- Email: [email protected]
Screening for Elderly Patients Admitted to the Emergency Department Requiring Specialized Geriatric Care
The Journal of Emergency Medicine, Vol. 45, No. 5, pp. 739–745, 2013 Copyright Ó 2013 Elsevier Inc. Printed in the USA. All rights reserved 0736-4679/$ - see front matter
http://dx.doi.org/10.1016/j.jemermed.2012.11.110
Administration of Emergency Medicine
SCREENING FOR ELDERLY PATIENTS ADMITTED TO THE EMERGENCY DEPARTMENT REQUIRING SPECIALIZED GERIATRIC CARE Olivier Beauchet, MD, PHD,*†‡ Cyrille P. Launay, MD, MS,† Bruno Fantino, MD, PHD,†‡ Nicolas Lerolle, MD, PHD,§ Franck Maunoury, PHD,jj and Ce´dric Annweiler, MD, PHD, MS*†‡ *UPRES EA 4638, UNAM, Angers University, Angers, France, †Department of Neuroscience, Division of Geriatric Medicine, Angers Univiserty Hospital, Angers, France, ‡Angers University Memory Clinic, Angers, France, §University of Angers, UNAM, Department of Intensive Care, Angers University Hospital, Angers, France, and jjUNAM, Le Mans, France Reprint Address: Olivier Beauchet, MD, PHD, Department of Neuroscience, Division of Geriatric Medicine, Angers University Hospital, 49933 Angers Cedex 9, France; E-mail: [email protected]
, Abstract—Background: There is a need for a brief geriatric assessment (BGA) tool to screen elderly patients admitted to the Emergency Department (ED) for their risk of a long hospital stay. Objective: To examine whether a BGA administered to elderly patients admitted to the ED may predict the risk of a long hospital stay in the geriatric acute care unit. Methods: This study had a prospective cohort study design, enrolling 424 elderly patients (mean age 84.0 ± 6.5 years, 31.6% male) who were evaluated in the ED using a BGA composed of the following items: age, gender, number of medications taken daily, history of falls during the past 6 months, Mini-Mental State Examination (MMSE) score, and non-use of home-help services (i.e., living alone without using any formal or informal home services or social help). The length of stay (LOS) was calculated in days. Patients were separated into three groups based on LOS: low (<8 days), intermediate (8–13 days), and high (>13 days). Results: The prevalence of male gender was higher among patients with the longest LOS compared to those with intermediate LOS (p = 0.002). There were more patients with a history of falls in the high LOS group compared to the intermediate LOS group (p = 0.001) and the low LOS group (p < 0.001). The classification tree showed that male patients with an MMSE score <20 who fell with age under 85 years formed the end node with the greatest relative risk (RR) of a long hospital stay (RR = 14.3 with p < 0.001). Conclusions: The combination of a history of falls, male gender, cognitive impairment, and age under
85 years identified elderly ED patients at high risk of a long hospital stay. Ó 2013 Elsevier Inc. , Keywords—screening; elderly patients; medical emergency unit; long hospital stay
INTRODUCTION Adapted care plans for elderly patients admitted to Emergency Departments (ED) generally arise from the assessment process called the comprehensive geriatric assessment (CGA), which is a multidimensional, interdisciplinary diagnostic process to determine the medical, psychological, and functional capabilities of an elderly person, to develop a coordinated and integrated plan for treatment (1,2). This integration of CGA in the ED decision-making and care management improves inpatients’ health and functional status, and reduces mortality rate and health care expenditures (3–6). Implementation of a systematic CGA for every elderly inpatient admitted to the ED remains difficult in daily practice due to a number of issues. First, although the number of older patients keeps increasing, the number of health care professionals with geriatric skills does not (3,4). Second, CGA is a complex and timeconsuming process (2,5–8). Third, CGA requires
RECEIVED: 24 February 2011; FINAL SUBMISSION RECEIVED: 11 June 2011; ACCEPTED: 29 November 2012 739
740
O. Beauchet et al.
a multidisciplinary geriatric team that cannot support alone the care of all frail older inpatients due to the limited number of team members (5–8). The use of non-geriatric specialists in the CGA process is therefore implied (9,10). Recently, it was confirmed that CGA cannot be applied to every elderly inpatient, and that the best compromise could be the use of a two-step approach (5,6). The first step is the identification by nongeriatric specialists of elderly inpatients at high risk for adverse outcomes according to a screening tool, with the second step being the administration of a CGA by geriatric specialists with a diagnosis purpose. Health care professionals working in the ED need a simple, standardized, and brief geriatric assessment tool that will enable them to quickly identify the frail elderly patients requiring specialized geriatric care. The length of stay (LOS) could be considered as a surrogate measure of state of health and functional status in elderly inpatients because prolonged LOS has been identified as both a consequence and a cause of adverse health outcomes. For example, a long LOS has been related to a high morbidity burden, polypharmacy, and cognitive impairment (10–16). It has also been shown that the likelihood of a long LOS increases with the number of risk factors present, but previous models have not provided information about the specific combinations of identified risk factors (4,6,16). Thus, identification of elderly inpatients at risk for a long LOS may be combined with the identification of frail inpatients requiring specialized geriatric care. Based on this reasoning, we hypothesized that a screening tool for elderly patients administered in the ED could be built using specific combinations of the 6 following previously identified risk factors for long hospital stay: oldest-old patients (i.e., $ 85 years), male gender, polypharmacy (i.e., at least five medications taken each day), cognitive impairment, history of falls during the past 6 months, and non-use of home-help services (8–14). The aim of this study was to examine whether a brief geriatric assessment (BGA) grouping six binary items (i.e., yes or no) administered to elderly patients admitted to the ED may predict the relative risk of a long hospital stay in a geriatric acute care unit, to develop a specialized geriatric integrated plan for treatment. METHODS Participants and Settings The study sample was a convenience sample of all unplanned inpatients admitted to the geriatric acute care unit via the medical emergency unit of the ED of Angers University Hospital, France, between December 1, 2008
Inpatients admitted into the geriatric acute care (n=820) 334 excluded inpatients: − Age <75 (n=56) − Admitted via general practitioner (n=97) − Programmed hospitalization (n=181) Inpatients meeting the inclusion criteria and agreeing to participate (n=486) 62 inpatients withdrawn: - Died (n=26) - Data not available (n=36) Final inpatients used in the analyses (n=424)
Figure 1. Flow chart describing the selection of inpatients included in the study.
