IAGS 38:1073-1081, 1990
CLINICAL INVESTIGATION
The Senior Care Studv A Controlled Trial of a Con<ative/Unit-Based Geriatric Assessment Program in Acute Care Marsha D. Fretwell, MD,*f Patricia M. Raymond, PhD, f Stephen T. McGarvey, PhD, MPH, Norma Owens, PharnzD, * Mark Traines, MD, * f Rebecca A. Silliman, MD, PhD, * f and Vincent Mor, PhDf Successful models of inpatient geriatric assessment have often involved long hospital stays, specialized interdisciplinary care, and prolonged follow-up, which are difficult to achieve within a prospective payment system. A randomized clinical trial was undertaken to evaluate the efficacy (maintenance or improvement in mental, emotional, and physical function) of using a geriatric assessment process in acute hospital care without increasing hospital charges or lengths of stay. Four hundred thirty-six patients 2 75 years of age were randomly allocated to treatment (n = 221) or control (n = 215) conditions. Patients in the treatment group were admitted to a special unit and evaluated on admission by an interdisciplinary team, which developed a care plan. Although primary care was provided by the patient’s own physician, the team fol-
lowed the patients as consultants on the unit in the hospital, and by telephone for 2 months after discharge. The control group was placed on other units and received usual hospital care. The treatment and control groups were similar at study entry. A t follow-up, there were no significant differences between the groups with respect to lengths of stay, hospital charges, mortality, change in physical function, or change in mental function. The treatment group changed more often in measured emotional function k2= 6.213, P = .045). This study indicates that it is feasible to implement consultative interdisciplinary team care in the acute-care hospital, but that its efficacy may be limited when applied to an unselected group of older patients. J Am Geriatr SOC38:1073-
he geriatric assessment model of care has become a major focus in research, education, and clinical practice in the growing field of geriatric medicine. Over the past decade, experience with the application and evaluation of geriatric assessment programs in various patient-care settings has grown, with reports of its usefulness in improving patient outcomes such as diagnostic accuracy, placement location, functional and mental status, medication use, survival, and appropriate use of health-care resources.1-8 In October 1987, a Consensus Develop-
ment Conference on comprehensive geriatric assessment, sponsored by the National Institutes of Health, concluded that the geriatric assessment process, when coupled with careful implementation of a specific care plan, is effective for selected groups of older persons.9 Although the number of units providing this service is increasing rapidly, reimbursement within the thirdparty health-care payment system has not been secured.10Thus, for the most part, the programs have been restricted to academic medical centers and Veterans’ Administration hospitals, Between November 18, 1985, and December 18, 1987, we undertook a randomized clinical trial at the Roger Williams General Hospital, a Brown University affiliate, to evaluate the impact of integrating a geriatric assessment team process that used existing hospital staff in routine acute hospital care. Our hypothesiswas that if the assessment was initiated early in the patient’s stay, utilized existing personnel, and was integrated into the everyday practice of hospital staff, an interdisciplinary geriatricassessment process could prevent the decline of the older patient’s physical, mental, and emotionalfunctions without increasing length of stay or hospital
T
From the *Roger Williams General Hospital, Division of Geriatrics and the tCenter for Gerontology and Health Care Research, Brown University, Providence, Rhode Island. Supportedin part by the National Institute of Aging, Geriahic Medicine Academic Award, grant 5 KO7 AG00249-05 (M.D.F.);the Robert Wood JohnsonFoundation, “Enhancing Hospital Care for the Older Patient,” grant 13583 (M.D.F.,S.T.M.,N.O.);and theNationa1 Center for Health Services Research and Health Care Technology Assessment, “Post-Doctoral Health Services Research Training Program in Gerontology and Geriatrics,” grant T32 HSOOOll (P.M.R.,R.A.S.). Address correspondence and reprint requests to Marsha Fretwell, MD, Roger Williams General Hospital, 825 Chalkstone Blvd., Providence, RI 02908. 0 1990 by the American Geriatrics Society
1081,1990
OOO2-8614/90/$3.50
1074 FRETWELL ET AL
charges. Mortality was expected to be similar for the two groups. METHODS Description of the Experimental Intervention The philosophy of the Senior Care Unit (SCU) was to integrate a psychosocial and functional orientation to care &thin the traditional model of patient management. Patients in the experimental or treatment group were admitted to the SCU, a regular 18-bed medical ward. Activities on this unit that distinguished it from the control units included: (1)a functional assessment by nurses within their routine admission evaluations of older patients; (2) 4-month rotations of experienced nurses as coordinators of the geriatric assessment team; and (3) three clinical-team meetings and one administrativeteam meeting per week. Patients randomized to the treatment group were evaluated by the geriatric assessment team, which included a physician specializingin geriatric medicine, the nurse coordinator, a physical therapist, a clinical pharmacist, a dietitian, and a social worker. The screening functional assessment was administered by the patient’s primary nurse and reviewed within 24 hours of admission by the nurse coordinator. During the next 48 hours, each patient was evaluated by all members of the team who, approximately 72 hours after randomization, participated in an interdisciplinary team conference facilitated by the nurse coordinator. Aside from the nursing functional assessment, no specific instruments were used for clinical evaluation. The nursing assessment included a modified Folstein MiniMental State,ll questions about mood and behavior, and the Katz Activities of Daily Living Index.12The team systematically reviewed medical diagnoses, medications, and problems in six areas of concern: nutrition, continence, cognition, emotion, mobility, and social support. The patient’s preferences concerning the use of extraordinary means of treatment were also routinely addressed. A comprehensive data base was generated, and an individualized care plan was developed. Consultation recommendations were then placed in each patient’s chart. Intervention patients were treated directly by all members of the team except the geriatrician. Action on all recommendations was at the discretion of the resident house staff and the patient’s private attending physician. Before patient discharge, an updated care plan documenting the problems that remained unresolved at discharge was prepared. The nurse coordinator provided telephone follow-up for a 2-month interval. This 2-month follow-up time was chosen to provide postdischarge monitoring for the average duration of an acute episode of illness in older patients whose lengths of stay might be shortened. Follow-up telephone calls were made weekly for 1month, and once 2 months after hospital discharge. However, patients who were unsta-
IAGS-OCTOBER 1990-VOL 38, NO.10
ble at the 1-month call received weekly calls until they were stable or the 2-month limit had been reached. After each contact, a written summary of the patient’s follow-up status was sent to the attending physician, other team members, and staff nurses. Urgent problems were communicated directly to the patient‘s physician by telephone. Subjects Patients were eligible for randomization into the study if their physician had given consent for all of his or her patients to participate, if they were at least 75 years of age, were not on protocol treatment, and, on admission, did not require coronary or intensive care. Patients also became eligible when ready to transfer out of the intensive or coronary-care units. Physician consent was obtained by contactingall attending physicians who, during the year before the study, had admitted three or more patients 75 years of age or older. One hundred percent of those contacted agreed to participate. All subjects were assured that their physicians would continue to be ”in charge” of their care whether or not they participated in the study. Patients receiving the intervention were told that they had been admitted to a special unit in the hospital where nurses were trained to care for older people. Control patients were housed on traditional medical and surgical floors and received the standard medical care of the hospital. A small number of (n = 15) control patients had consultation assessments by geriatricians, but they did not receive the organized team intervention or follow-up that was provided for the treatment patients. Patients were randomized only when both a treatment and a control bed were available and when their admission occurred between 7 A.M. on Sunday and 3 P.M. on Friday. Over the 2-year period, 4,101 patients (62.5% women) 75 years of age or older were admitted to the hospital. Because of the ”two-bed availability rule,” only 549 patients were randomized, 282 experimentals and 267 controls. Subject refusal rate during a 6-month pilot period where patients received the intervention, but no research interview, was < 1%. Because obtaining consent in the admitting area was difficult, subjects were randomized before patient consent during the 2-year study period.13 Subject flow for entry into the study is shown in Figure 1. There were no statistically significant differences between the study cohort and patients who (1)refused; (2) were unable to give consent; or (3)were randomized in error, in mean age, length of hospital stay, DRGassigned length of hospital stay, actual hospital charges, DRG-assigned reimbursement amounts, or in the proportions of men and women. Data Collection All patients were interviewed by a trained research assistant, in person, within 24 hours of admission, at 6 weeks after randomization, and by telephone at 3 months and 6 months after randomization.
