Predictors of Mental Health Service Use across the Life Course: Evidence from a Multigroup Analysis Based on Andersen's Behavioral Model

Background and Purpose: Mental health problems are increasingly prevalent across all age groups in South Korea, yet service utilization remains critically low. Despite 27.8% of the population experiencing mental health issues across their lifetime, only 4.8% have utilized mental health services. Prior research has focused on specific vulnerable populations or single age groups, leaving limited understanding of how predisposing, enabling, and need factors operate across life course stages among the general population and thereby constraining the development of effective life course-sensitive service strategies. Grounded in Andersen's Behavioral Model, this study examined predictors of mental health service use across three life course stages—young adulthood (ages 20–39), middle adulthood (ages 40–64), and older adulthood (ages 65 and above)—and tested whether their effect structure differed significantly across groups.

Methods: Data were drawn from the 2024 Korea National Health and Nutrition Examination Survey (KNHANES), a nationally representative dataset based on a rolling two-stage stratified cluster sampling design (N=5,367; young adults n=1,229, middle-aged n=2,460, older adults n=1,678). The outcome was past-year mental health service use. Predictors were organized according to Andersen's model: predisposing factors (age, sex, education, household type), enabling factors (income, marital status, insurance type), and need factors (depression by PHQ-9, anxiety by GAD-7, perceived stress). Known-class multigroup logistic regression was conducted in Mplus 8.10 using MLR estimation, and between-group differences in regression coefficients were tested simultaneously using a Wald test.

Results: Mental health service utilization differed significantly by life course stage (^χ²=51.812, p<.001), with rates of 7.9% among young adults, 4.3% among middle-aged adults, and 2.3% among older adults. Among young adults, only need factors were significant—depression (b=0.118, p<.01) and anxiety (b=0.127, p<.001). Among middle-aged adults, education (b=0.484, p<.01), insurance type (b=1.660, OR=5.259, p<.001), depression (b=0.124, p<.001), anxiety (b=0.098, p<.001), and perceived stress (b=0.404, p<.05) were positively associated with service use, whereas household type (b=-0.822, p<.01) was negatively associated. Among older adults, sex (b=-0.717, p<.05), anxiety (b=0.088, p<.05), and perceived stress (b=0.556, OR=1.743, p<.01) were significant. Anxiety was the only predictor significant across all life stages. The Wald test indicated no significant between-group differences in regression coefficients (^χ²=28.204, p=.105).

Conclusions and Implications: Although the composition of significant predictors varied by life course stage, the overall effect structure did not differ significantly across groups, suggesting that utilization differences reflect variation in population composition and resource access rather than fundamentally distinct mechanisms. This study provides a systematic, life course-comparative analysis using multigroup modeling, demonstrating the value of life course-informed approaches in designing targeted mental health service delivery systems. Findings support life course-sensitive strategies: promoting mental health literacy for young adults, implementing multilevel interventions to address informational, economic, and clinical barriers for middle-aged adults, and strengthening community-based outreach for older adults. This is particularly important given their lowest utilization rates and the lower likelihood of service use among women despite the strongest stress-related association across all groups.