REGULAR RESEARCH ARTICLES
Development and Cross-Validation of the UPSA Short Form for the PerformanceBased Functional Assessment of Patients With Mild Cognitive Impairment and Alzheimer Disease Jesus J. Gomar, M.S., Philip D. Harvey, Ph.D., Maria T. Bobes-Bascaran, M.S., Peter Davies, Ph.D., Terry E. Goldberg, Ph.D.
Background: Functional capacity includes basic and complex behaviors necessary to independently live in the community. It has been found that patients with cognitive impairment have daily living functional skills altered at very early stages of illness. Objectives: 1) To develop and validate a brief scale derived from the University of California, San Diego, performance-based skills assessment (UPSA); 2) to crossvalidate this new UPSA short form with an independent healthy elderly sample. Method: Fifty-one healthy elderly subjects, 26 mild cognitive impairment (MCI) subjects defined per Petersen’s criteria, and 22 probable Alzheimer Disease (AD) subjects according to National Institute of Neurological and Communicative Disorders and Stroke–AD and Related Disorders Association criteria were included. For crossvalidation purpose, a comparison group of 108 older healthy subjects with MiniMental scores of 25 or greater was also recruited. A modified four-functional domain version of the UPSA was administered. Results: Communication and comprehension/planning domains accounted for almost 90% of the variance (R2 = 0.89) and in all models entered first and second, respectively. An UPSA short form using these two domains was significantly correlated with the full UPSA scale in all the groups examined: 0.86 for healthy controls; 0.87 for MCI; and 0.88 for AD. Acceptable sensitivity and specificity values for the UPSA short form were found in receiver operating characteristic (ROC) analysis. A correlation of 0.80 was found between the short and the full UPSA scales in the cross-validation sample. Conclusions: The UPSA short form is a rapid, reliable, and efficient measure of functional capacity that is able to detect performance impairment in an ecologically valid setting in much less time compared with the extended form of the scale. Furthermore, it demonstrated adequate discriminative properties among healthy subjects, MCI patients, and AD patients. (Am J Geriatr Psychiatry 2011; 19:915–922)
Received February 16, 2010; accepted July 31, 2010. From the Litwin Zucker Alzheimer’s Disease Center, Feinstein Institute, Manhassett, NY (JJG, MTB-B, PD, TEG); Benito Menni Complex Assistencial en Salut Mental, Barcelona, Spain (JJG); Centro de Investigaci´ on Biom´edica en Red de Salud Mental, CIBERSAM, Spain (JJG, MTB-B); Department of Psychiatry and Behavioral Sciences, Emory University School of Medicine, Atlanta, GA (PDH); and Servicio de Psiquiatria, Hospital Cl´ınico Universitario de Valencia, Valencia, Spain (MTB-B). Send correspondence and reprint requests to Terry E. Goldberg, Litwin Zucker Alzheimer’s Disease Center, Feinstein Institute, Manhassett, NY. e-mail:
[email protected] c 2011 American Association for Geriatric Psychiatry DOI: 10.1097/JGP.0b013e3182011846
Am J Geriatr Psychiatry 19:11, November 2011
915
Copyright © American Association for Geriatric Psychiatry. Unauthorized reproduction of this article is prohibited.
Development and Cross-Validation of the UPSA Key Words: Alzheimer disease, cognitive impairment, functional assessment, instrumental activities, mild cognitive impairment, UPSA
E
veryday function refers to the self-initiated ability to perform those basic and complex behaviors necessary to live independently in the community. Impairments on these kinds of skills are of crucial importance, because it is, along with cognitive deficits, one of the first indicators of the development of mild forms of dementia,1 and second because it is one of the criteria for the diagnosis of dementia according to the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, (DSM-IV-TR)2 and the National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer’s Disease and Related Disorders Association.3 Moreover, impairments in the ability to perform activities of daily living result in many of the economic burdens of dementia4 and are a major contributors to caregiver burden as well. Although typically basic activities of daily living (ADL) (i.e., eating, continence, dressing, and bathing) remain preserved until the progression of Alzheimer disease (AD) reaches moderate stages, recent studies have shown that the more complex instrumental ADL (cooking, using telephone, using public transportation, shopping, managing money, etc) are often altered in early stages.5,6 As poorer functional abilities are associated with higher direct costs over time, interventions may be particularly useful to reduce morbidity if they are targeted in the areas of basic and instrumental ADL.7 In addition, some reports indicate that disability on instrumental ADL is a stronger predictor of mortality than disability on ADL in a 4-year follow-up.8 Of importance for understanding morbidity associated with dementing conditions is the emergence of an abnormal cognitive condition different from the normal aging process: the concept of mild cognitive impairment (MCI). Although recognized as an heterogeneous condition,9–11 a particular subtype of MCI, the amnesic subtype, where memory impairment is the core feature, has been found to lead to an elevated rate of progression into AD.12 Individuals with MCI, in contrast to people with AD, have relatively subtle degrees of functional impairment
916
that makes them difficult to distinguish from normal functional changes associated with aging. These indices of disability are difficult to capture with informant reports. Traditionally, the approach used to assess functional capacity of people suffering from dementia has involved informant-based measures, including a widely employed questionnaire, the Alzheimer’s Disease Cooperative Study–Activities of Daily Living.5 Although informant-based measures have been widely employed, several criticisms have arisen, mainly due to the insensitivity of those measures to subtle deficits in complex instrumental outcomes at earlier stages of the illness and biased or inaccurate reports of the patient’s functional ability by the caregiver.13,14 To avoid the limitations of the informant-based measures, other methods for functional assessment have been developed. One particularly interesting method that has shown greater ecological validity is the use of performance-based measures of critical skills, in which the subject is asked to perform a variety of everyday tasks, and the quality of performance, relative to normative standards, is examined.15,16 These tasks are similar in many ways to the direct, performance-based assessment of cognitive abilities that comprise neuropsychological testing. These issues have recently been highlighted.17,18 In a previous report,19 the University of California, San Diego, performance-based skills assessment (UPSA),20 a performance-based measure of everyday living skills that was developed for use in older individuals with severe mental illness has shown acceptable psychometric properties in a mixed sample of MCI, AD patients, and healthy comparison participants. Briefly, the UPSA was found to have good psychometric properties: no ceiling and floor effects, acceptable coefficient of variation, and normality of distribution as evidenced by lack of high skewness or kurtosis. The UPSA measures functional capacity through five domains: comprehension/planning, transportation, communication, household (kitchen) tasks, and financial procedures. It takes approximately 30 minutes to administer.