IAGS-OCTOBER 1990-VOL 38, NO. 10
54 9
A TRIAL OF CONSULTATIVE GERIATRIC ASSESSMENT 1075
Patients randomized into the study.
38 refusals (17 treatment,21 control)
511
45 unable to obtain consent (25 treatment, 20 control)
r-
FIGURE 1. Subject pow for entry into the study.
4 68
436
30 randomization errors (19 treatment, 11 control)
Eligible, consenting patients (221 treatment, 215 control)
Information collected for research purposes was not shared with the caretakers of either the treatment or the control group. The structuredinterview included demographicinformation as well as a four-item version of the Mini-Mental State Exam (MMSE),11 ADLs,12 and IADLs*~ and an interview version of the Zung Self-Rating Depression Scale (SDS).15Interviewerswere trained by the first author; interrater reliability was 95% for the entire interview instrument. If a patient scored 5 8 on the MMSE, a proxy was interviewed and the SDS was omitted. The MMSE and the SDS were only administered in person on admission, discharge, and at 6 weeks after randomization. In addition to the comprehensive patient questionnaire, data were also collected from the financial office of the hospital, the patients’ medical records, care-plan compliance forms lilled out by the nurse coordinator, and from death certificatesat the Bureau of Vital Statistics of the Rhode Island Department of Health.
OUTCOME MEASURES Outcomes at Hospital Discharge Length of stay was calculated from the time of study admission to the time of hospital discharge; this included some patients with long stays in intensive care before study partiapation. Hospital charges were calculated as the dollar amount over DRG-assigned reimbursement. Propor-
tions of patients exceeding DRG reimbursement and DRG length of stay were calculated in both groups. Outcomes at Follow-Up Activities of Daily Living. The Katz index was divided into four categories- 0 - 1 dependencies, 2 - 3 dependencies, 4 - 5 dependencies, and 6 dependencies-due to a high proportion of subjects falling in the “others” category in scoring. This categorizationis highly correlated with the hierarchical approach.12 The effect of the intervention on physical functional status was examined by calculating the proportions of those in the treatment and control groups within each category at 6 weeks, 3 months, and 6 months after randomization. We tested the hypothesis that our intervention could prevent iatrogenic functional decline by dividing the sample into those who improved, those who stayed the same, and those who declined in function from baseline to the follow-up interval, and by comparingthe differences between those in the treatment group and those in the control group. Modified Mini-Mental State Exam (MMSE).The modified MMSE included items that assess orientation, serial subtraction, and immediate and delayed recall.” A person who correctly answered at least one of the three orientation questions and answered all other items correctly, was considered “intact.” The ”somewhat impaired’ group consisted of those who correctly an-
1076 FRETWELL ET AL
IAGS-OCTOBER 1990-VOL 38, NO. 10
swered at least one of the orientation questions, remembered at least one, but not all three, of the objects in delayed recall, and correctly answered all other items. The remaining patients were classified as "impaired." This coding assumes that the average older individual who is neither demented nor delirious may miss some orientation questions in the setting of acute illness and that a summary score in this scale may be insensitive to changes (i.e., improvement or decline) in mental state function. In the analysis of the effect on mental status, we compared proportions of patients in the "intact" category on the modified MMSE with those in the "somewhat impaired" and "impaired"categories. Change in mental status was then examined using a strategy similar to that described for physical function. The Zung Self-RatingDepression Scale (SDS). Only 19 of the original 20 items were administered in this interview; the SDS was considered complete if at least 17 of the 19 items were answered. A conservative cut-off point (60) for depression was selected because of the reported inflating influence of age on SDS scores.16-18 Prevention of decline by the intervention was examined by categorizing the patients into those scoring in the nondepressed range of less than 50, those in the mildly depressed group (50 - 59), and those moderately-toseverely depressed (60 - loo), and by examining for changes in mood.
RESULTS
Statistical Analysis All 436 patients who were randomized without error and who consented were included for all analyses except those involving mood. The analysis of the effect on emotional status was limited to a sample of 161 patients, 78 from the treatment group and 83 from the control group, who were admitted from home and had completed the SDS at baseline as well as at 6 weeks after randomization. The mood subsample was comparable to the total sample with respect to age, sex, race, marital status, and education, but were more independent in activities of daily living and had a higher proportion of patients without cognitive impairment. No statistical control for baseline differences was employed because the randomization of eligible subjects yielded comparable groups at baseline for functional, cognitive, and mood status, as well as key sociodemographic variables for the treatment and control groups at each interval. Simple T-tests, therefore, were used to examine between-group differences in outcomes measured by interval scales. Contingency table analyses using Pearson's x z statistic as well as the MannWhitney test of ranks were employed to examine differences in dichotomous and ordinal scale outcomes. Significance was set at conventional levels (P < .05). SystatI9and SPSS-PC Version 2.020 were the statistical packages used for all analyses.
Sample Description The sociodemographiccharacteristics of the treatment and control groups at baseline are shown at the top of Table 1. With the exception of race, there were no differences between the two groups in baseline characteristics. There were seven treatment patients who were not white, in contrast to no control patients. Because function was measured at hospital admission, the high proportion of patients with dependencies reflects both the acute episode that precipitated hospitalization and the premorbid level of functioning. Attrition Table 2 shows the effects of attrition on sample size at 6 weeks, 3 months, and 6 months after randomization. There were no differences in attrition between treatment and control groups during any of the follow-up assessment intervals. At each follow-up point, the treatment and control groups had comparable baseline sociodemographic characteristics, as well as similar cognitive, emotional, and physical functions. Compliance with the Intervention House staff participated in team meetings 20% of the time. Although 221 patients were randomized to the initial intervention, 185 (84%)were discussed at the initial interdisciplinary conference and 202 (91%) at a brief discharge conference. The compliance ratio for recommendations made at the interdisciplinary conference was 0.65; on average 9.3 f 5.3 (mean k SD) recommendations were generated per patient and 6.1 k 4.7 recommendations were implemented. Outcomes at Hospital Discharge There were no differences in hospital mortality between the treatment and control groups (Table 3). Hospital charges and lengths of stay were also comparable. The highly skewed distributions for these measures are noteworthy and stem from the very long hospitalizations of some of these frail older patients. Similar proportions from the treatment and control groups were discharged home or to nursing homes. Of those residing at home before this hospital admission, 79% of each group returned to their homes at discharge. Similarly, the majority of those coming from nursing homes were discharged back to nursing homes. Outcomes at Follow-Up Patient outcomes at the three follow-up intervals (6 weeks, 3 months, and 6 months after randomization) are shown in Table 4. No significant mortality differences were observed in any interval or in cumulative death rates. No differences in residence between the experimental and control groups were observed at any of the follow-up assessments. Approximately 55% to 60% of each group were at home at each of the intervals, with progressively fewer patients in hospitals. The remainder of patients were in nursing homes. Over 90% of those at home or in nursing homes
,..,,-vcl~LltK
A TRIAL OF CONSULTATIVE GERIATRIC ASSESSMENT
1990-VOL 38, NO. 10
1077
TABLE 1. BASELINE VARIABLES IN TREATMENT AND CONTROL PATIENTS
Demographic variables Age (mean [SD]) Sex (% female) Race (% white) Marital status (% married) Education (Y' O < high school diploma) Preadmission residence 96 from home YO from nursing home Functional status Katz ADL Index 0 - 1 dependency 2 - 3 dependencies 4 - 5 dependencies Totally dependent Mental Status Intact Somewhat impaired Impaired Emotional Status Not depressed Mildly depressed Depressed
Treatment (n = 221)
Controls (n = 215)
83.5 (5.3) 71.5 96.8 21.2 75.6
83.0 (5.7) 71.6 100 18.3 70.3
70.5 29.5
74.9 25.1
n = 215 73 (3446) 42 (2096) 72 (34%) 28 (13%) n = 199 30 (15%) 26 (1396) 143 (729/0) n = 78 38 (49%) 22 (2896) 18 (2396)
n = 214
Comparisons t
X'
1.04 0.01 5.1" 3.5 1.3 1.4
72 (34%) 40 (19%) 61 (28%) 41 (1990) n = 191 33 (1796) 27 (1496) 131 (69'Yo) n = 83 45 (54%) 22 (2696) 16 (19%)
3.4
0.52
2.0
P < .05.
