Assessing Agreement between the K10 and MHI-5 Measures of Psychological Wellbeing

Abstract

Responses to the Kessler Psychological Distress Scale (K10) and the Mental Health Index-5 (MHI-5) mental health subscale of the Medical Outcomes Study Short Form (SF-36) survey cover broadly similar constructs. The aim of this paper is to use the equipercentile method to produce a concordance between K10 and MHI-5 across the whole score distribution. Comparisons were made using survey data from the Australian Longitudinal Study on Women’s Health (ALSWH) and the Household, Income and Labour Dynamics in Australia (HILDA) survey, which used both the K10 and the MHI-5 measures at the same time for the same participants. Agreement was assessed with Bland-Altman plots. The differences between MHI-5 scores and transformed K10 scores were assessed with paired t-tests. For the ALSWH data there is good agreement between MHI-5 scores and scores equated from the K10. The mean of the differences is −0.15 with standard deviation 10.8. Using the ALSWH transformation on the HILDA data for different age groups and genders, the mean differences ranged from 0.43 to 3.38 with corresponding standard deviations of 11.1 to 9.9. The concordance table was very similar to one obtained by the more complicated item response theory method. The results suggest that longitudinal surveys of the same respondents that previously used MHI-5 could adopt the K10 (or vice versa) without loss of comparability.