Professional Documents
Culture Documents
FITZHUGH C. PANNILL
III,
PURPOSE: Although multi-disciplinary geriatric assessment of elderly patients has been shown to be effective in identifying new diagnoses and previously unknown disabilities and in decreasing hospit~iiT~tion and mortality, time and financial constraints prevent most internists and office practitioners from using this approach with their older patients. Several instruments to screen older persons for functional disability have been proposed, but there are limited data regarding their utility or effectiveness in clinical medicine. This study developed a short, patient-completed screening assessment instrument (the Functional Assessment Screen), compared it to a standard, multi-disciplinery geriatric evaluation, and determined the screening instrument's ability to predict future use of home care services in a group of elderly patients. PATIENTS AND METHODS:The screening instrum e n t was piloted retrospectively using data from patients seen in the previous 2 years at a hospital-based geriatrics clinic in Wisconsin. Using these results, a revised instrument was developed and mailed to 80 consecutive new patients who presented to the clinic for multi-disciplina~T geriatric assessment and primary care. These patients were interviewed 18 months later to determine use of home services, institutionaliT~tion, and death after the initial visit~ RESULTS: Fifty-eight of 80 eligible patients (72%) completed beth the clinic evaluation and 18-month follow-up. The patients were an elderly (mean age of 76), frail (average of three medical diagnoses), functionally disabled group (dependent in an average of 3.7 instrumental activities of d~ily living and 2.7 activities of daily
From the Department of Medicine, Yale University School of Medicine, New Haven, Connecticut, and the West Haven Veterans Administration Medical Center, West Haven, Connecticut. Dr. Pannill was a full-time employee of the Veterans Administration when this work was done. Part of this work was presented at the Annual Meeting of the Society for General Internal Medicine, Washington, D.C., 1988. Requests for reprints should be addressed to Fitzhugh C. Pannill III, M.D., West Haven Veterans Administration Medical Center/111C, 752 Campbell Avenue, West Haven, Connecticut 06516. Manuscript submitted Marl:h 15, 1990, and accepted in revised form November 19, 1990.
living). Nine of the 58 enrolled patients (15%) were institutionAliT~d, five (9%) died, and 31 (53%) required new home services after 18 months. The screening variables were sensitive but less specific than clinic providers' judgment in identifying abnormalities in social, economic, or physical health status. The relative risk of eventual home service use was elevated in patients reporting poor health status (relative risk of 3.5, 95% confidence interval [CI] 9.9 to 1"2), and dependency in housework (relative risk of 3.0, 95% CI 5.1 to 1.7), shopping (relative risk of 2.6, 95% CI 4.7 to 1.5), meals (relative risk of 2.4, 95% CI 3.4 to 1.7), dressing (relative risk of 2"2, 95% CI 3.0 to 1.6), or bathing (relative risk of 2.2, 95% CI 3.2 to 1.5). Home services were used in 16% of patients with no positive responses to a subset of four of the screening questions; usage rose to 22 % with one positive response, and to 89% (relative risk of 4.5, 95% CI 9.2 to 2.1) with two or more positive responses. CONCLUSIONS:This screening instrtunent identiffed a group of elderly patients at much higher risk for increased home service use than other patients in a geriatrics clinic. If validated in other populations, such an instrument may identify frail, elderly patients in office practice at high risk for use of home services. These patients could be targeted for more complete multi-disciplinnry geriatric assessment to identify and treat disease and disability responsible for increased service use and declining health.
5]. This process identifieselderly patients at high risk for institutionalizationand other adverse outcomes, and has been shown to decrease hospital use and mortality, compared with that in control groups [1-5]. These assessments have not been standardized but usually include an evaluation of medical conditions, functional, mental, and emotionalstatus,and economic and socialwell-beingby a multi-disciplinaryteam of physicians, nurses, socialworkers, and other professionals.These specialized assessments are usually time- and personnel-
320
March 1991
Volume 90
FUNCTIONALDISABILITYSCREENING/ PANNILL
intensive, and not reimbursed by t h i r d - p a r t y payors [2,3]. Currently, however, the vast majority of older patients who may need these evaluations receive their care in private, office-based medical practices that are unequipped to accommodate such a comprehensive 3- or 4-hour multi-disciplinary approach. Several studies [6,7] have shown office practitioners do a relatively poor job of identifying functional disabilities in their older patients. Therefore, treatable functional disabilities may go unrecognized in office practice, resulting in increased health service use, institutionalization, and mortaliW that could be prevented by geriatric assessment. A short, simple, and effective screening instrument for geriatric assessment is needed to identify functionally disabled patients who are at highest risk for health service use, institutionalization, and adverse outcomes. As a first step in designing such a screening instrument, this study had three purposes: (1) to develop a short, patient-completed,'geriatric assessment screening instrument incorporating measurements of functional and health status and economic and social resources; (2) to validate the new instrument against clinical judgment (defined as the result of a standard comprehensive geriatric assessment) in identifying functional disability, poor physical health, and poor social or economic resources; and (3) to determine the validity of the screening instrument for predicting use of home care services and future institutionalization in a group of frail patients in a geriatrics clinic.