at any one time reported the same type of residence at the next follow-up interval, suggesting a very stable residential status. Functional Status Comparisonsat all three intervals yielded no significant differences between the groups in their functional status (Table 4). Comparisons of functional change across all intervals also revealed no significant differences (Table 5). TABLE 2. SAMPLE ATTRITION AFTER 6-WEEK, 3-MONTH, AND 6-MONTH FOLLOW-UP
6 weeks Cumulative death Lost* Active cases 3 months Cumulative death Lost Active cases 6 months Cumulative death Lost Active cases
Treatment
Controls
32 7 182
31 9 175
47 6 162
38 8 167
57 21 135
47 18 144
* Cases lost at each interval are not presented cumulatively because patients who refused to participate at one interval were recontacted and some agreed to participate at the next follow-up.
Mental Status There were no differences between treatment and control groups at 6 weeks (Table4). There was also no significant difference between the treatment and control groups with respect to the proportions of patients whose original mental status declined, was maintained, or improved (Table 5). Emotional Status Mean scores of both groups at 6 weeks were similar and in the nondepressed range (Table 4). The proportion of the treatment and control patient groups whose mood declined or improved versus those whose mood stayed the same from baseline to six weeks was significantly different (Table 5). However, the fraction of the treatment group patients whose mood improved was not significantlydifferent from the "improvers" in the control group. Finally, those in the treatment group who were depressed at baseline (n = 18) were more likely to improve in mood than were the depressed controls (n = 16,67% versus 31%,x2 = 4.25, P = .039).
DISCUSSION Our findings confirm that a comprehensive geriatric consultation by an interdisciplinary team can be instituted within a nursing unit of a community hospital without increasing lengths of stay or hospital charges. We found no significant differences in discharge destinations for those participating in the intervention. Gay-
IAGS-OCTOBER 1990-VOL.3U,
1078 FRETWELL ET AL
IVu.
I”
TABLE 3. OUTCOMES AT HOSPITAL DISCHARGE* Outcome Variables Mortality (n) Hospital charges exceeding DRG reimbursement (mean f SD) Length of stay (mean f SD) Total hospital days Study hospital days Discharge destination (n) Prior residence: home Home Nursing home Prior residence: nursing home Home Nursing home
Treatment (n = 221)
(n = 215)
Control Treatment YO
Control Oh
22
20
10.0
9.3
$3,148 f 7,210
$4,163 k 18,406
69.7$
71.2$
12.9 f 12.9 11.6 f 12.2
14.7 f 17.4 12.8 k 15.8
61.5t
67.4t
113 30
121 33
79.0 21.0
78.6 21.4
1 40
4 52
2.4 97.6
7.1 92.9
Comparison of outcome variables by Mann- Whitney or xzshow no significant differences.
t Percent of patients exceeding DRG assigned length of stay. $ Percent of patients exceeding DRG reimbursement.
ton et a1 used a similar consultation/treatment model of care on a random sample of patients over the age of 69 admitted to the emergency room and also found no effect on lengths of stay nor in disposition at discharge.21 Our hypothesis was that the geriatric assessmentteam care process would help improve or prevent the decline of older patients’ physical, mental, and emotional function during acute-carehospitalization.No significant differences were found, however, in physical functional status (ADL) or in the mental status of the treatment and control groups. The only significant effect was the higher rate of improvement in mood of those
patients, followed up by the geriatric assessment team, who were depressed at admission. The 21.1% rate of depression observed in our sample is consistent with the most recent geriatric assessment literature.22.23 This study demonstrates for the first time that a geriatric team consultation can identify and beneficially affect older patients who are depressed as they enter the hospital for acute care. The fact that we observed no significant differences in the prevention of decline in activities of daily living or cognitive function in our study may be explained in several ways: (1)possible failure of the attending physi-
TABLE 4. PATIENT OUTCOMES AT 6 WEEKS, 3 MONTHS, AND 6 MONTHS 6 Weeks
Outcome Variables* Cumulative mortality (YO) Sample size X’ Residence (YO home) Sample size
T
C
T
C
14.5 221
14.4 215 0.00 60.3 174 2.41 50.5 174 3.45 20.0 175 1.02 18.1 83
21.3 221
17.7 215 0.90 61.1 167 0.92 52.7 167 0.48
25.8 221
21.4 215 1.17 61.1 144 0.68 65.0 143 5.72
55.8 181 43.4 182
X‘
Mental Status (YO intact) Sample size
21.4 182
X‘
Emotional Status (% depressed) Sample size X2
6 Months
Ct
X’
ADL Status (96 5 3 dependencies) Sample size
3 Months
T
16.7 78 -
56.9 162 49.4 162
56.3 135 51.1 135
0.05
The outcomes listed in this table reflect functional statusgroup differences at fixed points in time rather than measures ofchange in status over time. Mental and emotional status were only measured at 6 weeks. Differences are not statistically significant. t T, experimental or treatment; C, control patients.
A TRIAL OF CONSULTATIVE GERIATRIC ASSESSMENT 1079
[AGS-OCTOBER 1990-VOL 38, NO.10
TABLE 5. TREATMENT AND CONTROL DIFFERENCES IN FUNCTIONAL STATUS AT 6 WEEKS, 3 MONTHS, AND 6 MONTHS Baseline to 3 Months
Baseline to 6 Weeks Function Percent Change ADL status Improved Maintained Declined
T
C*
T
C
T
C
29.2 53.9 16.9
31.0 56.3 12.7 1.24
31.6 51.9 16.5
30.1 58.4 11.4 2.15
31.8 53.0 15.1
33.6 57.3 9.1 2.40
18.0 70.7 11.4
15.2 70.9 13.9 0.79
30.8 49.8 19.4
21.7 68.7 9.5 6.2$
XZ
Mental statust Improved Maintained Declined X2
Emotional status Improved Maintained Declined X2
Baseline to 6 Months
T,treatment, C, control patients.
t Mental and emotional status were only measured during the interviews that were done in person at baseline and at 6 weeks. P = .045.