and October 30, 2009. As shown in Figure 1, all unplanned elderly inpatients (i.e., age $ 75 years) admitted to the geriatric acute care unit came from the medical emergency unit of the ED. Primary inclusion criteria were: evaluation by a nurse of the mobile geriatric team in the ED, unplanned admission to the geriatric acute care unit via the ED, age 75 years and over, willingness to participate and, finally, survival to discharge. Of the 820 admitted inpatients, 59.3% (n = 486) met the primary inclusion criteria and agreed to participate. Of this subset, all data were available for 450 (54.9%) participants, 26 of whom died during hospitalization. Finally, 424 participants (51.7%) were used in the analyses (Figure 1). The local ethics committee approved the project. Brief Geriatric Assessment The BGA was composed of the following six items: age coded as a binary variable (i.e., $ or <85 years), gender (i.e., male vs. female), number of medications taken daily (i.e., $ or <5), history of falls during the past 6 months (i.e., yes or no), Mini-Mental State Examination (MMSE) score (i.e., $ or <20), and non-use of homehelp services (i.e., yes or no). We chose these items because each of them has been separately associated with a prolonged hospitalization (4–8,10–16). During the ED assessment by the nurse, each patient’s age, gender, and number of medications taken daily were recorded. Older inpatients were defined as those aged 85 years and older. Polypharmacy was defined as a number of medications taken per day greater than five. In addition, all participants underwent a cognitive assessment consisting of the MMSE (17). A score <20 of 30 identified a moderate-to-severe cognitive impairment. History
Screening Tool for Long Hospital Stay
741
of falls in the past 6 months was also recorded. A fall was defined as unintentionally coming to rest on the ground, floor, or other lower level. The non-use of home-help services, defined as living alone without using any formal or informal home services or social help, was also assessed (18). Information on falls and formal or informal home services were obtained from the patient, or from a close person who lived with the patient. The LOS was calculated using the administrative registry of Angers University Hospital, and was defined as the number of days between the first day of admission into the ED and the last day of hospitalization in the geriatric acute care unit. Outcome Measures Outcome measures were: the six items of the BGA, i.e., age (mean 6 SD and percentage of participants aged 85 years and older); male gender; number of medications taken daily (mean 6 SD as well as percent of participants taking more than five medications per day), history of falls during the past 6 months (i.e., experienced at least one fall, expressed in percentage), MMSE score (mean 6 SD as well as percentage of participants with a score <20) and non-use of home-help services (expressed in percentage); and the LOS (expressed in number of days, and in tertiles). Statistical Analysis The participants’ baseline characteristics were summarized using means and SDs or frequencies and percent-
ages, as appropriate. Participants were separated into three groups based on the LOS: low (<8 days; n = 140), intermediate (8–13 days; n = 151), and high (> 13 days; n = 133). First, comparisons among groups were performed using one-way analysis of variance with Bonferroni corrections, or chi-squared, as appropriate. Second, a classification tree algorithm (CHAID algorithm) was performed. This technique splits a parent group into two subgroups (called ‘‘nodes’’) within which covariates are homogeneous, and between which outcome is distinct. The partitioning algorithm started with the covariate and split threshold that best maximized the difference in the outcome between the two subgroups. Fourth, for each end node, the relative risk of a long hospital stay was calculated using the end node with the lowest proportion, not including zero, of participants with a long LOS as a reference (Node 26 in our study). p-Values < 0.05 were considered statistically significant. All statistics were performed using R: A Language and Environment for Statistical Computing (version 2.10.0). RESULTS Patients were older in the group with intermediate LOS compared to those with low LOS (p = 0.012) (Table 1). The prevalence of male gender was higher among patients with high LOS compared to those with intermediate LOS (p = 0.002). There were more elderly patients with a history of falls in the group with high LOS compared to intermediate LOS (p = 0.001) and low LOS
Table 1. Comparison of Baseline Characteristics of Geriatrics Inpatients (n = 424) Separated into Three Groups Based on Length of Hospital Stay p-Value*
Length of Stay
Age (years) Mean 6 SD $85 years, n (%) Male gender, n (%) Number of medications taken daily Mean 6 SD >5, n (%) Falls during the past 6 months, n (%) Mini-Mental State Examination score (/30) Mean 6 SD <20, n (%) Non-use home-help servicesjj Length of hospital stay (days), Mean 6 SD
Short† n = 140
Intermediate‡ n = 151
Long§ n = 133
Overall
Short vs. Intermediate
Short vs. Long
83.4 6 6.4 59 (42.1) 42 (30.0
84.5 6 6.7 86 (57.0) 37 (24.5)
84.0 6 6.4 65 ( 48.9) 55 (41.4)
0.332 0.041{ 0.008{
0.012{ 0.292
0.173 0.050
0.264 0.002{
6.0 6 2.2 74 (52.9) 36 (25.7)
6.4 6 3.2 42 (27.8) 42 (27.8)
6.3 6 3.1 86 (64.7) 64 (46.6)
0.594 0.072 <0.001{
0.686
<0.001{
0.001{
18.8 6 6.3 73 (52.1) 83 (59.3) 4.4 6 2.6
18.2 6 7.0 78 (58.7) 103 (68.2) 10.3 6 1.8
17.1 6 6.4 84 (63.2) 91 (68.4) 21.5 6 9.3
0.100 0.095 0.185 <0.001{
<0.001{
<0.001{
<0.001{
* Comparison based on one-way analysis of variance with Bonferroni corrections, or chi-squared test, as appropriate. † Length of stay <8 days. ‡ Length of stay between 8 and 13 days. § Length of stay >13 days. jj Living alone without using any formal or informal home services and social help. { p significant (i.e., <0.05).