use. The A D L s included eating, bathing, dressing, grooming, walking, toileting,and transferring ability, Each patient was rated on three summary scores of social resources, economic resources (rated by a social worker), and physical health status (rated by the nurse). These ratings consisted of a five-point scale: excellent, good, fair,moderately impaired, or poor (scales available from the author). A pilot study was designed to test the utilityof a set of eight questions for geriatricassessment developed by Pearlman [9] in a Delphi survey of 20 geriatric clinicians in six subspecialties. M y colleagues and I identified eight questions on the clinic'sassessment instrument that were identical or close approximations to these consensus questions. These eight questions were tested for validity in predicting the clinician-generated summary ratings of five areas (social resources, economic resources, physical health status, and IADL and ADL averages), using data from patients seen in the clinic between 1983 and 1985. Based on this analysis, we retained all the original questions and added one additional economic question ("luxuries"), three additional IADL questions (shopping, housework, and getting around), and two additional ADL questions (dressing and bathing). The final "Functional Assessment Screen" is shown in the Appendix. The eight pilot questions are numbers 1, 2, 3, 5, 6, 7, 8, 96, and 9-7.
Prospective Study
The sample for the prospective evaluation consisted of 80 community-living patients who presented consecutively to the geriatrics clinic from March 1, 1985, to M a y 15, 1985, for new patient evaluations and ongoing care. Patients who presented specifically for assistance in nursing home placement were excluded. T w o weeks before the firstvisit,allnew patients were mailed the one-page screening instrument with an introductory letter. Patients were requested to get assistance from family members to complete the screen if necessary. Family members were asked to complete the screen for their relatives if the appointment was for the evaluation of memory loss,confusion, or dementia. These instruments were collected by the clinicsecretary at the first visit,and not available to clinicians. The nurse and social worker performing the comprehensive clinic assessment were unaware of the study's purpose and did not see the completed screening questionnaire. The data from the comprehensive assessment were used extensively to develop plans for ongoing care and in multi-disciplinary staffing meetings held for all patients. This was standard clinic practice before and during the study. The clinicserved as the major source of pri321
mary care for almost all (90%) of the patients. Clinic staff arranged hospitalization, specialty referrals, home care services, and nursing home placement, if needed. Eighteen months after the initial clinic visit, a trained research assistant, unaware of both the screen and comprehensive clinic assessment data, attempted to contact all patients by telephone. After obtaining patient consent, the research assistant determined the patient's current living situation, use of help or services in the home, and the types and sources of both formal (by paid agencies or nonfamily members) and informal (family member or friend) daily assistance used since the patient's first clinic visit. The interviewer also inquired about hospitalizations, and changes in place and type of residence, and specifically asked if the patient needed or had used additional help at home with any dhily tasks such as housework, transportation, meals, shopping, chores, medications, personal care, walking, health problems, or money management. Patients who indicated such use were asked about the types of service and how it was provided. The families were interviewed if the patient was dead or institutionalized or had memory loss or dementia. Interviews were conducted with a standard protocol. A blinded chart review was conducted to determine medical diagnoses and hospitalizations since the first clinic visit. All provider notes were reviewed for mention of the use of formal services. Chart copies of records from visiting nursing agencies were reviewed when available. Services in place before the first clinic visit were specifically documented by the assessment on the first visit and were excluded from analysis. A specific search\was not made for services delivered at other hospitals or hospital clinics not recorded in our chart, or by providers unknown to our clinic staff or not mentioned in the interviews. As Almost all patients attended the clinic for primary care and had no other providers, information on most major services were available. The procedures and interview protocol were approved by the Clinical Investigations Committee of M o u n t Sinai Medical Center, Milwaukee, Wisconsin.
