A randomised test of the effect of medical v. lay idiom on assessment of perceived mental health condition in the USA

AIMS: To test the impact of using different idioms in epidemiological interviews on the prevalence and correlates of poor mental health and mental health service use. METHODS: We conducted a randomised methodological experiment in a nationally representative sample of the US adult population, comparing a lay idiom, which asked about ‘problems with your emotions or nerves’ with a more medical idiom, which asked about ‘problems with your mental health’. Differences across study arms in the associations of endorsement of problems with the Kessler-6 (a validated assessment of psychological distress), demographic characteristics, self-rated health and mental health service use were examined. RESULTS: Respondents were about half as likely to endorse a problem when asked with the more medical idiom (18.1%) than when asked with the lay idiom (35.1%). The medical idiom had a significantly larger area under the ROC curve when compared against a validated measure of psychological distress than the lay idiom (0.91 v. 0.87, p = 0.012). The proportion of the population who endorsed a problem but did not receive treatment in the past year was less than half as large for the medical idiom (7.90%) than for the lay idiom (20.94%). Endorsement of problems differed in its associations with age, sex, race/ethnicity and self-rated health depending on the question idiom. For instance, the odds of endorsing problems were threefold higher in the youngest than the oldest age group when the medical idiom was used (OR = 3.07; 95% CI 1.47–6.41) but did not differ across age groups when the lay idiom was used (OR = 0.76; 95% CI 0.43–1.36). CONCLUSION: Choice of idiom in epidemiological questionnaires can affect the apparent correlates of poor mental health and service use. Cultural change within populations over time may require changes in instrument wording to maintain consistency in epidemiological measurement of psychiatric conditions.