Intermediate vs. Long
742
O. Beauchet et al.
Node 0 n = 424 31.4% (133/424) Falls during the past 6 months (p < 0.001) Yes
No
Node 1 n = 140 44.3% (62/140)
Node 16 n = 284 25.0% (71/284) MMSE score (/30) < 20 (p < 0.001)
Male gender (p = 0.025) No
Yes Node 2 n = 44 50.0% (22/44)
Node 3 n = 96 41.7% (40/96)
MMSE score (/30) < 20 (p = 0.046) Yes No
Non-use home-help services (p = 0.127) No Yes
Node 5 n = 16 37.5% (6/16)
Node 6 n = 62 43.6% (27/62)
Number of drugs > 5 (p = 0.097) No
MMSE score (/30) < 20 (p = 0.098) Yes No
Node 4 n = 28 57.1% (16/28) Age > 85 years ( p = 0.033) Yes
No
Node 8 n = 18 44.4% (8/18)
Node 9 n = 10 80.0% (8/10)
RR = 7.9 [1.1;56.6] p = 0.021
RR = 14.3 [2.1;97.6] p < 0.001
Yes
No
Yes Node 17 n = 152 29.6% (45/152)
Non-use home-help services (p = 0.014)
Number of drugs > 5 Yes
No
Node 7 n = 34 38.2% (13/34)
Node 19 n = 97 35.1% (34/97)
MMSE score (/30) < 20 (p = 0.075) No
Non-use home-help services ( p = 0.015) No Yes
Yes
Node 18 n = 132 19.7% (26/132)
No
Yes
Non-use home-help services ( p = 0.023) No Yes
Node 22 n = 56 14.3% (8/56)
Node 21 n = 76 23.7% (18/76)
Node 20 n = 55 20.0% (11/55)
Number of drugs > 5 ( p = 0.022) No
Yes
Node 10 n=7 57.1% (4/7)
Node 11 n=9 22.2% (2/9)
Node 12 n = 33 39.4% (13/33)
Node 13 n = 29 48.3% (14/29)
Node 14 n = 22 45.4% (10/22)
Node 15 n = 12 25.0% (3/12)
Node 23 n = 74 31.1% (23/74)
Node 24 n = 23 47.8% (11/23)
Node 25 n = 37 27.0% (10/37)
Node 26 n = 18 5.6% (1/18)
Node 27 n = 49 18.4% (9/49)
Node 28 n = 27 33.3% (9/27)
RR = 10.2 [1.4.;75.5] p = 0.019
RR = 4.0 [0.4;37.9] p = 0.516
RR = 7.0 [4.1;49.1] p = 0.024
RR = 8.6 [1.2;59.6] p = 0.006
RR = 8.1 [1.2;57.1] p = 0.014
RR = 4.5 [0.5;37.8] p = 0.329
RR = 5.6 [0.8;38.1] p = 0.056
RR = 8.5 [1.2;59.7] p = 0.009
RR = 4.8 [0.7;34.6] p = 0.131
RR = 1.0
RR = 3.3 [0.5;23.9] p = 0.359
RR = 5.9 [0.8;42.7] p = 0.067
Yes
Male gender ( p = 0.046)
Node 29 n = 14 0.0% (0/14)
No
Node 30 n = 42 19.1% (8/42) RR = 3.4 [0.5;25.1] p = 0.344
Figure 2. Classification tree for the prediction of risk of a long hospital stay (i.e., > 13 days) among studied geriatric inpatients (n = 424). CHAID Algorithm. RR = relative risk; [;] = 95% confidence interval. Circles indicate the intermediate nodes, and squares indicate the end nodes. Circles and squares contain the group size and the proportion of participants with a long hospital stay. At each node, split variable and the p-value of comparison of proportion of participants with a long hospital stay between the two resulting branches are shown. p Significance (i.e., <0.05) indicated in bold. The square gray node was used as reference value for calculating the relative risks. MMSE = Mini-Mental State Examination.
(p < 0.001). The mean LOS increased significantly across the groups (p < 0.001 for all comparisons). There was no significant difference in the other characteristics. Figure 2 shows the classification tree results for predicting a long hospital stay (i.e., >13 days). It consists of 30 end groups, each having a different risk of a long hospital stay. The root node (n 0) comprising the entire sample (n = 424) was first separated into two nodes (n 1) and (n 2) by the first major split, which was ‘‘History of falls during the past 6 months.’’ The analysis identified 15 end nodes for participants who fell at least one time during the past 6 months, and also 15 end nodes for those who did not fall. Among patients who fell, those who were male, had an MMSE score <20, and were under 85 years of age formed the end node (n 9) with the greatest relative risk (RR) of a long hospital stay (RR = 14.3 with p < 0.001). Conversely, among those who did not fall, those who had a MMSE score $20, used no homehelp services, and who were male (node n 29) formed the most favorable group because 100% of the participants had a short hospital stay, even though this combination was not significant (RR = 0 with p = 1.000). In addition, the RR of a long hospital stay was significant for node n 8 (RR = 7.9 with p = 0.021), node n 10 (RR = 10.2 with p = 0.019), node n 12 (RR = 7.0 with
p = 0.024), node n 13 (RR = 8.6 with p = 0.006), node n 14 (RR = 8.1 with p = 0.014), and node n 24 (RR = 8.5 with p = 0.009). DISCUSSION Our results show that the six-item BGA was significantly associated with the LOS. The classification tree algorithm highlighted that the risk of long hospital stay changed with the different combinations of the six BGA items, a history of falls being the major split item for a long stay. Furthermore, the combination of a history of falls, male gender, cognitive impairment, and age under 85 years identified elderly ED patients with the highest risk of a long hospital stay. In our study, a history of falls was a key predictor of a long hospital stay. This relationship may be explained in part by the fact that falls in older adults are a marker of morbidities whose consequences are gait and balance disorders, but also a loss of independence (18,19). Indeed, most identified risk factors for falls are intrinsic (i.e., subject-related) and usually include polypharmacy, psychoactive medications, muscle weakness, sensory deficits, and a cognitive decline (18–20). The state of health of patients with a history of recent falls is
Screening Tool for Long Hospital Stay
therefore characterized by a cumulative effect of chronic diseases (21). These effects become even more pronounced and diverse with age, contributing to a vicious cycle of increasing frailty and increasing risk of dependence, which may increase the LOS (21,22). Male gender was also associated with a long hospital stay in our study. The prevalence of male gender was significantly higher in the group of inpatients with long LOS compared to those with intermediate LOS (p = 0.002), and there was a significant trend to have more men in the long LOS group compared to the short LOS group (p = 0.050). A higher age-related morbidity burden in men compared to women could be the main explanation for this finding. Indeed, it recently has been reported that among elderly patients, male gender is associated with a greater number of morbidities on admission than female gender, and greater impairments in biological markers of clinical severity such as serum albumin, as well as in functional status (23). In addition, male gender has been identified as a strong predictor for adverse events after an organ failure (23–26). Moderate-to-severe cognitive impairment was associated with a long hospital stay in our study. Acute medical illnesses and a change of environment often lead to an increased risk of confusion, agitation, and behavioral troubles in persons with dementia (27). Adverse outcomes related to cognitive decline could explain the long LOS found in our study. We also suggest that an increase in disability in demented inpatients may help to explain the long hospital stay. Indeed, older patients often experience