Differential item functioning (DIF) of the Kessler psychological distress scale (K6) among adults using National Health Interview Survey (NHIS) data

The Kessler K6 is widely used to measure non-specific psychological distress in U.S. national surveys, yet measurement equivalence across key health and sociodemographic subgroups has not been evaluated. Group comparisons can be biased if items are interpreted differently across populations. Evaluating differential item functioning across diverse subgroups is essential to ensure the validity of distress estimates. Using data from 164,291 adults (ages 18-85 years, median 50 years, 55.1% female) from the 2013-2017 National Health Interview Survey, we examined item bias in the Kessler K6 with item response theory-based graded response models. Uniform and non-uniform differential item functioning were evaluated across chronic health condition status, sex, age, educational attainment, geographic region, survey language, and survey year using likelihood ratio tests and pseudo-R 2 statistics. Monte Carlo simulations were used to empirically calibrate effect size thresholds to account for large sample sizes. Likelihood ratio χ² tests identified statistically significant uniform and non-uniform differential item functioning across several Kessler K6 items and subgroups; however, the magnitude of differential item functioning was consistently negligible across all analyses. Small differences in item thresholds were observed across chronic health condition status and sociodemographic groups (e.g., sex, age, education). However, Monte Carlo simulation analyses indicated that observed differential item functioning was trivial and unlikely to meaningfully affect comparisons of psychological distress across groups. The Kessler K6 demonstrates strong measurement equivalence across key health and sociodemographic subgroups, supporting the validity of estimated psychological distress in population-based research. These findings reinforce the suitability of the Kessler K6 for national surveillance efforts.