Identifying long-term psychological distress from single measures: evidence from a nationally representative longitudinal survey of the Australian population.

backgroundSingle time-point assessments of psychological distress are often used to indicate chronic mental health problems, but the validity of this approach is unclear. The aims of this study were to investigate how a single assessment of distress relates to longer-term assessment and quantify misclassification from using single measures to indicate chronic distress.

methodsData came from the Household, Income and Labour Dynamics in Australia Survey, a nationally representative study of Australian adults. Psychological distress, measured with the Kessler10 and categorised into low (scores:10- < 12), mild (12- < 16), moderate (16- < 22) and high (22-50), has been assessed in the Survey biennially since wave 7. Among respondents who were aged ≥25 years and participated in all waves in which distress was measured, we describe agreement in distress categories, and using a mixed linear model adjusting for age and sex we estimate change in scores, over a two-, four-, six- and eight-year follow-up period. We applied weights, benchmarked to the Australian population, to all analyses.

resultsTwo-years following initial assessment, proportions within identical categories of distress were 66.0% for low, 54.5% for mild, 44.0% for moderate and 50.3% for high, while 94.1% of those with low distress initially had low/mild distress and 81.4% with high distress initially had moderate/high distress. These patterns did not change materially as follow-up time increased. Over the full eight-year period, 77.3% of individuals with high distress initially reported high distress on ≥1 follow-up occasion. Age-and sex- adjusted change in K10 scores over a two-year period was 1.1, 0.5, - 0.7 and - 4.9 for low, mild, moderate and high distress, respectively, and also did not change materially as follow-up time increased.

conclusionIn the absence of repeated measures, single assessments are useful proxies for chronic distress. Our estimates could be used in bias analyses to quantify the magnitude of the bias resulting from use of single assessments to indicate chronic distress.