochastic curtailment to reduce assessment length by stopping the assessment once there is a 99% or 95% chance of the respondent scoring positive or negative for aberrant medication-related behavior without greatly reducing sensitivity or specificity.

• Feasibility of tablet computer screening for opioid abuse in the emergency department

  Weiner SG, Horton LC, Green TC, Butler SF. Western Journal of Emergency Medicine. 2015. 16(1): 18-23. doi: 10.5811/westjem.2014.11.23316

  Summary: Researchers collected data about participant completion of the SOAPP-R and participant-reported ease of use to evaluate the feasibility of administering the SOAPP-R on a tablet in emergency departments. Indicators of feasibility included whether participants could complete the tablet-administered SOAPP-R without help, how many participants could complete the tablet-administered SOAPP-R in less than 5 minutes, and participant-reported ease of use. Participants were asked one question about the ease of completing the SOAPP-R with a 5-point Likert scale (very easy to very difficult) on the tablet after completing the SOAPP-R. All participants completed the SOAPP-R without help. The median time to complete the assessment was 2 minutes and 28 seconds, with 78 (95.1%) participants completing the SOAPP-R in less than 5 minutes. Most participants (93%) reported that the SOAPP-R was very easy to complete.

  Take Away: Tablet administration of the SOAPP-R is feasible and easy to use for most participants.

• A comparison of an opioid abuse screening tool and prescription drug monitoring data in the emergency department

  Weiner SG, Horton LC, Green TC, Butler SF. Drug and Alcohol Dependence. 2016. 159: 152-157. doi: 10.1016/j.drugalcdep.2015.12.007

  Summary: Researchers evaluated the psychometric properties of the tablet-administered SOAPP-R in emergency department patients. Researchers recruited 82 emergency department patients being discharged with a prescription opioid based on identification in the medical record, confirmed by a physician. Participants were asked to complete the 24-item SOAPP-R on a tablet. Researchers did not help participants complete the SOAPP-R and could not see their final scores. Researchers used a previously-validated cut-off for predicting AMRB of 18 on the SOAPP-R to classify patients as high-risk. Researchers compared participant responses on the SOAPP-R to state prescription drug monitoring program (PDMP) data for each participant, with 4 or more prescriptions and 4 or more providers as indicators of high-risk behavior. The SOAPP-R had a sensitivity of 53.9%, a specificity of 71.0%, a positive predictive value (percentage of true positives given a positive screen) of 25.9%, and a negative predictive value 89.1% (percentage of true negatives given a negative screen). The correlations between SOAPP-R score and numbers of opioid prescriptions (r2=.05) and number of prescribers (r2=.08) were both statistically significant. The adjusted odds ratio for the relationship between a positive SOAPP-R score and high-risk PDMP data was 1.39, which did not reach significance. The SOAPP-R with a cut-off score of 18 achieved a “good fit” for identifying participants with 4 or more opioid prescriptions or providers (area under the curve=0.81). A cut-off score of 20 on the SOAPP-R achieved the best psychometrics (sensitivity=53.9%, specificity=73.9%).

  Take away: Tablet-administration of the SOAPP-R in emergency departments showed mixed psychometrics, suggesting a need for evaluating the suitability of PDMP data as an objective indicator of risk for aberrant medication-related behavior in emergency departments.

• Computer-based testing and the 12-item Screener and Opioid Assessment for Patients with Pain-Revised: A combined approach to improving efficiency

  Kulich RJ, Butler SF, Smits N, Weiner SG. Journal of Applied Behavioral Research. 2018. e12145. doi: 10.1111/jabr.12145

  Summary: Researchers analyzed the psychometrics of the full-length and short-form SOAPP-R as regular assessments, using curtailment, and using stochastic curtailment to improve assessment efficiency while maintaining psychometrics. Researchers retrospectively analyzed 3 data sets of pain patients (N=622) recruited from pain clinics (n=428), a hospital-based pain center (n=84), and primary care clinics (n=110) who had completed the paper-based full-length SOAPP-R. Participant responses on the SOAPP-R were compared to other validated measures of aberrant medication-related behavior (e.g., Aberrant Drug Behavior Index, Prescription Drug Use Questionnaire, Addiction Behaviors Checklist, Current Opioid Misuse Measure) and urine screening. Using participant SOAPP-R responses, researchers analyzed how participants would have scored on the short-form SOAPP-R and on the full-length and short-form SOAPP-R using curtailment and stochastic curtailment with cut-off criterions of 99% (SC-99) and 95% (SC-95) for all 3 samples. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for the short-form SOAPP-R (Respectively: .67-.8, .46-.59, .42-.62, .62-.85) were within .03 of the full-length SOAPP-R (respectively: .67-.79, .44-.59, .41-.63, .63-.85) for all 3 samples. Additionally, curtailment and SC-99 maintained the psychometrics of the non-curtailed assessments for all 3 samples. Shortening the full-length and short-form SOAPP-R using SC-95 resulted in psychometrics within .04 of the non-curtailed assessments. Curtailment, SC-99, and SC-95 reduced the average length of the full-length SOAPP-R to 15.9-16.9, 12.8-13.6, and 8.8-10.4 items, respectively. Curtailment, SC-99, and SC-95 reduced the average length of the short-form SOAPP-R reduced the average assessment length to 7.8-8.4, 7-7.6, and 4.9-6.2 items, respectively.

  Take Away: Curtailment and stochastic curtailment, particularly using a cut-off criterion of 99%, can reduce the length of the full-length and short-form SOAPP-R assessment without significantly affecting the psychometrics.

• Related Articles

  An investigation of completion times on the Screener and Opioid Assessment for Patients with Pain-Revised (SOAPP-R). Finkelman MD, Kulich RJ, Butler SF, et al. Journal of Pain Research. 2016. 9: 1163-1171. doi: 10.2147/JPR.S123454