cian to implement the recommendations of the consultation team; (2) contamination, i.e., improvement of the care delivered to the control group; (3) the inclusion of individuals in the study who might not have been able to respond to the interventions; (4) an insufficient intervention; and (5) insensitivity of our outcome measures to improvements that did occur. The most successfulintervention reported to date was characterized by direct control over patient care and therefore had 100% compliance.6 Because the medical, cognitive, emotional, physical, and social problems in frail older patients are often interactive, failure to comply with recommendations in one area may adversely affect outcomes in other areas. For instance, if a patient has multiple dependencies in ADLs because of depression, ordering a physical therapy consultation without treating the depression is unlikely to be effective. The treatment effect may also have been attenuated by the impact of this special care unit and team consultation on the usual care for controls in the hospital as a result of an effect on the attending and resident physicians. Both Wood-DauphineeZ4and Allen,' however, have shown that the influence of a special team in an acute-care hospital does not necessarily generalize to other areas of the hospital, even if the same personnel function in the team as well as in the control wards. Therefore, it is unlikely that our findings are limited by an inadvertent effect on the usual standard of care in the rest of the hospital. A recent review by Schmitt et a1of the methodological
problems inherent in evaluating the effect of geriatric assessment teams suggests that the effects of team care may be diluted if the subpopulations most likely to benefit from the assessment are not selected and if the team does not focus on the specific treatable problems most likely to be present in the selected population.= The large numbers of patients in our sample who were either too able or too dependent in ADLs to respond to an intervention focused on improvement could have washed out the effect of the intervention on those in the midband of ADL function with more potential for change. In the most successful controlled clinical trial, the patients were a carefully targeted group of men in a hospital setting with prolonged lengths of stay. Our patients had a much more limited exposure to the intervention, with an average length of stay of 12 days as compared to 85.1 for geriatric assessment unit patients in the report by Rubenstein et a1.6 Another limitation in the design of our study may have been our lack of control over posthospital care. The brief follow-up in our study, which focused only on the acute episode of illness, may have limited the possibility of a significant impact on the level of functional dependence for those patients whose functional impairments were chronic in nature. Finally, there was a strong consensus among the team members that certain meaningful improvements in patient functions did occur but were imperceptible by the Katz Index and the MMSE measures. For instance, in nursing home patients, control of disruptive behavior
IAGS-OCTOBER 1990-VOL. 38, NO. 10
1080 FRETWELL ET AL
was often achieved, but was not reflected in measurable change in mental status. Table 5 shows that the patients receiving the intervention were more likely to change (both to improve and to decline) in emotional status then were controls; the bidirectional response makes the significance of the x 2 problematic. This concern is increased in light of the fact that the fraction of intervention subjects that showed improvement in mood was no different from that in the control group. The finding that study patients who were severely depressed at admission and received the team intervention were more likely to improve than the depressed people in the control group suggests that the geriatric assessment team did effectively identify and treat depression in acutely ill older patients. On the other hand, some of those who were not depressed, or only mildly so, in the treatment group experienced a decline in emotional status during the study. Given baseline comparability of the total group receiving the Zung and the subgroup of study patients who were very depressed at admission, and assuming that the effect of any measurement error on the change scores is equal, there may be differences in the effectiveness of our geriatric assessment process for different groups of patients. Without an intensive follow-up component, the team would have been unable to prevent decline of mood function occumng at the end of hospitalization or after discharge. Failure of the team to identify depression after the initial team conference or poorer compliance by the patient’s physician with team recommendations for treatment of the depression that develops during hospitalization (often termed “situational” or ”reactive” depression) are also possible explanations. Supporting the latter explanation was the finding that the compliance rate among depressed treatment patients was 83% compared to an overall treatment compliance rate of 65%. One final explanation for an apparent treatment effect in the area of emotional status was the effect of our decision to limit the application of the Zung instrument to only those study patients who had intact cognitive function. That process selected a group of individuals who were from home and had better physical and cognitive function than the overall study population. In other words, this criterion “targeted” (for the measurement of mood outcomes) the group for whom the intervention of geriatric assessment has been most consistently effective, when evaluated in randomized controlled trials.21,23,27 Future analyses of our study data will focus on the identification of the following groups: (1)those patients at risk for functional decline in the acute-care hospital, and (2) within this high-risk group, those who are most likely to benefit from the team consultation. The next chapter of geriatric assessment research should focus on the development of successful protocols for multidi-
mensional screening, assessment and treatment plans for frail patients with common profiles of functional impairments and medical problems, and the improvement of compliance with these treatment plans. Integration of the primary-care physician and nurse into the structure and process of interdisciplinary assessment and care teams may provide the most efficient and effective application of this comprehensive health-care intervention.26 REFERENCES 1. Allen CM, Becker PM, McVey LJ,et al: A randomized, controlled clinical trial of a geriatric consultation team: compliance with recommendations. JAMA 255:2617,1986 2. Rubenstein LZ,Josephson K, Wieland GD, et al: Geriatricassessment on a subacute hospital ward. Clin Geriatr Med 3:131,1987 3. Tulloch AJ, Moore V A randomized controlled trial of geriatric screening and surveillancein general practice. J R Coll Gen Pract 29:733, 1979 4. Williams ME: Outpatient geriatric evaluation. Clin Geriatr Med 3:175, 1987 5. Hendriksen C, Lund E, Stromgard E: Consequences of assessment and intervention among elderly people: three-year randomised controlled trial. Br Med J 289:1522,1984 6. Rubenstein LZ,Josephson KR, Wieland GD, et al: Effectiveness of a geriatric evaluation unit: a randomized clinical trial. N Engl J Med 311:1664,1984 7. Collard AF, Bachman SS, Beatrice D F Acute care delivery for the geriatric patient: an innovative approach. Quality Rev Bull 2(6):180, 1985 8 . Vetter NJ, Jones DA, Victor CR: Effect of health visitors working with elderly patients in general practice: a randomised controlled trial. Br Med J 288:369,1984 9. National Institutes of Health Consensus Development Conference Statement: Geriatric assessment methods for clinical decision-making (Consensus Development Panel, David Soloman, MD, Chairman). J Am Geriatr Soc 36:342,1988 10. Epstein AM, Hall JA, Besdine R, et al: The emergenceof geriatric assessment units: the ”new technologyof geriatrics.“Ann Intern Med 106(2):299, 1987 11. Klein LE, Roca RP, McArthur J, et al: Diagnosing dementia: u N variate and multivariate analyses of the mental status examination. J Am Geriatr Soc 33:483, 1985 12. Katz S,Akpom C A A measure of primary sociobiologicalfunctions. Int J Health Serv 6:493,1976 13. Glen M: A new design for randomized clinical trials. N Engl J Med 300:1242, 1979 14. Lawton P, Moss M, Fulcomer M, Lenban MH: Research and service oriented multilevel assessment instrument. J Gerontol 3791, 1982 15. Zung WWK: A self-rating depression scale. Arch Gen Psychiatry 1263, 1965 16. Okimoto JT,Barnes RF, Veith RC, et al: Screeningfor depression in geriatric medical patients. Am J Psychiatry 139:799, 1982 17. Kitchell MA, Barnes RF, Veith RC. et al: Screeningfor depression in hospitalized geriatric medical patients. J Am Geriatr Soc 30:174, 1982 18. Magni G, DeLeo D, Schifano F Depression in geriatric and adult medical inpatients. J Clin Psycho1 41:337,1985 19. Wilkinson L: SYSTAT The system for statistics. Evanston, IL, SYSTAT Inc, 1987 20. SPSS Inc: SPSS/PC V2.0 Base Manual. Chicago, SPSS, Inc., 1988. 21. Gayton D, Wood-Dauphinee S, de Lorimer M, et al: Trial of a geriatricconsultation team in an acute care hospital. J Am Geriatr Soc 35:726, 1987 22. Katz PR, Dube DH, Calkins E: Use of a structured functional assessment format in a geriatric consultative service. J Am Geriatr Soc 33:681, 1985 23. Becker PM, McVey LJ, Saltz CC, et al: Hospital-acquiredcompli-
+
IAGS-OCTOBER 1990-VOL 38, NO. 10
A TRIAL O F CONSULTATIVE GERIATRIC ASSESSMENT
cations in a randomized controlled clinical trial of a geriatric consultation team. JAMA2572313,1987 24. Wood-Dauphinee S, Shapiro S, Bass E,et al: A randomized trial of team care following stroke. Stroke 15:864,1984 25. Schmitt MH, Famll MP, Heinemann GD. Conceptual andmethodological problems in studying the effects of interdisciplinary geriatric teams. Gerontologist 28:753, 1988 26. Campion EW, Jette A, Berkman B An interdisciplinary geriatric
1081
consultative service: a controlled trial. J Am Geriatr Soc 31:792, 1983 27. Hogan DB, Fox
RA, Badley BWD, et al: Effect of a geriatric consultative service on management of patients in an acute care hospital. Can Med Assoc J 136:713, 1987 28. Fretwell MD: The consensus conference on comprehensive geriatric assessment: a dialogue is the beginning of consensus. J Am Geriatr Soc 36377,1988
