Representativeness of participants in the Danish National Health Survey in relation to mental disorders and a range of other medical conditions and mortality

Abstract Although the Danish National Health Survey (DNHS) uses survey weights to reduce self-selection bias, differences in health status may remain. We explored whether DNHS participants are representative of the general population in Denmark regarding morbidity, particularly mental disorders, and mortality. By individual linkage to administrative registers, we compared DNHS participants ( N = 183,194) with the general population aged 16 years or older ( N = 4,697,047). For 202 health conditions, we estimated period prevalence ratios (PPRs) of having received a diagnosis in the 5 years prior to survey participation and hazard ratios (HRs) of receiving a diagnosis in the subsequent 5 years. Additionally, we estimated mortality HRs, overall and among individuals with prior diagnoses of the specific conditions. Both original weights developed for the DNHS and updated weights better accounting for mental disorders were applied. Overall, weighted survey participants had similar prevalence compared to the general population, but mental disorders were less common before (PPR: 0.86 [0.85; 0.88]) and after (HR: 0.86 [0.84; 0.88]) survey participation. Participants experienced lower all-cause mortality rates than the general population (HR: 0.80 [0.79; 0.82]), especially among those with prior mental disorders (HR: 0.62 [0.59; 0.66]). Applying updated weights resulted in similar mental disorder prevalence (PPR: 0.99 [0.98; 1.01]) but unchanged mortality estimates. The lower mortality among survey participants compared to the general population, especially for individuals with prior mental disorders, suggests that self-selection into participation is not fully accounted for by the original or updated weights. This has implications for the generalisability of DNHS-based analyses.