Shortening the Current Opioid Misuse Measure via computer-based testing: a retrospective proof-of-concept study

Authors: Finkelman, M. D.; Kulich, R. J.; Zoukhri, D.; Smits, N.; Butler, S. F.
BMC Medical Research Methodology, 2013, 13:126, https://doi.org/10.1186/1471-2288-13-126

Background

The Current Opioid Misuse Measure (COMM) is a self-report questionnaire designed to help identify aberrant drug-related behavior in respondents who have been prescribed opioids for chronic pain. The full-length form of the COMM consists of 17 items. Some individuals, especially compromised individuals, may be deterred from taking the full questionnaire due to its length. This study examined the use of curtailment and stochastic curtailment, two computer-based testing approaches that sequentially determine the test length for each individual, to reduce the respondent burden of the COMM without compromising sensitivity and specificity.

Methods

Existing data from n = 415 participants, all of whom had taken the full-length COMM and had been classified via the Aberrant Drug Behavior Index (ADBI), were divided into training (n = 214) and test (n = 201) sets. Post-hoc analysis of the test set was performed to evaluate the screening results and test lengths that would have been obtained, if curtailment or stochastic curtailment had been used. Sensitivity, specificity, and average test length were calculated for each method and compared with the corresponding values of the full-length test.

Results

The full-length COMM had a sensitivity of 0.703 and a specificity of 0.701 for predicting the ADBI. Curtailment reduced the average test length by 22% while maintaining the same sensitivity and specificity as the full-length COMM. Stochastic curtailment reduced the average test length by as much as 59% while always obtaining a sensitivity of at least 0.688 and a specificity of at least 0.701 for predicting the ADBI.

Conclusions

Curtailment and stochastic curtailment have the potential to achieve substantial reductions in respondent burden without compromising sensitivity and specificity. The two sequential methods should be considered for future computer-based administrations of the COMM.