CLINICAL INVESTIGATION
The Senior Care Studv A Controlled Trial of a Con<ative/Unit-Based Geriatric Assessment Program in Acute Care Marsha D. Fretwell, MD,*f Patricia M. Raymond, PhD, f Stephen T. McGarvey, PhD, MPH, Norma Owens, PharnzD, * Mark Traines, MD, * f Rebecca A. Silliman, MD, PhD, * f and Vincent Mor, PhDf Successful models of inpatient geriatric assessment have often involved long hospital stays, specialized interdisciplinary care, and prolonged follow-up, which are difficult to achieve within a prospective payment system. A randomized clinical trial was undertaken to evaluate the efficacy (maintenance or improvement in mental, emotional, and physical function) of using a geriatric assessment process in acute hospital care without increasing hospital charges or lengths of stay. Four hundred thirty-six patients 2 75 years of age were randomly allocated to treatment (n = 221) or control (n = 215) conditions. Patients in the treatment group were admitted to a special unit and evaluated on admission by an interdisciplinary team, which developed a care plan. Although primary care was provided by the patient’s own physician, the team fol-
lowed the patients as consultants on the unit in the hospital, and by telephone for 2 months after discharge. The control group was placed on other units and received usual hospital care. The treatment and control groups were similar at study entry. A t follow-up, there were no significant differences between the groups with respect to lengths of stay, hospital charges, mortality, change in physical function, or change in mental function. The treatment group changed more often in measured emotional function k2= 6.213, P = .045). This study indicates that it is feasible to implement consultative interdisciplinary team care in the acute-care hospital, but that its efficacy may be limited when applied to an unselected group of older patients. J Am Geriatr SOC38:1073-
he geriatric assessment model of care has become a major focus in research, education, and clinical practice in the growing field of geriatric medicine. Over the past decade, experience with the application and evaluation of geriatric assessment programs in various patient-care settings has grown, with reports of its usefulness in improving patient outcomes such as diagnostic accuracy, placement location, functional and mental status, medication use, survival, and appropriate use of health-care resources.1-8 In October 1987, a Consensus Develop-
ment Conference on comprehensive geriatric assessment, sponsored by the National Institutes of Health, concluded that the geriatric assessment process, when coupled with careful implementation of a specific care plan, is effective for selected groups of older persons.9 Although the number of units providing this service is increasing rapidly, reimbursement within the thirdparty health-care payment system has not been secured.10Thus, for the most part, the programs have been restricted to academic medical centers and Veterans’ Administration hospitals, Between November 18, 1985, and December 18, 1987, we undertook a randomized clinical trial at the Roger Williams General Hospital, a Brown University affiliate, to evaluate the impact of integrating a geriatric assessment team process that used existing hospital staff in routine acute hospital care. Our hypothesiswas that if the assessment was initiated early in the patient’s stay, utilized existing personnel, and was integrated into the everyday practice of hospital staff, an interdisciplinary geriatricassessment process could prevent the decline of the older patient’s physical, mental, and emotionalfunctions without increasing length of stay or hospital
T
From the *Roger Williams General Hospital, Division of Geriatrics and the tCenter for Gerontology and Health Care Research, Brown University, Providence, Rhode Island. Supportedin part by the National Institute of Aging, Geriahic Medicine Academic Award, grant 5 KO7 AG00249-05 (M.D.F.);the Robert Wood JohnsonFoundation, “Enhancing Hospital Care for the Older Patient,” grant 13583 (M.D.F.,S.T.M.,N.O.);and theNationa1 Center for Health Services Research and Health Care Technology Assessment, “Post-Doctoral Health Services Research Training Program in Gerontology and Geriatrics,” grant T32 HSOOOll (P.M.R.,R.A.S.). Address correspondence and reprint requests to Marsha Fretwell, MD, Roger Williams General Hospital, 825 Chalkstone Blvd., Providence, RI 02908. 0 1990 by the American Geriatrics Society
1081,1990
OOO2-8614/90/$3.50
1074 FRETWELL ET AL
charges. Mortality was expected to be similar for the two groups. METHODS Description of the Experimental Intervention The philosophy of the Senior Care Unit (SCU) was to integrate a psychosocial and functional orientation to care &thin the traditional model of patient management. Patients in the experimental or treatment group were admitted to the SCU, a regular 18-bed medical ward. Activities on this unit that distinguished it from the control units included: (1)a functional assessment by nurses within their routine admission evaluations of older patients; (2) 4-month rotations of experienced nurses as coordinators of the geriatric assessment team; and (3) three clinical-team meetings and one administrativeteam meeting per week. Patients randomized to the treatment group were evaluated by the geriatric assessment team, which included a physician specializingin geriatric medicine, the nurse coordinator, a physical therapist, a clinical pharmacist, a dietitian, and a social worker. The screening functional assessment was administered by the patient’s primary nurse and reviewed within 24 hours of admission by the nurse coordinator. During the next 48 hours, each patient was evaluated by all members of the team who, approximately 72 hours after randomization, participated in an interdisciplinary team conference facilitated by the nurse coordinator. Aside from the nursing functional assessment, no specific instruments were used for clinical evaluation. The nursing assessment included a modified Folstein MiniMental State,ll questions about mood and behavior, and the Katz Activities of Daily Living Index.12The team systematically reviewed medical diagnoses, medications, and problems in six areas of concern: nutrition, continence, cognition, emotion, mobility, and social support. The patient’s preferences concerning the use of extraordinary means of treatment were also routinely addressed. A comprehensive data base was generated, and an individualized care plan was developed. Consultation recommendations were then placed in each patient’s chart. Intervention patients were treated directly by all members of the team except the geriatrician. Action on all recommendations was at the discretion of the resident house staff and the patient’s private attending physician. Before patient discharge, an updated care plan documenting the problems that remained unresolved at discharge was prepared. The nurse coordinator provided telephone follow-up for a 2-month interval. This 2-month follow-up time was chosen to provide postdischarge monitoring for the average duration of an acute episode of illness in older patients whose lengths of stay might be shortened. Follow-up telephone calls were made weekly for 1month, and once 2 months after hospital discharge. However, patients who were unsta-
IAGS-OCTOBER 1990-VOL 38, NO.10
ble at the 1-month call received weekly calls until they were stable or the 2-month limit had been reached. After each contact, a written summary of the patient’s follow-up status was sent to the attending physician, other team members, and staff nurses. Urgent problems were communicated directly to the patient‘s physician by telephone. Subjects Patients were eligible for randomization into the study if their physician had given consent for all of his or her patients to participate, if they were at least 75 years of age, were not on protocol treatment, and, on admission, did not require coronary or intensive care. Patients also became eligible when ready to transfer out of the intensive or coronary-care units. Physician consent was obtained by contactingall attending physicians who, during the year before the study, had admitted three or more patients 75 years of age or older. One hundred percent of those contacted agreed to participate. All subjects were assured that their physicians would continue to be ”in charge” of their care whether or not they participated in the study. Patients receiving the intervention were told that they had been admitted to a special unit in the hospital where nurses were trained to care for older people. Control patients were housed on traditional medical and surgical floors and received the standard medical care of the hospital. A small number of (n = 15) control patients had consultation assessments by geriatricians, but they did not receive the organized team intervention or follow-up that was provided for the treatment patients. Patients were randomized only when both a treatment and a control bed were available and when their admission occurred between 7 A.M. on Sunday and 3 P.M. on Friday. Over the 2-year period, 4,101 patients (62.5% women) 75 years of age or older were admitted to the hospital. Because of the ”two-bed availability rule,” only 549 patients were randomized, 282 experimentals and 267 controls. Subject refusal rate during a 6-month pilot period where patients received the intervention, but no research interview, was < 1%. Because obtaining consent in the admitting area was difficult, subjects were randomized before patient consent during the 2-year study period.13 Subject flow for entry into the study is shown in Figure 1. There were no statistically significant differences between the study cohort and patients who (1)refused; (2) were unable to give consent; or (3)were randomized in error, in mean age, length of hospital stay, DRGassigned length of hospital stay, actual hospital charges, DRG-assigned reimbursement amounts, or in the proportions of men and women. Data Collection All patients were interviewed by a trained research assistant, in person, within 24 hours of admission, at 6 weeks after randomization, and by telephone at 3 months and 6 months after randomization.