a decline in function associated with hospitalization for acute medical illnesses, precipitating disability in activities of daily living (27–29). It has been shown that dementia independently predicts failure to recover in activities of daily living that may explain in part the prolonged hospital stay among inpatients with moderate-to-severe cognitive decline (29). The general idea is that the LOS is associated with the degree of frailty, and that frail elderly inpatients have a longer LOS compared to transitionally frail inpatients, who themselves have a longer LOS than vigorous inpatients. However, our results showed that this association is not linear. Indeed, between-group comparisons highlighted that the prevalence of oldest-old inpatients (i.e., $ 85 years) was higher within the group of intermediate LOS compared to short LOS (p = 0.012) but not compared to long LOS (p = 0.264). This result demonstrates that aging should not be considered a disease, and confirmed that oldest-old inpatients have a heterogeneous frailty status. Our results also show that the risk of a long LOS may change according to the combinations of the BGA items. The history of falls was the primary segmentation process, and participants with a history of falls in the
743
past 6 months had a broadly higher risk of a long hospital stay than those with no history of falls. Actually, the risk depended on the combination of having had a fall with the other risk factors, as shown by the classification tree algorithm. For example, men with a history of recent falls but no moderate-to-severe cognitive impairment and no polypharmacy had a low though non-significant RR for a long hospital stay (RR = 4.0 with p = 0.516), although those who did not fall but presented with cognitive impairment, polypharmacy, or no social isolation had a surprisingly higher and significant RR of a long hospital stay (RR = 8.5 with p = 0.009). Our approach of utilizing the BGA to identify older inpatients at high risk of a long hospital stay provides several advantages over previous CGAs and regressions models. First, the BGA is based on only six items, is easy to measure, accessible, takes little time to administer, and requires no specific competence or equipment. Second, the classification tree algorithm is simple and requires no score to be calculated, as the risk profile gives rise to an end group with a matching risk for a long hospital stay. Furthermore, it does not require any a priori distributional assumption and knowledge about the underlying relationships between dependent (i.e., long hospital stay) and independent (i.e., BGA items) variables. Hence, this method is useful in situations where there are interactions among variables; the cases are partitioned and each group is analyzed separately. In addition, classification tree algorithm allows the construction of directly applicable risk profiles for long hospital stay. Compared to regression analyses where all predictors must be measured to identify the risk profile, few predictors – in our case the BGA six items – are required to recognize the risk profile. Lastly, although risk profiles do not provide information about causal relationships, knowledge about the combination of the 6 BGA items of end groups might provide a more purposeful assignment of recommended standards of care among older patients admitted to the ED. Limitations Some limitations of this study need to be considered. First, selected older inpatients probably were not representative of the general population of older adults admitted to the ED. Only half of inpatients met inclusion criteria, and our study was limited to one single ED and one single geriatric acute care unit. Also, given the many differences in health care systems across countries, the generalization of our findings could be limited to the French health care system. Second, as the number of participants in end nodes was low, equivocal or nonsignificant results could be the result of a lack of power. Third, the six BGA items did not take into account the
744
O. Beauchet et al.
causes (i.e., organ failure and geriatric syndromes) of hospitalization of participants. Fourth, one of the six BGA items was the reported history of falls. This information was collected by questioning each subject or caregiver, which may have led to a recall bias (30). Fifth, although the MMSE is the best-studied and most widely used instrument to screen cognitive impairment, its main limitation, especially in the ED context, is that it is a timeconsuming test (e.g., requires about 15 min to perform). CONCLUSION In conclusion, prediction of LOS with a six-item BGA was possible in the studied sample of older inpatients admitted to the ED. The risk of a long hospital stay changed depending on the different combinations of the six items on the BGA. The combination of a history of a recent fall, male gender, cognitive impairment, and age under 85 years identified the elderly ED patients with the highest risk of a long hospital stay requiring geriatric care and planning for discharge. Acknowledgment—The author would like to thank all participants involved in the present study.
REFERENCES 1. Li CM, Chen CY, Li CY, Wang WD, Wu SC. The effectiveness of a comprehensive geriatric assessment intervention program for frailty in community-dwelling older people: a randomized, controlled trial. Arch Gerontol Geriatr 2010;50:S39–42. 2. Van Craen K, Braes T, Wellens N, et al. The effectiveness of inpatient geriatric evaluation and management units: a systematic review and meta-analysis. J Am Geriatr Soc 2010;58:83–92. 3. McCusker J, Verdon J, Tousignant P, de Courval LP, Dendukuri N, Belzile E. Rapid emergency department intervention for older people reduces risk of functional decline: results of a multicenter randomized trial. J Am Geriatr Soc 2001;49:1272–81. 4. Lang PO, Zekry D, Michel JP, et al. Early markers of prolonged hospital stay in demented inpatients: a multicentre and prospective study. J Nutr Health Aging 2010;14:141–7. 5. Graf CE, Zekry D, Giannelli S, Michel JP, Chevalley T. Efficiency and applicability of the comprehensive geriatric assessment in the emergency department: a systematic review. Aging Clin Exp Res 2011;23:244–54. 6. Graf CE, Zekry D, Giannelli S, Michel JP, Chevalley T. Comprehensive geriatric assessment in the emergency department. J Am Geriatr Soc 2010;58:2032–3. 7. Hoogerduijn JG, Schuurmans MJ, Korevaar JC, Buurman BM, de Rooij SE. Identification of older hospitalised patients at risk for functional decline, a study to compare the predictive values of three screening instruments. J Clin Nurs 2010;19:1219–25. 8. Hoogerduijn JG, Schuurmans MJ, Duijnstee MS, de Rooij SE, Grypdonck MF. A systematic review of predictors and screening instruments to identify older hospitalized patients at risk for functional decline. J Clin Nurs 2007;16:46–57.