worse social or economic resources (a score of 4 or higher) or "fair" physical health or worse (a score of 4 or higher). IADL or ADL impairments were defined as the need for assistance in any two of the seven IADLs or ADLs. A formal service was defined as a service provided by a paid third party other than immediate family members for support in IADL or ADL functions or patient supervision, a living situation with such IADL or ADL assistance (senior apartment with meals, for example), or an in-home evaluation by a health care professional other than clinic staff (nurse or therapist, for example). Institutionalization was defined as nursing home placement at skilled, intermediate, or domiciliary (board and care) levels. Bivariate analysis compared the association between clinical characteristics, comprehensive prod d e r assessment, and the screen variables and the outcomes of formal service use and institutionalization. Statistical significance (p -- 0.05) was evaluated using the Mann-Whitney U test, or the t-test, as appropriate. For statistically significant variables, the relative risk was calculated to determine the association of the presence of a particular feature (i.e., living alone) with home services use and institutionalization. Relative risks are reported with 95% confidence intervals (CI) [10]. A short index of specific screening instrument questions with the highest relative risks was developed to predict forreal home service use. The variables were added incrementally, based on highest relative risk, to maximize the number of patients with service use identified.
RESULTS
Two patients were excluded from the analysis for not completing the comprehensive assessment. The characteristics of all 78 patients and their outcomes after 18 months are shown in Table I. These patients' characteristics are not significantly different from those of patients seen in the clinic in the previous 3 years and seem a suitably representative sample. Of-the 78 patients, 58 (74%) returned the screening instrument. The 20 patients who refused did not differ significantly from the group as a whole, in general demographic data, assessment results, medical diagnoses, or outcome. Data Analysis The 78 patients were a frail, elderly group with The prospective analysis examined the validity of frequent medical problems and a high prevalence of the screening instrument to predict provider-deter- functional disability. Major medical diagnoses inmined abnormalities in the corresponding areas of cluded cardiac disease (59%), dementia (26%), arthe comprehensive assessment (social, economic, thritis (19%), diabetes (18%), and other neurologic physical health, IADLs, and ADLs), and the utility disease (14%). Almost 60% were rated by the nurse of the screen to predict use of formal home services as having fair or worse physical health. Impaired and nursing home placement. Abnormal provider social and economic resources were common. The ratings were defined as "moderately impaired" or patients were dependent in 3.8 of seven IADL items
322 March 1991 The American Journal of Medicine Volume 90
and 2.2 of seven ADL items on average. Thirty-four percent had urinary incontinence. Thirty-two percent were followed in the clinic for more than 18 months, and 55% of the patients had visited the clinic within 6 months of the end of the study, indicating most were still regular clinic patients. Fortythree (55%) reported or had chart documentation of formal service use during the 18 months. Eleven of the 78 (14%) required nursing home placement during the 18 months, and seven (9%) of the patients died. These outcomes are not mutually exclusive, but of the 17 patients who died or were institutionalized, 11 (65%) had received home services. The formal services required included home health aides, day care, and other major services (Table I). Use of Meals on Wheels, a senior center, or a senior apartment were each reported by four of 78 patients (5%). Evaluations by home physical therapists or social workers, Title XIX or Title X X evaluation, or help with finances were each reported by three of the 78 patients (4%). One patient used telephone reassurance.
TABLE I Characteristics of 78 Patients on Admission to Clinic* Completed Screen (n -- 58) Mean age 76.8=1=8.3 Mean number of admissiondiagnoses 2.9 =t=1.4 Abnormal social resources 18 (31%) (moderate or greater impairment) 16 (27%) Abnormal economic resources (moderate or greater impairment) 33 (57%) Abnormal physicalhealth status (fair or worse health) Averagenumber of IADLdependencies 3.8 + 4.7 Average number of ADL dependencies 1.9 =t=2.3 Average mental status score 22.7 =t=7.3 Patient outcomes at 18 months Patients requiringformal services Patients hospitalized Patients institutionalized Deaths Home care services Home health aide Registered nurse Day care Transportation Homemaker Shoppingassistance 31 (53%) 24 (41%) 9 (15%) 5 (9%) ]7 (29%) 9 (16%) 9 (16%) 6 (10%) 5 (9%) 4 (7%) Did Not CompleteScreen (n = 20) 75.7 =E 10.1 3.1 4- 1.2 7 (35%) 7 (35%) 13 (65%) 3.7 -4-3.5 2.7 4- 3.8 21.5 =t=8. I 12 (60%) 7 (35%) 2 (10%) 2 (10%) 5 (25%) 4 (20%) 0 (0%) 0 (0%) I (5%) 2 (10%)