IAGS-OCTOBER 1990-VOL 38, NO. 10
54 9
A TRIAL OF CONSULTATIVE GERIATRIC ASSESSMENT 1075
Patients randomized into the study.
38 refusals (17 treatment,21 control)
511
45 unable to obtain consent (25 treatment, 20 control)
r-
FIGURE 1. Subject pow for entry into the study.
4 68
436
30 randomization errors (19 treatment, 11 control)
Eligible, consenting patients (221 treatment, 215 control)
Information collected for research purposes was not shared with the caretakers of either the treatment or the control group. The structuredinterview included demographicinformation as well as a four-item version of the Mini-Mental State Exam (MMSE),11 ADLs,12 and IADLs*~ and an interview version of the Zung Self-Rating Depression Scale (SDS).15Interviewerswere trained by the first author; interrater reliability was 95% for the entire interview instrument. If a patient scored 5 8 on the MMSE, a proxy was interviewed and the SDS was omitted. The MMSE and the SDS were only administered in person on admission, discharge, and at 6 weeks after randomization. In addition to the comprehensive patient questionnaire, data were also collected from the financial office of the hospital, the patients’ medical records, care-plan compliance forms lilled out by the nurse coordinator, and from death certificatesat the Bureau of Vital Statistics of the Rhode Island Department of Health.
OUTCOME MEASURES Outcomes at Hospital Discharge Length of stay was calculated from the time of study admission to the time of hospital discharge; this included some patients with long stays in intensive care before study partiapation. Hospital charges were calculated as the dollar amount over DRG-assigned reimbursement. Propor-
tions of patients exceeding DRG reimbursement and DRG length of stay were calculated in both groups. Outcomes at Follow-Up Activities of Daily Living. The Katz index was divided into four categories- 0 - 1 dependencies, 2 - 3 dependencies, 4 - 5 dependencies, and 6 dependencies-due to a high proportion of subjects falling in the “others” category in scoring. This categorizationis highly correlated with the hierarchical approach.12 The effect of the intervention on physical functional status was examined by calculating the proportions of those in the treatment and control groups within each category at 6 weeks, 3 months, and 6 months after randomization. We tested the hypothesis that our intervention could prevent iatrogenic functional decline by dividing the sample into those who improved, those who stayed the same, and those who declined in function from baseline to the follow-up interval, and by comparingthe differences between those in the treatment group and those in the control group. Modified Mini-Mental State Exam (MMSE).The modified MMSE included items that assess orientation, serial subtraction, and immediate and delayed recall.” A person who correctly answered at least one of the three orientation questions and answered all other items correctly, was considered “intact.” The ”somewhat impaired’ group consisted of those who correctly an-
1076 FRETWELL ET AL
IAGS-OCTOBER 1990-VOL 38, NO. 10
swered at least one of the orientation questions, remembered at least one, but not all three, of the objects in delayed recall, and correctly answered all other items. The remaining patients were classified as "impaired." This coding assumes that the average older individual who is neither demented nor delirious may miss some orientation questions in the setting of acute illness and that a summary score in this scale may be insensitive to changes (i.e., improvement or decline) in mental state function. In the analysis of the effect on mental status, we compared proportions of patients in the "intact" category on the modified MMSE with those in the "somewhat impaired" and "impaired"categories. Change in mental status was then examined using a strategy similar to that described for physical function. The Zung Self-RatingDepression Scale (SDS). Only 19 of the original 20 items were administered in this interview; the SDS was considered complete if at least 17 of the 19 items were answered. A conservative cut-off point (60) for depression was selected because of the reported inflating influence of age on SDS scores.16-18 Prevention of decline by the intervention was examined by categorizing the patients into those scoring in the nondepressed range of less than 50, those in the mildly depressed group (50 - 59), and those moderately-toseverely depressed (60 - loo), and by examining for changes in mood.
RESULTS
Statistical Analysis All 436 patients who were randomized without error and who consented were included for all analyses except those involving mood. The analysis of the effect on emotional status was limited to a sample of 161 patients, 78 from the treatment group and 83 from the control group, who were admitted from home and had completed the SDS at baseline as well as at 6 weeks after randomization. The mood subsample was comparable to the total sample with respect to age, sex, race, marital status, and education, but were more independent in activities of daily living and had a higher proportion of patients without cognitive impairment. No statistical control for baseline differences was employed because the randomization of eligible subjects yielded comparable groups at baseline for functional, cognitive, and mood status, as well as key sociodemographic variables for the treatment and control groups at each interval. Simple T-tests, therefore, were used to examine between-group differences in outcomes measured by interval scales. Contingency table analyses using Pearson's x z statistic as well as the MannWhitney test of ranks were employed to examine differences in dichotomous and ordinal scale outcomes. Significance was set at conventional levels (P < .05). SystatI9and SPSS-PC Version 2.020 were the statistical packages used for all analyses.
Sample Description The sociodemographiccharacteristics of the treatment and control groups at baseline are shown at the top of Table 1. With the exception of race, there were no differences between the two groups in baseline characteristics. There were seven treatment patients who were not white, in contrast to no control patients. Because function was measured at hospital admission, the high proportion of patients with dependencies reflects both the acute episode that precipitated hospitalization and the premorbid level of functioning. Attrition Table 2 shows the effects of attrition on sample size at 6 weeks, 3 months, and 6 months after randomization. There were no differences in attrition between treatment and control groups during any of the follow-up assessment intervals. At each follow-up point, the treatment and control groups had comparable baseline sociodemographic characteristics, as well as similar cognitive, emotional, and physical functions. Compliance with the Intervention House staff participated in team meetings 20% of the time. Although 221 patients were randomized to the initial intervention, 185 (84%)were discussed at the initial interdisciplinary conference and 202 (91%) at a brief discharge conference. The compliance ratio for recommendations made at the interdisciplinary conference was 0.65; on average 9.3 f 5.3 (mean k SD) recommendations were generated per patient and 6.1 k 4.7 recommendations were implemented. Outcomes at Hospital Discharge There were no differences in hospital mortality between the treatment and control groups (Table 3). Hospital charges and lengths of stay were also comparable. The highly skewed distributions for these measures are noteworthy and stem from the very long hospitalizations of some of these frail older patients. Similar proportions from the treatment and control groups were discharged home or to nursing homes. Of those residing at home before this hospital admission, 79% of each group returned to their homes at discharge. Similarly, the majority of those coming from nursing homes were discharged back to nursing homes. Outcomes at Follow-Up Patient outcomes at the three follow-up intervals (6 weeks, 3 months, and 6 months after randomization) are shown in Table 4. No significant mortality differences were observed in any interval or in cumulative death rates. No differences in residence between the experimental and control groups were observed at any of the follow-up assessments. Approximately 55% to 60% of each group were at home at each of the intervals, with progressively fewer patients in hospitals. The remainder of patients were in nursing homes. Over 90% of those at home or in nursing homes
,..,,-vcl~LltK
A TRIAL OF CONSULTATIVE GERIATRIC ASSESSMENT
1990-VOL 38, NO. 10
1077
TABLE 1. BASELINE VARIABLES IN TREATMENT AND CONTROL PATIENTS
Demographic variables Age (mean [SD]) Sex (% female) Race (% white) Marital status (% married) Education (Y' O < high school diploma) Preadmission residence 96 from home YO from nursing home Functional status Katz ADL Index 0 - 1 dependency 2 - 3 dependencies 4 - 5 dependencies Totally dependent Mental Status Intact Somewhat impaired Impaired Emotional Status Not depressed Mildly depressed Depressed
Treatment (n = 221)
Controls (n = 215)
83.5 (5.3) 71.5 96.8 21.2 75.6
83.0 (5.7) 71.6 100 18.3 70.3
70.5 29.5
74.9 25.1
n = 215 73 (3446) 42 (2096) 72 (34%) 28 (13%) n = 199 30 (15%) 26 (1396) 143 (729/0) n = 78 38 (49%) 22 (2896) 18 (2396)
n = 214
Comparisons t
X'
1.04 0.01 5.1" 3.5 1.3 1.4
72 (34%) 40 (19%) 61 (28%) 41 (1990) n = 191 33 (1796) 27 (1496) 131 (69'Yo) n = 83 45 (54%) 22 (2696) 16 (19%)
3.4
0.52
2.0
P < .05.
