Coman Emil N, Wu Helen Z, Assari Shervin
UConn Health Disparities Institute, University of Connecticut Health Center, Hartford, CT 06106, USA.
Department of Psychiatry, University of Connecticut Health Center, Farmington, CT 06030, USA.
Brain Sci. 2018 Nov 28;8(12):207. doi: 10.3390/brainsci8120207.
Poor comparability of social groups is one of the major methodological problems that threatens the validity of health disparities (HD) research findings. We illustrate a methodological solution that can additionally unpack the mechanisms behind differential effects on depression and anxiety. We describe racial/ethnic differences in the prevalence of depression and anxiety scores between Black and White women using classic methods, and then we illustrate a 1:1 matching procedure that allows for building of individual-level difference scores, i.e., actual HD difference score variables, for each pair of comparable participants. We compare the prevalence of depression disorder between Black and White young women after matching them 1:1 on common socio-economic characteristics (age, employment, education, and marital status). In essence, we follow matching or stratification methods, but make a step further and match cases 1:1 on propensity scores, i.e., we create Black⁻White 'dyads'. Instead of concluding from plain comparisons that 11% more White young women (18⁻30 years old) report a depressive disorder than Black young women, the matched data confirms the trend, but provides more nuances. In 27% of the pairs of comparable pairs the White woman was depressed (and the comparable Black woman was not), while in 15% of the pairs the Black woman was depressed (and the comparable White woman was not). We find that Black-to-White disparities in neighborhood disorder do not predict depression differences (HDs), while such an effect is evident for anxiety HDs. The 1:1 matching approach allows us to examine more complex HD effects, like differential mediational or resilience mechanisms that appear to be protective of Black women's mental health.
社会群体可比性差是威胁健康差异（HD）研究结果有效性的主要方法学问题之一。我们阐述了一种方法学解决方案，该方案还可以揭示对抑郁和焦虑产生不同影响背后的机制。我们使用经典方法描述了黑人和白人女性在抑郁和焦虑得分患病率上的种族/民族差异，然后说明了一种1:1匹配程序，该程序允许为每对可比参与者构建个体水平的差异得分，即实际的HD差异得分变量。我们在将黑人和白人年轻女性在共同的社会经济特征（年龄、就业、教育和婚姻状况）上进行1:1匹配后，比较她们之间抑郁症的患病率。本质上，我们遵循匹配或分层方法，但更进一步，根据倾向得分将病例1:1匹配，即我们创建黑-白“二元组”。不是从简单比较中得出18至30岁的白人年轻女性报告抑郁症的比例比黑人年轻女性高11%的结论，匹配后的数据证实了这一趋势，但提供了更多细微差别。在27%的可比对中，白人女性患有抑郁症（而可比的黑人女性没有），而在15%的对中，黑人女性患有抑郁症（而可比的白人女性没有）。我们发现邻里混乱方面的黑人与白人差异并不能预测抑郁差异（HDs），而这种影响在焦虑HDs方面很明显。1:1匹配方法使我们能够研究更复杂的HD效应，比如似乎对黑人女性心理健康有保护作用的差异中介或复原力机制。