Validation of the Functional Assessment Screen The screening instrument was compared with the provider's rating of the patients based on the five summary areas of the comprehensive assessment. The results show that the screen was generally more sensitive than specific for the five areas of provideridentified abnormalities (Table II). Overall, the screen was 91% sensitive and 64% specific for an abnormality in any of the five areas on the providerconducted assessment. Only four of the 47 (9%) patients with one or more positive screen questions were judged by the nurse and social worker to be normal in all five areas (false positives). Four of 11 patients (36%) judged to be abnormal by the nurse or social worker had no positive answer on the screen (false negatives). Table III shows the clinical and comprehensive assessment summary variables from the provider assessment that had a statistically significant association between patients who used formal services and those who did not. The relative risk of formal service use was significantly increased (relative risk of 1.6 to 2.1) with a provider-determined ADL or IADL impairment, a provider rating of impaired social resources, or a diagnosis of dementia. Impaired physical health status was marginally significant (p = 0.03). Age, sex, number of medical diagnoses, economic resources rating, or other mddical diagnoses did not show a statistically significant relationship. Relative risks were 1.3 or less with the other variables. The screen variables closely reproduced these associations. Table IV lists the screen variables with
IADL= instrumental activitiesof dailyliving;ADL= activitiesof dailyliving. ' Resultsexpressedasmean~ SD. TABLE II Sensitivity and Specificity of Screen Items Compared with Comprehensive Provider Assessment Sensitivity Provider-determined abnormality in: Social resources Economicresources Physical health IADLs ADLs Any of the above Abbreviations asin TableI. TABLE III Relationship Between Clinical and Comprehensive Assessment Variables and Formal Service Use Number 95% of Service Use Relative Confidence Patients Number (%) Risk Interval 22 54 37 39 25 47 20 60 46 30 19 22 27 14 18 19 16 27 30 12 (86) (41) (73) (36) (72) (40) (80) (45) (65) (40) 2.1 2.0 1.8 1.8 1.6 (3.0-1.5) (3.2-1.3) (2.7-1.2) (2.5-1.3) (2.7-1.0) Specificity
72% (13/ 18) 88% (14/16) 88% (29/33) 73% (19/26) 79% (15/19) 91% (43/47)
60% (21 / 35) 30% (11/37) 43% (10/23) 74% (23/31) 67% (26/39) 64% (7/1 I)
Variable ADLs Help with >2 Help with _<2 IADLs Help with >2 Help with _<2 Social resources Impaired Normal Diagnosisof dementia Present Absent Physical health status Impaired Normal Abbreviations asinTableI.
March 1991
323
TABLEIV
Variable Health status (self-reported) Poor Normal Housework Dependent Independent Shopping Dependent Independent Meals Dependent Independent Dressing Dependent Independent Bathing Dependent Independent Bathroom Dependent Independent Hospital days >1 <1 Getting around Dependent Independent
40 16 24 34 28 30 12 46 9 49 13 45 5 53 5 53 16 42
26 3 21 10 22 9 12 19 9 22 12 19 5 26 5 26 13 18
(65) (19) (87) (29) (79) (30) (100) (41) (100) (45) (92) (42) (100) (49) (100) (49) (81) (43)
TABLE V
Multivariate Index for Determining Need for Increased Formal Services tn 58 Geriatric Patients
Number of Index Features Present* 0 1 2 3 4 2-4
Using the nine screen variables with significantly increased relative risks, an index was developed to stratify patients into groups with and without service use. The results of this analysis are shown in T a b l e V. Patients with two or more positive answers of poor self-reported health status, or assistance in shopping, housekeeping, or meals demonstrated a markedly increased likelihood of using formal services (relative risk of 4.7 to 5.9) compared to patients with no positive answers. Overall, a positive response to any two or more of the four questions identified a group that accounted for 89% of the formal service use, with an overall relative risk of 4.5 (95% CI, 9.2 to 2.1). An analysis using just the IADL variables (shopping, housekeeping, and meal preparation) demonstrated much lower relative risks of 2.4 to 2.8 corresponding to two to three positive answers. Addition of any of the other five significant screen items did not increase the accuracy of the index. Only 11 patients were institutionalized during the 18-month follow-up period, limiting the usefulness of any analysis. These patients were more likely to be dependent in IADLs and ADLs, to have a reduced physical health rating, and to suffer from dementia. No differences were seen in other clinical variables, or in any screening questionnaire variable except self-reported dependency in "getting around" (relative risk = 5.3). The infrequency of nursing home placement limited the significance of further analysis.
