Royall D R, Cabello M, Polk M J
Department of Psychiatry, The Audie L. Murphy VA Geriatric Research Education and Clinical Center, University of Texas Health Science Center at San Antonio 78264-7792, USA.
J Am Geriatr Soc. 1998 Dec;46(12):1519-24. doi: 10.1111/j.1532-5415.1998.tb01536.x.
To examine the relative contributions of Executive Control Function (ECF), general cognition, mood, problem behavior, physical disability, demographic variables, and the number of prescribed medications to the level of care received by older retirees.
Multivariate regression and discriminant modeling.
A single Continuing Care Retirement Community (CCRC) in San Antonio, Texas.
A total of 107 older retirees (mean age = 83.7+/-7.2 years), including 17 community-dwelling, well, older controls and 90 CCRC residents. CCRC subjects represented a convenience sample of consecutive referrals for geropsychiatric assessment. Sixty-one subjects resided at a noninstitutionalized level of care, and 46 were institutionalized.
Tests of ECF (the Executive Interview (EXIT25)), general cognition (the Mini-Mental State Examination (MMSE)), mood (the Geriatric Depression Scale short-form (sGDS)), problem behavior (the Nursing Home Behavior Problem Scale (NHBPS)), physical disability (the Cumulative Illness Rating Scale (CIRS)), age, gender, years of education, and the number of prescribed medications were studied.
All variables except gender and education varied significantly across level of care. Four variables made significant independent contributions; EXIT25 score (r2 = .48, P< .001), medication usage (partial r2 = .11, P<.001), sGDS score (partial r2 = .06, P = .001), and problem behavior (partial NHBPS r2 = .04, P<.04). These variables accounted for 69% of the total variance in level of care (R2 = .69; F (df 7,99) = 32.1, P<.001). A discriminant model based on the number of prescribed medications, EXIT25, sGDS, and NHBPS scores classified 83.2% of cases correctly (Wilke's lambda = .50, F(5,101) = 20.1; P<.001). The MMSE enters but fails to contribute significantly, independent of the other variables. Age and CIRS scores fail to enter.
Cognitive (particularly ECF) impairment contributes most to the observed variance in level of care received by older retirees in this CCRC. In contrast, markers of general cognition, depression, and physical illness contributed relatively little additional variance. ECF is not detected well by traditional cognitive measures and must be sought by specific tests. Further study is needed to replicate these findings in other populations.
探讨执行控制功能（ECF）、一般认知、情绪、问题行为、身体残疾、人口统计学变量以及处方药数量对老年退休人员所接受护理水平的相对贡献。
多元回归和判别建模。
德克萨斯州圣安东尼奥市的一个单一的持续照料退休社区（CCRC）。
总共107名老年退休人员（平均年龄 = 83.7±7.2岁），包括17名居住在社区的健康老年对照者和90名CCRC居民。CCRC受试者是连续接受老年精神病学评估的便利样本。61名受试者居住在非机构化护理水平，46名被机构化。
研究了ECF测试（执行访谈（EXIT25））、一般认知（简易精神状态检查表（MMSE））、情绪（老年抑郁量表简表（sGDS））、问题行为（养老院行为问题量表（NHBPS））、身体残疾（累积疾病评定量表（CIRS））、年龄、性别、受教育年限以及处方药数量。
除性别和教育程度外，所有变量在护理水平上均有显著差异。四个变量做出了显著的独立贡献；EXIT25评分（r2 = 0.48，P <.001）、药物使用情况（偏r2 = 0.11，P <.001）、sGDS评分（偏r2 = 0.06，P = 0.001）以及问题行为（偏NHBPS r2 = 0.04，P <.04）。这些变量占护理水平总方差的69%（R2 = 0.69；F（自由度7,99） = 32.1，P <.001）。基于处方药数量、EXIT25、sGDS和NHBPS评分的判别模型正确分类了83.2%的病例（威尔克斯lambda = 0.50，F(5,101) = 20.1；P <.001）。MMSE进入模型但未能独立于其他变量做出显著贡献。年龄和CIRS评分未能进入模型。
认知（尤其是ECF）损害对该CCRC中老年退休人员所接受护理水平的观察方差贡献最大。相比之下，一般认知、抑郁和身体疾病的指标贡献的额外方差相对较小。传统认知测量方法不能很好地检测到ECF，必须通过特定测试来寻找。需要进一步研究以在其他人群中复制这些发现。