at any one time reported the same type of residence at the next follow-up interval, suggesting a very stable residential status. Functional Status Comparisonsat all three intervals yielded no significant differences between the groups in their functional status (Table 4). Comparisons of functional change across all intervals also revealed no significant differences (Table 5). TABLE 2. SAMPLE ATTRITION AFTER 6-WEEK, 3-MONTH, AND 6-MONTH FOLLOW-UP
6 weeks Cumulative death Lost* Active cases 3 months Cumulative death Lost Active cases 6 months Cumulative death Lost Active cases
Treatment
Controls
32 7 182
31 9 175
47 6 162
38 8 167
57 21 135
47 18 144
* Cases lost at each interval are not presented cumulatively because patients who refused to participate at one interval were recontacted and some agreed to participate at the next follow-up.
Mental Status There were no differences between treatment and control groups at 6 weeks (Table4). There was also no significant difference between the treatment and control groups with respect to the proportions of patients whose original mental status declined, was maintained, or improved (Table 5). Emotional Status Mean scores of both groups at 6 weeks were similar and in the nondepressed range (Table 4). The proportion of the treatment and control patient groups whose mood declined or improved versus those whose mood stayed the same from baseline to six weeks was significantly different (Table 5). However, the fraction of the treatment group patients whose mood improved was not significantlydifferent from the "improvers" in the control group. Finally, those in the treatment group who were depressed at baseline (n = 18) were more likely to improve in mood than were the depressed controls (n = 16,67% versus 31%,x2 = 4.25, P = .039).
DISCUSSION Our findings confirm that a comprehensive geriatric consultation by an interdisciplinary team can be instituted within a nursing unit of a community hospital without increasing lengths of stay or hospital charges. We found no significant differences in discharge destinations for those participating in the intervention. Gay-
IAGS-OCTOBER 1990-VOL.3U,
1078 FRETWELL ET AL
IVu.
I”
TABLE 3. OUTCOMES AT HOSPITAL DISCHARGE* Outcome Variables Mortality (n) Hospital charges exceeding DRG reimbursement (mean f SD) Length of stay (mean f SD) Total hospital days Study hospital days Discharge destination (n) Prior residence: home Home Nursing home Prior residence: nursing home Home Nursing home
Treatment (n = 221)
(n = 215)
Control Treatment YO
Control Oh
22
20
10.0
9.3
$3,148 f 7,210
$4,163 k 18,406
69.7$
71.2$
12.9 f 12.9 11.6 f 12.2
14.7 f 17.4 12.8 k 15.8
61.5t
67.4t
113 30
121 33
79.0 21.0
78.6 21.4
1 40
4 52
2.4 97.6
7.1 92.9
Comparison of outcome variables by Mann- Whitney or xzshow no significant differences.
t Percent of patients exceeding DRG assigned length of stay. $ Percent of patients exceeding DRG reimbursement.
ton et a1 used a similar consultation/treatment model of care on a random sample of patients over the age of 69 admitted to the emergency room and also found no effect on lengths of stay nor in disposition at discharge.21 Our hypothesis was that the geriatric assessmentteam care process would help improve or prevent the decline of older patients’ physical, mental, and emotional function during acute-carehospitalization.No significant differences were found, however, in physical functional status (ADL) or in the mental status of the treatment and control groups. The only significant effect was the higher rate of improvement in mood of those
patients, followed up by the geriatric assessment team, who were depressed at admission. The 21.1% rate of depression observed in our sample is consistent with the most recent geriatric assessment literature.22.23 This study demonstrates for the first time that a geriatric team consultation can identify and beneficially affect older patients who are depressed as they enter the hospital for acute care. The fact that we observed no significant differences in the prevention of decline in activities of daily living or cognitive function in our study may be explained in several ways: (1)possible failure of the attending physi-
TABLE 4. PATIENT OUTCOMES AT 6 WEEKS, 3 MONTHS, AND 6 MONTHS 6 Weeks
Outcome Variables* Cumulative mortality (YO) Sample size X’ Residence (YO home) Sample size
T
C
T
C
14.5 221
14.4 215 0.00 60.3 174 2.41 50.5 174 3.45 20.0 175 1.02 18.1 83
21.3 221
17.7 215 0.90 61.1 167 0.92 52.7 167 0.48
25.8 221
21.4 215 1.17 61.1 144 0.68 65.0 143 5.72
55.8 181 43.4 182
X‘
Mental Status (YO intact) Sample size
21.4 182
X‘
Emotional Status (% depressed) Sample size X2
6 Months
Ct
X’
ADL Status (96 5 3 dependencies) Sample size
3 Months
T
16.7 78 -
56.9 162 49.4 162
56.3 135 51.1 135
0.05
The outcomes listed in this table reflect functional statusgroup differences at fixed points in time rather than measures ofchange in status over time. Mental and emotional status were only measured at 6 weeks. Differences are not statistically significant. t T, experimental or treatment; C, control patients.
A TRIAL OF CONSULTATIVE GERIATRIC ASSESSMENT 1079
[AGS-OCTOBER 1990-VOL 38, NO.10
TABLE 5. TREATMENT AND CONTROL DIFFERENCES IN FUNCTIONAL STATUS AT 6 WEEKS, 3 MONTHS, AND 6 MONTHS Baseline to 3 Months
Baseline to 6 Weeks Function Percent Change ADL status Improved Maintained Declined
T
C*
T
C
T
C
29.2 53.9 16.9
31.0 56.3 12.7 1.24
31.6 51.9 16.5
30.1 58.4 11.4 2.15
31.8 53.0 15.1
33.6 57.3 9.1 2.40
18.0 70.7 11.4
15.2 70.9 13.9 0.79
30.8 49.8 19.4
21.7 68.7 9.5 6.2$
XZ
Mental statust Improved Maintained Declined X2
Emotional status Improved Maintained Declined X2
Baseline to 6 Months
T,treatment, C, control patients.
t Mental and emotional status were only measured during the interviews that were done in person at baseline and at 6 weeks. P = .045.