9. Reuben DB. Medical care for the final years of life: ‘‘When you’re 83, it’s not going to be 20 years’’ JAMA 2009;302:2686–94. 10. Reuben DB. Better care for older people with chronic diseases: an emerging vision. JAMA 2007;298:2673–4. 11. Covinsky KE, Palmer RM, Fortinsky RH, et al. Loss of independence in activities of daily living in older adults hospitalized with medical illness: increased vulnerability with age. J Am Geriatr Soc 2003;51:451–8. 12. Zanocchi M, Maero B, Maina P, et al. Factor predicting a prolonged hospital stay in elderly patients. Minerva Med 2002;93:135–43. 13. Hirsch CH, Sommers L, Olsen A. The natural history of functional morbidity in hospitalized older patients. J Am Geriatr Soc 1996;38: 1296–303. 14. Aminzadeh F, Dalziel WB. Older adults in the emergency department: a systematic review of patterns of use, adverse outcomes, and effectiveness of interventions. Ann Emerg Med 2002;39:238–47. 15. McCusker J, Cardin S, Bellavance F, et al. Return to the emergency department among elders: patterns and predictors. Acad Emerg Med 2000;7:249–59. 16. Hastings SN, Schmader KE, Sloane RJ, et al. Adverse health outcomes after discharge from the emergency department—incidence and risk factors in a veteran population. J Gen Intern Med 2007;22: 1527–31. 17. Folstein MF, Folstein SE, McHugh PR. ‘‘Mini-mental state’’. A practical method for grading the cognitive state of the patient for the clinician. J Psychiatr Res 1975;12:189–98. 18. American Geriatrics Society, British Geriatrics Society, and American Academy of Orthopedic Surgeons Panel on Falls Prevention. Guideline for the prevention of falls in older persons. J Am Geriatr Soc 2001;49:664–772. 19. Oliver D, Daly F, Martin FC, McMurdo ME. Risk factors and risk assessment tools for falls in hospital in-patients: a systematic review. Age Ageing 2004;33:122–30. 20. Bloem BR, Steijns JA, Smits-Engelsman BC. An update on falls. Curr Opin Neurol 2003;16:15–26. 21. Fried LP, Bandeen-Roche K, Kasper JD, Guralnik JM. Association of comorbidity with disability in older women: the Women’s Health and Aging Study. J Clin Epidemiol 1999;52:27–37. 22. Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56: M146–56. 23. Brown CJ, Redden DT, Flood KL, Allman RM. The underrecognized epidemic of low mobility during hospitalization of older adults. J Am Geriatr Soc 2009;57:1660–5. 24. Rozzini R, Sleiman I, Maggi S, Noale M, Trabucchi M. Gender differences and health status in old and very old patients. J Am Med Dir Assoc 2009;10:554–8. 25. Pitt B. Male gender, diabetes, COPD, anemia, and creatinine clearance < 30 mL/min predicted hospitalization after heart failure diagnosis. Ann Intern Med 2010;152:JC4–2, JC4-3. 26. Lee TC, Wang HP, Chiu HM, et al. Male gender and renal dysfunction are predictors of adverse outcome in nonpostoperative ischemic colitis patients. J Clin Gastroenterol 2010;44:e96–100. 27. Zekry D, Herrmann FR, Grandjean R, et al. Does dementia predict adverse hospitalization outcomes? A prospective study in aged inpatients. Int J Geriatr Psychiatry 2009;24:283–91. 28. Ehlenbach WJ, Hough CL, Crane PK, et al. Association between acute care and critical illness hospitalization and cognitive function in older adults. JAMA 2010;303:763–70. 29. Boyd CM, Landefeld CS, Counsell SR, et al. Recovery of activities of daily living in older adults after hospitalization for acute medical illness. J Am Geriatr Soc 2008;56:2171–9. 30. Cummings SR, Nevitt MC, Kidd S. forgetting falls: the limited accuracy of recall of falls in the elderly. J Am Geriatr Soc 1988; 36:613–6.
Screening Tool for Long Hospital Stay
ARTICLE SUMMARY 1. Why is this topic important? Because of the increased number of frail older adults hospitalized in the Emergency Department, current hospital care systems are driven by the high burden of chronic diseases and their adverse consequences, worsening the patient’s quality of life and healthcare professional practice. 2. What does this study attempt to show? This study attempts to show that a brief geriatric assessment carried out in older patients admitted in the ED may predict the risk of a long hospital stay in a geriatric acute care unit. 3. What are the key findings? The combination of a history of falls, male gender, cognitive impairment and age under 85 years, identified elderly patients admitted and at highest risk of long hospital stay. 4. How is patient care impacted? Use of a brief geriatric assessment (BGA) to identify frail older patients early could improve their discharge planning.
745
http://dx.doi.org/10.1016/j.jemermed.2012.11.110
Administration of Emergency Medicine
SCREENING FOR ELDERLY PATIENTS ADMITTED TO THE EMERGENCY DEPARTMENT REQUIRING SPECIALIZED GERIATRIC CARE Olivier Beauchet, MD, PHD,*†‡ Cyrille P. Launay, MD, MS,† Bruno Fantino, MD, PHD,†‡ Nicolas Lerolle, MD, PHD,§ Franck Maunoury, PHD,jj and Ce´dric Annweiler, MD, PHD, MS*†‡ *UPRES EA 4638, UNAM, Angers University, Angers, France, †Department of Neuroscience, Division of Geriatric Medicine, Angers Univiserty Hospital, Angers, France, ‡Angers University Memory Clinic, Angers, France, §University of Angers, UNAM, Department of Intensive Care, Angers University Hospital, Angers, France, and jjUNAM, Le Mans, France Reprint Address: Olivier Beauchet, MD, PHD, Department of Neuroscience, Division of Geriatric Medicine, Angers University Hospital, 49933 Angers Cedex 9, France; E-mail: [email protected]
, Abstract—Background: There is a need for a brief geriatric assessment (BGA) tool to screen elderly patients admitted to the Emergency Department (ED) for their risk of a long hospital stay. Objective: To examine whether a BGA administered to elderly patients admitted to the ED may predict the risk of a long hospital stay in the geriatric acute care unit. Methods: This study had a prospective cohort study design, enrolling 424 elderly patients (mean age 84.0 ± 6.5 years, 31.6% male) who were evaluated in the ED using a BGA composed of the following items: age, gender, number of medications taken daily, history of falls during the past 6 months, Mini-Mental State Examination (MMSE) score, and non-use of home-help services (i.e., living alone without using any formal or informal home services or social help). The length of stay (LOS) was calculated in days. Patients were separated into three groups based on LOS: low (<8 days), intermediate (8–13 days), and high (>13 days). Results: The prevalence of male gender was higher among patients with the longest LOS compared to those with intermediate LOS (p = 0.002). There were more patients with a history of falls in the high LOS group compared to the intermediate LOS group (p = 0.001) and the low LOS group (p < 0.001). The classification tree showed that male patients with an MMSE score <20 who fell with age under 85 years formed the end node with the greatest relative risk (RR) of a long hospital stay (RR = 14.3 with p < 0.001). Conclusions: The combination of a history of falls, male gender, cognitive impairment, and age under
85 years identified elderly ED patients at high risk of a long hospital stay. Ó 2013 Elsevier Inc. , Keywords—screening; elderly patients; medical emergency unit; long hospital stay
INTRODUCTION Adapted care plans for elderly patients admitted to Emergency Departments (ED) generally arise from the assessment process called the comprehensive geriatric assessment (CGA), which is a multidimensional, interdisciplinary diagnostic process to determine the medical, psychological, and functional capabilities of an elderly person, to develop a coordinated and integrated plan for treatment (1,2). This integration of CGA in the ED decision-making and care management improves inpatients’ health and functional status, and reduces mortality rate and health care expenditures (3–6). Implementation of a systematic CGA for every elderly inpatient admitted to the ED remains difficult in daily practice due to a number of issues. First, although the number of older patients keeps increasing, the number of health care professionals with geriatric skills does not (3,4). Second, CGA is a complex and timeconsuming process (2,5–8). Third, CGA requires
RECEIVED: 24 February 2011; FINAL SUBMISSION RECEIVED: 11 June 2011; ACCEPTED: 29 November 2012 739
