Background: The Minimal Insomnia Symptom Scale (MISS) has been suggested as a brief 3-item screening tool for detecting insomnia. Each item has an ordered 5-category (0-4) response scale and the instrument yields a total score between 0-12 (higher scores = more insomnia). Available MISS evaluations have been based on classical test theory (CTT) approaches. Different cut-offs for identifying insomnia have been suggested for adults (aged 20-64) and elderly (aged 65+). For adults, a cut-off of ≥6 has been suggested, while a cut-off of ≥7 has been suggested for the elderly, as determined from applications of the Youden index.

Aim: To test the measurement properties of the MISS using the Rasch measurement model, with special emphasis on Differential Item Functioning (DIF) by age, and to explore implications for the two suggested cut-off scores.

Design: Cross-sectional MISS data from adult (n=1075) and elderly (n=548) populations were analysed by the unrestricted polytomous Rasch measurement model using the RUMM2030 software program. To avoid an inflated type I error rate, sample size was algebraically adjusted to 500 in the calculation of P-values while leaving all other aspects of data (e.g., locations, fit residuals) unaltered.

Results: Mean person location was -1.095 (SD, 1.28), i.e. items tend to represent more severe levels of insomnia than that experienced by the sample. However, for the purpose of screening this may be considered acceptable. There were no statistically significant deviations from model expectations, with a non-significant overall item-trait interaction (χ² = 26.94, P=0.173). Reliability (PSI) was 0.59 suggesting that the MISS can separate approximately two statistically distinct groups of people (1.92 strata). The highest Information Function (IF) was found at -0.2 logits. There were no disordered response category thresholds. There was uniform DIF by age for all three items, which disappeared following adjustment (split by age group) for the most pronounced DIF, suggesting that DIF was artificial for two items. Examination of raw scores-to-location estimates between the two age groups revealed differences at the lower and higher ends of the scale. The raw score cut-off of ≥6 was associated with a smaller logit difference between age groups than the ≥7 cut-off (0.09 vs. 0.23). That is, at a raw score of 6 the two age groups were comparable regarding their logit location estimates. This raw score (representing a logit value of -0.03 for the pooled sample) was also the one closest to the location of the highest IF (i.e., -0.2 logits).

Summary and implications: This study provides general support for the measurement properties of the MISS. However, caution should be exercised in comparing MISS scores between age groups, but applying a ≥6 raw score cut-off appears to allow for valid comparisons between adults and elderly regarding the presence of insomnia. Nevertheless, additional studies are needed to determine the clinically optimal cut-score for identification of insomnia.