cian to implement the recommendations of the consultation team; (2) contamination, i.e., improvement of the care delivered to the control group; (3) the inclusion of individuals in the study who might not have been able to respond to the interventions; (4) an insufficient intervention; and (5) insensitivity of our outcome measures to improvements that did occur. The most successfulintervention reported to date was characterized by direct control over patient care and therefore had 100% compliance.6 Because the medical, cognitive, emotional, physical, and social problems in frail older patients are often interactive, failure to comply with recommendations in one area may adversely affect outcomes in other areas. For instance, if a patient has multiple dependencies in ADLs because of depression, ordering a physical therapy consultation without treating the depression is unlikely to be effective. The treatment effect may also have been attenuated by the impact of this special care unit and team consultation on the usual care for controls in the hospital as a result of an effect on the attending and resident physicians. Both Wood-DauphineeZ4and Allen,' however, have shown that the influence of a special team in an acute-care hospital does not necessarily generalize to other areas of the hospital, even if the same personnel function in the team as well as in the control wards. Therefore, it is unlikely that our findings are limited by an inadvertent effect on the usual standard of care in the rest of the hospital. A recent review by Schmitt et a1of the methodological
problems inherent in evaluating the effect of geriatric assessment teams suggests that the effects of team care may be diluted if the subpopulations most likely to benefit from the assessment are not selected and if the team does not focus on the specific treatable problems most likely to be present in the selected population.= The large numbers of patients in our sample who were either too able or too dependent in ADLs to respond to an intervention focused on improvement could have washed out the effect of the intervention on those in the midband of ADL function with more potential for change. In the most successful controlled clinical trial, the patients were a carefully targeted group of men in a hospital setting with prolonged lengths of stay. Our patients had a much more limited exposure to the intervention, with an average length of stay of 12 days as compared to 85.1 for geriatric assessment unit patients in the report by Rubenstein et a1.6 Another limitation in the design of our study may have been our lack of control over posthospital care. The brief follow-up in our study, which focused only on the acute episode of illness, may have limited the possibility of a significant impact on the level of functional dependence for those patients whose functional impairments were chronic in nature. Finally, there was a strong consensus among the team members that certain meaningful improvements in patient functions did occur but were imperceptible by the Katz Index and the MMSE measures. For instance, in nursing home patients, control of disruptive behavior
IAGS-OCTOBER 1990-VOL. 38, NO. 10
1080 FRETWELL ET AL
was often achieved, but was not reflected in measurable change in mental status. Table 5 shows that the patients receiving the intervention were more likely to change (both to improve and to decline) in emotional status then were controls; the bidirectional response makes the significance of the x 2 problematic. This concern is increased in light of the fact that the fraction of intervention subjects that showed improvement in mood was no different from that in the control group. The finding that study patients who were severely depressed at admission and received the team intervention were more likely to improve than the depressed people in the control group suggests that the geriatric assessment team did effectively identify and treat depression in acutely ill older patients. On the other hand, some of those who were not depressed, or only mildly so, in the treatment group experienced a decline in emotional status during the study. Given baseline comparability of the total group receiving the Zung and the subgroup of study patients who were very depressed at admission, and assuming that the effect of any measurement error on the change scores is equal, there may be differences in the effectiveness of our geriatric assessment process for different groups of patients. Without an intensive follow-up component, the team would have been unable to prevent decline of mood function occumng at the end of hospitalization or after discharge. Failure of the team to identify depression after the initial team conference or poorer compliance by the patient’s physician with team recommendations for treatment of the depression that develops during hospitalization (often termed “situational” or ”reactive” depression) are also possible explanations. Supporting the latter explanation was the finding that the compliance rate among depressed treatment patients was 83% compared to an overall treatment compliance rate of 65%. One final explanation for an apparent treatment effect in the area of emotional status was the effect of our decision to limit the application of the Zung instrument to only those study patients who had intact cognitive function. That process selected a group of individuals who were from home and had better physical and cognitive function than the overall study population. In other words, this criterion “targeted” (for the measurement of mood outcomes) the group for whom the intervention of geriatric assessment has been most consistently effective, when evaluated in randomized controlled trials.21,23,27 Future analyses of our study data will focus on the identification of the following groups: (1)those patients at risk for functional decline in the acute-care hospital, and (2) within this high-risk group, those who are most likely to benefit from the team consultation. The next chapter of geriatric assessment research should focus on the development of successful protocols for multidi-
mensional screening, assessment and treatment plans for frail patients with common profiles of functional impairments and medical problems, and the improvement of compliance with these treatment plans. Integration of the primary-care physician and nurse into the structure and process of interdisciplinary assessment and care teams may provide the most efficient and effective application of this comprehensive health-care intervention.26 REFERENCES 1. Allen CM, Becker PM, McVey LJ,et al: A randomized, controlled clinical trial of a geriatric consultation team: compliance with recommendations. JAMA 255:2617,1986 2. Rubenstein LZ,Josephson K, Wieland GD, et al: Geriatricassessment on a subacute hospital ward. Clin Geriatr Med 3:131,1987 3. Tulloch AJ, Moore V A randomized controlled trial of geriatric screening and surveillancein general practice. J R Coll Gen Pract 29:733, 1979 4. Williams ME: Outpatient geriatric evaluation. Clin Geriatr Med 3:175, 1987 5. Hendriksen C, Lund E, Stromgard E: Consequences of assessment and intervention among elderly people: three-year randomised controlled trial. Br Med J 289:1522,1984 6. Rubenstein LZ,Josephson KR, Wieland GD, et al: Effectiveness of a geriatric evaluation unit: a randomized clinical trial. N Engl J Med 311:1664,1984 7. Collard AF, Bachman SS, Beatrice D F Acute care delivery for the geriatric patient: an innovative approach. Quality Rev Bull 2(6):180, 1985 8 . Vetter NJ, Jones DA, Victor CR: Effect of health visitors working with elderly patients in general practice: a randomised controlled trial. Br Med J 288:369,1984 9. National Institutes of Health Consensus Development Conference Statement: Geriatric assessment methods for clinical decision-making (Consensus Development Panel, David Soloman, MD, Chairman). J Am Geriatr Soc 36:342,1988 10. Epstein AM, Hall JA, Besdine R, et al: The emergenceof geriatric assessment units: the ”new technologyof geriatrics.“Ann Intern Med 106(2):299, 1987 11. Klein LE, Roca RP, McArthur J, et al: Diagnosing dementia: u N variate and multivariate analyses of the mental status examination. J Am Geriatr Soc 33:483, 1985 12. Katz S,Akpom C A A measure of primary sociobiologicalfunctions. Int J Health Serv 6:493,1976 13. Glen M: A new design for randomized clinical trials. N Engl J Med 300:1242, 1979 14. Lawton P, Moss M, Fulcomer M, Lenban MH: Research and service oriented multilevel assessment instrument. J Gerontol 3791, 1982 15. Zung WWK: A self-rating depression scale. Arch Gen Psychiatry 1263, 1965 16. Okimoto JT,Barnes RF, Veith RC, et al: Screeningfor depression in geriatric medical patients. Am J Psychiatry 139:799, 1982 17. Kitchell MA, Barnes RF, Veith RC. et al: Screeningfor depression in hospitalized geriatric medical patients. J Am Geriatr Soc 30:174, 1982 18. Magni G, DeLeo D, Schifano F Depression in geriatric and adult medical inpatients. J Clin Psycho1 41:337,1985 19. Wilkinson L: SYSTAT The system for statistics. Evanston, IL, SYSTAT Inc, 1987 20. SPSS Inc: SPSS/PC V2.0 Base Manual. Chicago, SPSS, Inc., 1988. 21. Gayton D, Wood-Dauphinee S, de Lorimer M, et al: Trial of a geriatricconsultation team in an acute care hospital. J Am Geriatr Soc 35:726, 1987 22. Katz PR, Dube DH, Calkins E: Use of a structured functional assessment format in a geriatric consultative service. J Am Geriatr Soc 33:681, 1985 23. Becker PM, McVey LJ, Saltz CC, et al: Hospital-acquiredcompli-
+
IAGS-OCTOBER 1990-VOL 38, NO. 10
A TRIAL O F CONSULTATIVE GERIATRIC ASSESSMENT
cations in a randomized controlled clinical trial of a geriatric consultation team. JAMA2572313,1987 24. Wood-Dauphinee S, Shapiro S, Bass E,et al: A randomized trial of team care following stroke. Stroke 15:864,1984 25. Schmitt MH, Famll MP, Heinemann GD. Conceptual andmethodological problems in studying the effects of interdisciplinary geriatric teams. Gerontologist 28:753, 1988 26. Campion EW, Jette A, Berkman B An interdisciplinary geriatric
1081
consultative service: a controlled trial. J Am Geriatr Soc 31:792, 1983 27. Hogan DB, Fox
RA, Badley BWD, et al: Effect of a geriatric consultative service on management of patients in an acute care hospital. Can Med Assoc J 136:713, 1987 28. Fretwell MD: The consensus conference on comprehensive geriatric assessment: a dialogue is the beginning of consensus. J Am Geriatr Soc 36377,1988