740
O. Beauchet et al.
a multidisciplinary geriatric team that cannot support alone the care of all frail older inpatients due to the limited number of team members (5–8). The use of non-geriatric specialists in the CGA process is therefore implied (9,10). Recently, it was confirmed that CGA cannot be applied to every elderly inpatient, and that the best compromise could be the use of a two-step approach (5,6). The first step is the identification by nongeriatric specialists of elderly inpatients at high risk for adverse outcomes according to a screening tool, with the second step being the administration of a CGA by geriatric specialists with a diagnosis purpose. Health care professionals working in the ED need a simple, standardized, and brief geriatric assessment tool that will enable them to quickly identify the frail elderly patients requiring specialized geriatric care. The length of stay (LOS) could be considered as a surrogate measure of state of health and functional status in elderly inpatients because prolonged LOS has been identified as both a consequence and a cause of adverse health outcomes. For example, a long LOS has been related to a high morbidity burden, polypharmacy, and cognitive impairment (10–16). It has also been shown that the likelihood of a long LOS increases with the number of risk factors present, but previous models have not provided information about the specific combinations of identified risk factors (4,6,16). Thus, identification of elderly inpatients at risk for a long LOS may be combined with the identification of frail inpatients requiring specialized geriatric care. Based on this reasoning, we hypothesized that a screening tool for elderly patients administered in the ED could be built using specific combinations of the 6 following previously identified risk factors for long hospital stay: oldest-old patients (i.e., $ 85 years), male gender, polypharmacy (i.e., at least five medications taken each day), cognitive impairment, history of falls during the past 6 months, and non-use of home-help services (8–14). The aim of this study was to examine whether a brief geriatric assessment (BGA) grouping six binary items (i.e., yes or no) administered to elderly patients admitted to the ED may predict the relative risk of a long hospital stay in a geriatric acute care unit, to develop a specialized geriatric integrated plan for treatment. METHODS Participants and Settings The study sample was a convenience sample of all unplanned inpatients admitted to the geriatric acute care unit via the medical emergency unit of the ED of Angers University Hospital, France, between December 1, 2008
Inpatients admitted into the geriatric acute care (n=820) 334 excluded inpatients: − Age <75 (n=56) − Admitted via general practitioner (n=97) − Programmed hospitalization (n=181) Inpatients meeting the inclusion criteria and agreeing to participate (n=486) 62 inpatients withdrawn: - Died (n=26) - Data not available (n=36) Final inpatients used in the analyses (n=424)
Figure 1. Flow chart describing the selection of inpatients included in the study.
and October 30, 2009. As shown in Figure 1, all unplanned elderly inpatients (i.e., age $ 75 years) admitted to the geriatric acute care unit came from the medical emergency unit of the ED. Primary inclusion criteria were: evaluation by a nurse of the mobile geriatric team in the ED, unplanned admission to the geriatric acute care unit via the ED, age 75 years and over, willingness to participate and, finally, survival to discharge. Of the 820 admitted inpatients, 59.3% (n = 486) met the primary inclusion criteria and agreed to participate. Of this subset, all data were available for 450 (54.9%) participants, 26 of whom died during hospitalization. Finally, 424 participants (51.7%) were used in the analyses (Figure 1). The local ethics committee approved the project. Brief Geriatric Assessment The BGA was composed of the following six items: age coded as a binary variable (i.e., $ or <85 years), gender (i.e., male vs. female), number of medications taken daily (i.e., $ or <5), history of falls during the past 6 months (i.e., yes or no), Mini-Mental State Examination (MMSE) score (i.e., $ or <20), and non-use of homehelp services (i.e., yes or no). We chose these items because each of them has been separately associated with a prolonged hospitalization (4–8,10–16). During the ED assessment by the nurse, each patient’s age, gender, and number of medications taken daily were recorded. Older inpatients were defined as those aged 85 years and older. Polypharmacy was defined as a number of medications taken per day greater than five. In addition, all participants underwent a cognitive assessment consisting of the MMSE (17). A score <20 of 30 identified a moderate-to-severe cognitive impairment. History
Screening Tool for Long Hospital Stay
741
of falls in the past 6 months was also recorded. A fall was defined as unintentionally coming to rest on the ground, floor, or other lower level. The non-use of home-help services, defined as living alone without using any formal or informal home services or social help, was also assessed (18). Information on falls and formal or informal home services were obtained from the patient, or from a close person who lived with the patient. The LOS was calculated using the administrative registry of Angers University Hospital, and was defined as the number of days between the first day of admission into the ED and the last day of hospitalization in the geriatric acute care unit. Outcome Measures Outcome measures were: the six items of the BGA, i.e., age (mean 6 SD and percentage of participants aged 85 years and older); male gender; number of medications taken daily (mean 6 SD as well as percent of participants taking more than five medications per day), history of falls during the past 6 months (i.e., experienced at least one fall, expressed in percentage), MMSE score (mean 6 SD as well as percentage of participants with a score <20) and non-use of home-help services (expressed in percentage); and the LOS (expressed in number of days, and in tertiles). Statistical Analysis The participants’ baseline characteristics were summarized using means and SDs or frequencies and percent-
ages, as appropriate. Participants were separated into three groups based on the LOS: low (<8 days; n = 140), intermediate (8–13 days; n = 151), and high (> 13 days; n = 133). First, comparisons among groups were performed using one-way analysis of variance with Bonferroni corrections, or chi-squared, as appropriate. Second, a classification tree algorithm (CHAID algorithm) was performed. This technique splits a parent group into two subgroups (called ‘‘nodes’’) within which covariates are homogeneous, and between which outcome is distinct. The partitioning algorithm started with the covariate and split threshold that best maximized the difference in the outcome between the two subgroups. Fourth, for each end node, the relative risk of a long hospital stay was calculated using the end node with the lowest proportion, not including zero, of participants with a long LOS as a reference (Node 26 in our study). p-Values < 0.05 were considered statistically significant. All statistics were performed using R: A Language and Environment for Statistical Computing (version 2.10.0). RESULTS Patients were older in the group with intermediate LOS compared to those with low LOS (p = 0.012) (Table 1). The prevalence of male gender was higher among patients with high LOS compared to those with intermediate LOS (p = 0.002). There were more elderly patients with a history of falls in the group with high LOS compared to intermediate LOS (p = 0.001) and low LOS
Table 1. Comparison of Baseline Characteristics of Geriatrics Inpatients (n = 424) Separated into Three Groups Based on Length of Hospital Stay p-Value*
Length of Stay
Age (years) Mean 6 SD $85 years, n (%) Male gender, n (%) Number of medications taken daily Mean 6 SD >5, n (%) Falls during the past 6 months, n (%) Mini-Mental State Examination score (/30) Mean 6 SD <20, n (%) Non-use home-help servicesjj Length of hospital stay (days), Mean 6 SD
Short† n = 140
Intermediate‡ n = 151
Long§ n = 133
Overall
Short vs. Intermediate
Short vs. Long
83.4 6 6.4 59 (42.1) 42 (30.0
84.5 6 6.7 86 (57.0) 37 (24.5)
84.0 6 6.4 65 ( 48.9) 55 (41.4)
0.332 0.041{ 0.008{
0.012{ 0.292
0.173 0.050
0.264 0.002{
6.0 6 2.2 74 (52.9) 36 (25.7)
6.4 6 3.2 42 (27.8) 42 (27.8)
6.3 6 3.1 86 (64.7) 64 (46.6)
0.594 0.072 <0.001{
0.686
<0.001{
0.001{
18.8 6 6.3 73 (52.1) 83 (59.3) 4.4 6 2.6
18.2 6 7.0 78 (58.7) 103 (68.2) 10.3 6 1.8
17.1 6 6.4 84 (63.2) 91 (68.4) 21.5 6 9.3
0.100 0.095 0.185 <0.001{
<0.001{
<0.001{
<0.001{
* Comparison based on one-way analysis of variance with Bonferroni corrections, or chi-squared test, as appropriate. † Length of stay <8 days. ‡ Length of stay between 8 and 13 days. § Length of stay >13 days. jj Living alone without using any formal or informal home services and social help. { p significant (i.e., <0.05).