COMMENTS
The results indicate that the Functional Assessment Screen is a sensitive but somewhat nonspecific measurement of health provider ratings of impaired social, economic, and physical health status and functional disability in very frail patients from a geriatrics clinic. This screening questionnaire, and a subset of four questions in particular, identified frail, elderly patients who were much more likely to require home services. It is-not surprising that IADL variables were so strongly associated with home services use, as previous studies have shown significant relationships between ADL or IADL dependence and home services use [11], hospitalization [12], institutionalization [13], and mortality [12,14-16]. In a study of home service requirements in community elderly [11], ADLor IADL dependence, homebound status, social isolation, and mental status were significant predictors of utilization. The frail, clinically ill patients reported here had a much higher incidence of abnormalities than those in this community study, perhaps explaining the added importance of a global health measure--self-perceived health status--
Indexfeaturesinclude:healthtroublespreventeddoingthings,or neededassistancein anyof housework,shopping,or meals. t X2for lineartrend= 26.2, p < 0.00001.
significant association by descending relative risk of formal service use for the 58 patients who completed the screen. Poor self-reported health status, dependency in housework, shopping, meals, dressing, bathing, and bathroom, a reported hospital stay in the last 6 months, and dependency in getting around demonstrated relative risks over or close to 2.0, all statistically significant. Living alone, lack of available help, home ownership, income, luxuries, and sick days did not demonstrate significant associations. The screen as a whole had a sensitivity of 94% (29 of 31) and a specificity of 33% (nine of 27) for prediction of any formal service use.
324
and the relative lack of importance of social or economic variables. Based on our results, I A D L variables, although important, do not seem to be sufficient as the only c o m p o n e n t of a screening examination in this frail population. Several caveats are in order. Although the screen seemed to be widely accepted and was easily completed by patients or family members, "significant others" m a y not always be reliable informants about patient function [17]. Further evaluation is needed to determine ifour practice of allowing family members to answer for the patient is reasonable or makes some patients appear more dependent. While itisunlikely we missed data on home services in patients with normal functioning, thus introducing significant bias, future studies of larger, less frail populations m a y need a more comprehensive measure of home service use such as direct interviews or patient diaries. The generalizability of these results m a y be limited until the usefulness of the screen in predicting use of other.services,institutionalization,mortality, or preventable functional disabilityis studied. Its validity in other populations with a lower prevalence of disability and disease such as those in typical office practice also needs to be determined. One of the most significantobstacles to the widespread use of geriatricassessment by private, officebased practitioners m a y be the absence of a clear consensus on how this assessment process can be targeted to patients who need it the most and how to best recognize these patients in office practice. Physicians, in general, are not very good at recognizing functional disability in their own patients [6,7]. Identifying older persons at highest risk for functional deterioration and increasing need for expensive health services and institutionalizationwill require a simple, effective, and well-validated screening instrument that is not now available. This screen should (1) be sensitive to patient abnormalities, while being specific enough to avoid identifying a large number of "false positives" who need further evaluation, (2) require littleor no provider time to administer and be readily integrated into current officepractice, (3) be easilyunderstood by patients and their families and, ideally,equally valid ifcompleted either by a patient or by a family surrogate, and (4) point to different,specific interventions that can be easily introduced into office practice, or indicate which patients should be referred for more intensive assessment. The exact "target areas" of such a screening instrument are unclear. It may be difficultor impossible to design a screening questionnaire that can cover all of the areas evaluated in the usual geriatric assessment. Some, like dementia, depression, and
incontinence, have a clearly defined methodology available for evaluation and treatment that office practitioners are familiar with and can probably institute by themselves, using existing instruments [18].Others, such as "poor social resources" or "impaired shopping ability,"are markers for underlying processes or specific diseases that m a y need more in-depth, multi-disciplinary evaluation. A screening instrument m a y not be able to "cover the waterfront" and deal with all these areas at once. There are limited data, however, on the performance of screening instruments for specific diseases in ambulatory geriatricpatients [7],and even less data on screening for less specific conditions. Referral for a geriatric assessment m a y be more appropriate for all "positives." Given this uncertainty, it is not surprising that a wide variety of screening evaluations have been proposed [15,18-24],designed for different populations, target conditions, and settings. Few have been tested for validity against patient outcomes, however. Screening instruments designed solely for