Clinical and Population Health Research PhD Program, Graduate School of Biomedical Sciences, University of Massachusetts Medical School, Worcester, Massachusetts, USA.
Department of Population and Quantitative Health Sciences, University of Massachusetts Medical School, Worcester, Massachusetts, USA.
Int J Geriatr Psychiatry. 2020 Jul;35(7):769-778. doi: 10.1002/gps.5301. Epub 2020 Apr 17.
To identify subgroups of nursing home (NH) residents in the USA experiencing homogenous depression symptoms and evaluate if subgroups vary by cognitive impairment.
We identified 104 465 newly admitted, long-stay residents with depression diagnosis at NH admission in 2014 using the Minimum Data Set 3.0. The Patient Health Questionnaire-9 was used to measure depression symptoms and the Brief Interview of Mental Status for cognitive impairment (intact; moderately impaired; severely impaired). Latent class analysis (LCA) with logistic regression was used to: (a) construct the depression subgroups and (b) estimate adjusted odds ratios (aOR) and 95% confidence intervals (CI) of the associations between the subgroups and cognitive impairment level, adjusting for demographic and clinical characteristics.
The best-fitted LCA model suggested four subgroups of depression: minimal symptoms (latent class prevalence: 42.4%), fatigue (32.0%), depressed mood (14.5%), and multiple symptoms (11.2%). Odds of subgroup membership varied by cognitive impairment. Compared to residents with intact cognition, those with moderate or severe cognitive impairment were less likely to belong to the fatigue subgroup [aOR(95% CI): moderate: 0.75 (0.71-0.80); severe: 0.26 (0.23-0.29)] and more likely to belong to the depressed mood subgroup [aOR (95% CI): moderate: 4.54 (3.55-5.81); severe: 6.41 (4.86-8.44)]. Residents with moderate cognitive impairment had increased odds [aOR (95% CI): 1.19 (1.12-1.27)] while those with severe impairment had reduced odds of being in the multiple symptoms subgroup [aOR (95% CI): 0.63 (0.58-0.68)].
Findings provide a basis for improving depression management with consideration of both subgroups of depression symptoms and levels of cognitive function.
在美国，确定具有同质抑郁症状的养老院（NH）居民亚组，并评估亚组是否因认知障碍而有所不同。
我们使用最小数据集 3.0 识别了 2014 年 NH 入院时患有抑郁诊断的 104465 名新入院的长期居民。使用患者健康问卷-9 来衡量抑郁症状，使用简短的精神状态访谈来衡量认知障碍（完整；中度受损；严重受损）。使用逻辑回归进行潜在类别分析（LCA），以：（a）构建抑郁亚组；（b）估计亚组与认知障碍水平之间关联的调整优势比（aOR）和 95%置信区间（CI），并调整人口统计学和临床特征。
最佳拟合的 LCA 模型表明存在四种抑郁亚组：轻微症状（潜在类别患病率：42.4%）、疲劳（32.0%）、情绪低落（14.5%）和多种症状（11.2%）。亚组成员的可能性因认知障碍而异。与认知功能正常的居民相比，认知功能中度或重度受损的居民不太可能属于疲劳亚组[aOR（95%CI）：中度：0.75（0.71-0.80）；重度：0.26（0.23-0.29）]，而更有可能属于情绪低落亚组[aOR（95%CI）：中度：4.54（3.55-5.81）；重度：6.41（4.86-8.44）]。中度认知障碍患者的可能性增加[aOR（95%CI）：1.19（1.12-1.27）]，而严重认知障碍患者的可能性降低[aOR（95%CI）：0.63（0.58-0.68）]）属于多种症状亚组。
这些发现为改善抑郁症管理提供了依据，同时考虑了抑郁症症状和认知功能的亚组。