Intermediate vs. Long
742
O. Beauchet et al.
Node 0 n = 424 31.4% (133/424) Falls during the past 6 months (p < 0.001) Yes
No
Node 1 n = 140 44.3% (62/140)
Node 16 n = 284 25.0% (71/284) MMSE score (/30) < 20 (p < 0.001)
Male gender (p = 0.025) No
Yes Node 2 n = 44 50.0% (22/44)
Node 3 n = 96 41.7% (40/96)
MMSE score (/30) < 20 (p = 0.046) Yes No
Non-use home-help services (p = 0.127) No Yes
Node 5 n = 16 37.5% (6/16)
Node 6 n = 62 43.6% (27/62)
Number of drugs > 5 (p = 0.097) No
MMSE score (/30) < 20 (p = 0.098) Yes No
Node 4 n = 28 57.1% (16/28) Age > 85 years ( p = 0.033) Yes
No
Node 8 n = 18 44.4% (8/18)
Node 9 n = 10 80.0% (8/10)
RR = 7.9 [1.1;56.6] p = 0.021
RR = 14.3 [2.1;97.6] p < 0.001
Yes
No
Yes Node 17 n = 152 29.6% (45/152)
Non-use home-help services (p = 0.014)
Number of drugs > 5 Yes
No
Node 7 n = 34 38.2% (13/34)
Node 19 n = 97 35.1% (34/97)
MMSE score (/30) < 20 (p = 0.075) No
Non-use home-help services ( p = 0.015) No Yes
Yes
Node 18 n = 132 19.7% (26/132)
No
Yes
Non-use home-help services ( p = 0.023) No Yes
Node 22 n = 56 14.3% (8/56)
Node 21 n = 76 23.7% (18/76)
Node 20 n = 55 20.0% (11/55)
Number of drugs > 5 ( p = 0.022) No
Yes
Node 10 n=7 57.1% (4/7)
Node 11 n=9 22.2% (2/9)
Node 12 n = 33 39.4% (13/33)
Node 13 n = 29 48.3% (14/29)
Node 14 n = 22 45.4% (10/22)
Node 15 n = 12 25.0% (3/12)
Node 23 n = 74 31.1% (23/74)
Node 24 n = 23 47.8% (11/23)
Node 25 n = 37 27.0% (10/37)
Node 26 n = 18 5.6% (1/18)
Node 27 n = 49 18.4% (9/49)
Node 28 n = 27 33.3% (9/27)
RR = 10.2 [1.4.;75.5] p = 0.019
RR = 4.0 [0.4;37.9] p = 0.516
RR = 7.0 [4.1;49.1] p = 0.024
RR = 8.6 [1.2;59.6] p = 0.006
RR = 8.1 [1.2;57.1] p = 0.014
RR = 4.5 [0.5;37.8] p = 0.329
RR = 5.6 [0.8;38.1] p = 0.056
RR = 8.5 [1.2;59.7] p = 0.009
RR = 4.8 [0.7;34.6] p = 0.131
RR = 1.0
RR = 3.3 [0.5;23.9] p = 0.359
RR = 5.9 [0.8;42.7] p = 0.067
Yes
Male gender ( p = 0.046)
Node 29 n = 14 0.0% (0/14)
No
Node 30 n = 42 19.1% (8/42) RR = 3.4 [0.5;25.1] p = 0.344
Figure 2. Classification tree for the prediction of risk of a long hospital stay (i.e., > 13 days) among studied geriatric inpatients (n = 424). CHAID Algorithm. RR = relative risk; [;] = 95% confidence interval. Circles indicate the intermediate nodes, and squares indicate the end nodes. Circles and squares contain the group size and the proportion of participants with a long hospital stay. At each node, split variable and the p-value of comparison of proportion of participants with a long hospital stay between the two resulting branches are shown. p Significance (i.e., <0.05) indicated in bold. The square gray node was used as reference value for calculating the relative risks. MMSE = Mini-Mental State Examination.