Am J Geriatr Psychiatry 19:11, November 2011
Copyright © American Association for Geriatric Psychiatry. Unauthorized reproduction of this article is prohibited.
Gomar et al. The present study was carried out to accomplish two main goals. First, we aimed to develop and validate a brief scale derived from the UPSA with similar psychometric properties as compared with the longer version. The reason for this goal is that the long form of the UPSA has several subtests, such as household tasks, that require extensive utilization of props and are not amenable to assessment at field sites without a permanent assessment setup. Furthermore, some of the other UPSA domains have highly specific tasks that require adaptation to different geographic locations. The development of a shorter version could also avoid some difficulties associated to the cognitive evaluation of patients suffering from cognitive impairment: fatigue and attentional decline. Given the predictive validity of the diagnosis of some subtypes of the MCI in the evolution to the AD, early-screening tools, sensitive and quick to administer, are of importance to determine to what extent functional outcomes are compromised in the MCI. Of importance as well is the utility of a quick and reliable measure of functional disability for its use in clinical trials aimed to find potential medications for the treatment of demented conditions with an associated lack of functional capacity. Second, after obtaining a new UPSA short form and validating it with a mixed sample of the MCI and probable AD patients, we sought to cross-validate this new instrument with an independent healthy elderly sample to contrast the validity and psychometric properties of the UPSA short form with a different sample from that used to develop it.
METHODS Subjects Recruitment and demographics are described in detail in Goldberg et al.19 In brief, 26 MCI subjects defined following Petersen’s criteria,10 and 22 “probable” AD subjects according to National Institute of Neurological and Communicative Disorders and Stroke–AD and Related Disorders Association criteria3 were included. The MCI subjects were nondemented (Mini-Mental State Examination [MMSE] > 23)21 and had a Clinical Dementia Rating (CDR) score of 0.5. The CDR scores were obtained through interview of the patient and his or her relevant caregiver. The AD subjects were in the
Am J Geriatr Psychiatry 19:11, November 2011
mild-to-moderate range (12 > MMSE < 24) and had a CDR score of one or greater. In addition, 51 healthy subjects with a MMSE score equivalent or greater than 24 were recruited. For purposes of cross-validation, the results of this study were replicated in a separate sample of 108 healthy older individuals (60 women). These healthy older people were extensively screened to ensure that they did not have any current cognitive problems consistent with the MCI or dementia. They were assessed with the MMSE, and all had scores of more than 25 (mean = 28.4 ± 1.7). Mean age of the sample was 67.9 ± 11.5 years. Mean education level was 14.3 ± 2.5 years. This sample was collected as a comparison sample for a study of the course of cognitive impairment and performance-based measurement of functional capacity in older people with schizophrenia. They were all the residents of senior residential centers in Manhattan, New York. Performance-Based Functional Measure University of California, San Diego performance-based skills assessment. This is a performance-based measure of functional capacity and competence that is administered in analogue settings. We excluded the household chores subtest from the original UPSA because the analogue kitchen required would not have the portability to be used at field sites. This modified version was used in our previous reports with the UPSA.19,22,23 The scale has been validated in the AD and MCI patients.19 It consists of 27 items divided into four functional domains: comprehension/planning (six items), financial procedures (eight items), communication (eight items), and mobility (five items). Mean of percentage correct for each of the UPSA domains and a mean-derived composite score for the full scale were used as dependent variables. Statistical Analysis Strategy First, stepwise regression linear models with the functional domains of the UPSA were used to detect the best predictive model of the full UPSA scale. The UPSA full-scale score as dependent variable and the UPSA functional domain scores as independent variables were entered in the model, with age and education forced to enter in the regression model as they
917
Copyright © American Association for Geriatric Psychiatry. Unauthorized reproduction of this article is prohibited.
Development and Cross-Validation of the UPSA had been found to be related with functional outcome measures. (The same analysis was repeated without forcing age and education in the model.) Psychometric properties of the UPSA short form were analyzed by measuring ceiling effects in the comparison group (CG) and floor effects in the mixed sample of the MCI and AD patients. In addition, coefficients of variance (CV), skewness, and kurtosis were used to determine the normality of distribution of the short form for each of the three groups. Next, receiver operating characteristic (ROC) analyses were also conducted for the short form derived from the UPSA. Last, general linear models were fit to compare scores on the UPSA short form between groups adjusted by age and years of education. Finally, we sought to cross-validate the UPSA short form derived in an independent older sample. Pearson correlation coefficient was used to test whether the UPSA short form was also significantly correlated with the full UPSA in a different sample from that used to develop the short form.
RESULTS Demographic characteristics of the validation sample are shown in Table 1. A stepwise regression analysis of the functional domain scores on the full-scale score with age and education forced to enter was first conducted, and the model was fit with all the subjects. The model indicated that communication and comprehension/planning domains entered first and second at p