geriatric patients include questionnaires [15,19,21-24] and "performance-based" instruments [18,20].Fillenbaum [15],using three existing databases, found that five I A D L items--transportation, shopping, meal preparation ability, housework, and finances-demonstrated good reproducibility and construct and predictive validity in reference to IADL functioning and mortality a year later. These items were not tested prospectively or in a clinical setting. Freer [19] has proposed nine questions to use in identifying elderly patients in general practice who need more detailed assessments. These questions had many similarities with the Functional Assessment Screen described here, but no specific validation was reported. In a geriatric clinic similar to o u r s , W i l l i a m s [20] f o u n d t h a t a performance-based evaluation consisting of timing hand function was more useful than IADL or ADL variables in predicting nursing home utilization and service use. This hand function evaluation may have limited applicability in general office settings because of the need for special equipment and expert supervision. Williams proposed simpler measurements of hand function that may be more useful, but these still seem more complicated than a simple mailed questionnaire. This may be a problem with all "performance-based" instruments. Lachs et al [18] proposed a series of 11 items for identifying poor physical function, vision, hearing, IADLs, ADLs, mental status, and social functioning in older patients. The items are largely performance-based, but include simple questions and are designed to be incorporated into physicians' prac325
tices. This is a comprehensive approach trying to identify diseases (dementia, depression), functional disability, and poor resources. The validity and practicality of these items in the identification of frail, elderly patients in office practice have not been studied. Several short questionnaires have been designed to measure physical, psychologic, and social functioning in general populations [21-24], but could also have applicability to the frail elderly. Two of these include items on general health perceptions and social role functioning that would seem to have limited relevance to older adults [21,23], but the Functional Status Questionnaire (FSQ) [22,24] has an emphasis on questions for assessment of the elderly. The FSQ consists of 34 self-administered questions covering physical, psychologic, and social roles, some of which closely resemble ADL, IADL, and health evaluation items on geriatric assessments. Validation of the FSQ has shown it is a good marker of disability as measured by age, bed days, restricted activity days, and work limitations, but it has not been tested as a measure of health outcomes or death or compared to clinical judgment. The FSQ has been tested prospectively [24] and found to be acceptable to patients and physicians. Unfortunately, when physicians were given the additional , information on functional status from the FSQ, there was little difference in therapy or services prescribed or eventual patient function. This raises the serious issue of the capacity and interest of office practitioners to act on information about functional disability. It also strengthens contentions that a screen should prompt referral for specific geriatric assessment and management, ,and that reliance on nongeriatric practitioners to manage functional disabilities may be problematic. None of the proposed instruments, including our Functional Assessment Screen, can be recommended without reservations for widespread use. The approaches that have been most thoroughly validated seem either too complicated for widespread physician acceptance or, if easily accepted by physicians, have not been tested as predictors of future service use in elderly patients and may have little impact on patient care. The performance-based screen of Lachs et al [18] is properly focused on geriatric patients but needs to be validated. The data presented here establish the validity of the Functional Assessment Screen compared to provider judgment and future service use, but these conclusions are limited as to setting and population. Future evaluations of the Functional Assessment Screen need to be performed to determine accept-
ability to larger numbers of patients and physicians, reliability when comparing patient and surrogate respondents, validity in other clinical settings, especially in nongeriatric practice, and validity in predicting institutionalization, death, and other service use in larger populations with fewer frail elderly. Without these and other studies of screening instruments in general office practice, their usefulness in the mainstream of medicine will go untested and geriatric services may never be widely accepted by office generalists.
ACKNOWLEDGMENT
I wish to-thank the staff of the Geriatrics Institute, Sinai Samaritan Medical Center, Milwaukee, Wisconsin, for patient evaluations, Lee Ellington for assistance with data collection and analysis, Nancy Hetmanski for expert secretarial services, and Mary Tinetti, M.D., and Alvan Feinstein, M.D., for reviewing the manuscript.
REFERENCES
1. Williams TF. Comprehensive functional assessment: an overview. J Am Geriatr Soc ]983; 3]: 637-4]. 2. Rubenstein LZ. Comprehensive geriatric assessment. In: Solomon DH, moderator. New issues in geriatric care. Ann Intern Med 1988; 108: 718-32. 3. Rubenstein 17, Josephson KR, Wieland GD, et al. Effectiveness of a geriatric evaluation unit: a randomized clinical trial. N Engl J Med 1984; 31]: ].664-70. 4. Rubenstein 17. Geriatric assessment: an overview of its impacts. Clin Geriatr Med ]987; 3: 1-15. 5. Williams TF, Hill JG, Fairbank MF, et al. Appropriate placement of the chroni-
326
March 1991
Volume 90
March 1991
Volume 90
327