(p < 0.001). The mean LOS increased significantly across the groups (p < 0.001 for all comparisons). There was no significant difference in the other characteristics. Figure 2 shows the classification tree results for predicting a long hospital stay (i.e., >13 days). It consists of 30 end groups, each having a different risk of a long hospital stay. The root node (n 0) comprising the entire sample (n = 424) was first separated into two nodes (n 1) and (n 2) by the first major split, which was ‘‘History of falls during the past 6 months.’’ The analysis identified 15 end nodes for participants who fell at least one time during the past 6 months, and also 15 end nodes for those who did not fall. Among patients who fell, those who were male, had an MMSE score <20, and were under 85 years of age formed the end node (n 9) with the greatest relative risk (RR) of a long hospital stay (RR = 14.3 with p < 0.001). Conversely, among those who did not fall, those who had a MMSE score $20, used no homehelp services, and who were male (node n 29) formed the most favorable group because 100% of the participants had a short hospital stay, even though this combination was not significant (RR = 0 with p = 1.000). In addition, the RR of a long hospital stay was significant for node n 8 (RR = 7.9 with p = 0.021), node n 10 (RR = 10.2 with p = 0.019), node n 12 (RR = 7.0 with
p = 0.024), node n 13 (RR = 8.6 with p = 0.006), node n 14 (RR = 8.1 with p = 0.014), and node n 24 (RR = 8.5 with p = 0.009). DISCUSSION Our results show that the six-item BGA was significantly associated with the LOS. The classification tree algorithm highlighted that the risk of long hospital stay changed with the different combinations of the six BGA items, a history of falls being the major split item for a long stay. Furthermore, the combination of a history of falls, male gender, cognitive impairment, and age under 85 years identified elderly ED patients with the highest risk of a long hospital stay. In our study, a history of falls was a key predictor of a long hospital stay. This relationship may be explained in part by the fact that falls in older adults are a marker of morbidities whose consequences are gait and balance disorders, but also a loss of independence (18,19). Indeed, most identified risk factors for falls are intrinsic (i.e., subject-related) and usually include polypharmacy, psychoactive medications, muscle weakness, sensory deficits, and a cognitive decline (18–20). The state of health of patients with a history of recent falls is
Screening Tool for Long Hospital Stay
therefore characterized by a cumulative effect of chronic diseases (21). These effects become even more pronounced and diverse with age, contributing to a vicious cycle of increasing frailty and increasing risk of dependence, which may increase the LOS (21,22). Male gender was also associated with a long hospital stay in our study. The prevalence of male gender was significantly higher in the group of inpatients with long LOS compared to those with intermediate LOS (p = 0.002), and there was a significant trend to have more men in the long LOS group compared to the short LOS group (p = 0.050). A higher age-related morbidity burden in men compared to women could be the main explanation for this finding. Indeed, it recently has been reported that among elderly patients, male gender is associated with a greater number of morbidities on admission than female gender, and greater impairments in biological markers of clinical severity such as serum albumin, as well as in functional status (23). In addition, male gender has been identified as a strong predictor for adverse events after an organ failure (23–26). Moderate-to-severe cognitive impairment was associated with a long hospital stay in our study. Acute medical illnesses and a change of environment often lead to an increased risk of confusion, agitation, and behavioral troubles in persons with dementia (27). Adverse outcomes related to cognitive decline could explain the long LOS found in our study. We also suggest that an increase in disability in demented inpatients may help to explain the long hospital stay. Indeed, older patients often experience a decline in function associated with hospitalization for acute medical illnesses, precipitating disability in activities of daily living (27–29). It has been shown that dementia independently predicts failure to recover in activities of daily living that may explain in part the prolonged hospital stay among inpatients with moderate-to-severe cognitive decline (29). The general idea is that the LOS is associated with the degree of frailty, and that frail elderly inpatients have a longer LOS compared to transitionally frail inpatients, who themselves have a longer LOS than vigorous inpatients. However, our results showed that this association is not linear. Indeed, between-group comparisons highlighted that the prevalence of oldest-old inpatients (i.e., $ 85 years) was higher within the group of intermediate LOS compared to short LOS (p = 0.012) but not compared to long LOS (p = 0.264). This result demonstrates that aging should not be considered a disease, and confirmed that oldest-old inpatients have a heterogeneous frailty status. Our results also show that the risk of a long LOS may change according to the combinations of the BGA items. The history of falls was the primary segmentation process, and participants with a history of falls in the
743
past 6 months had a broadly higher risk of a long hospital stay than those with no history of falls. Actually, the risk depended on the combination of having had a fall with the other risk factors, as shown by the classification tree algorithm. For example, men with a history of recent falls but no moderate-to-severe cognitive impairment and no polypharmacy had a low though non-significant RR for a long hospital stay (RR = 4.0 with p = 0.516), although those who did not fall but presented with cognitive impairment, polypharmacy, or no social isolation had a surprisingly higher and significant RR of a long hospital stay (RR = 8.5 with p = 0.009). Our approach of utilizing the BGA to identify older inpatients at high risk of a long hospital stay provides several advantages over previous CGAs and regressions models. First, the BGA is based on only six items, is easy to measure, accessible, takes little time to administer, and requires no specific competence or equipment. Second, the classification tree algorithm is simple and requires no score to be calculated, as the risk profile gives rise to an end group with a matching risk for a long hospital stay. Furthermore, it does not require any a priori distributional assumption and knowledge about the underlying relationships between dependent (i.e., long hospital stay) and independent (i.e., BGA items) variables. Hence, this method is useful in situations where there are interactions among variables; the cases are partitioned and each group is analyzed separately. In addition, classification tree algorithm allows the construction of directly applicable risk profiles for long hospital stay. Compared to regression analyses where all predictors must be measured to identify the risk profile, few predictors – in our case the BGA six items – are required to recognize the risk profile. Lastly, although risk profiles do not provide information about causal relationships, knowledge about the combination of the 6 BGA items of end groups might provide a more purposeful assignment of recommended standards of care among older patients admitted to the ED. Limitations Some limitations of this study need to be considered. First, selected older inpatients probably were not representative of the general population of older adults admitted to the ED. Only half of inpatients met inclusion criteria, and our study was limited to one single ED and one single geriatric acute care unit. Also, given the many differences in health care systems across countries, the generalization of our findings could be limited to the French health care system. Second, as the number of participants in end nodes was low, equivocal or nonsignificant results could be the result of a lack of power. Third, the six BGA items did not take into account the
744
O. Beauchet et al.
causes (i.e., organ failure and geriatric syndromes) of hospitalization of participants. Fourth, one of the six BGA items was the reported history of falls. This information was collected by questioning each subject or caregiver, which may have led to a recall bias (30). Fifth, although the MMSE is the best-studied and most widely used instrument to screen cognitive impairment, its main limitation, especially in the ED context, is that it is a timeconsuming test (e.g., requires about 15 min to perform). CONCLUSION In conclusion, prediction of LOS with a six-item BGA was possible in the studied sample of older inpatients admitted to the ED. The risk of a long hospital stay changed depending on the different combinations of the six items on the BGA. The combination of a history of a recent fall, male gender, cognitive impairment, and age under 85 years identified the elderly ED patients with the highest risk of a long hospital stay requiring geriatric care and planning for discharge. Acknowledgment—The author would like to thank all participants involved in the present study.
REFERENCES 1. Li CM, Chen CY, Li CY, Wang WD, Wu SC. The effectiveness of a comprehensive geriatric assessment intervention program for frailty in community-dwelling older people: a randomized, controlled trial. Arch Gerontol Geriatr 2010;50:S39–42. 2. Van Craen K, Braes T, Wellens N, et al. The effectiveness of inpatient geriatric evaluation and management units: a systematic review and meta-analysis. J Am Geriatr Soc 2010;58:83–92. 3. McCusker J, Verdon J, Tousignant P, de Courval LP, Dendukuri N, Belzile E. Rapid emergency department intervention for older people reduces risk of functional decline: results of a multicenter randomized trial. J Am Geriatr Soc 2001;49:1272–81. 4. Lang PO, Zekry D, Michel JP, et al. Early markers of prolonged hospital stay in demented inpatients: a multicentre and prospective study. J Nutr Health Aging 2010;14:141–7. 5. Graf CE, Zekry D, Giannelli S, Michel JP, Chevalley T. Efficiency and applicability of the comprehensive geriatric assessment in the emergency department: a systematic review. Aging Clin Exp Res 2011;23:244–54. 6. Graf CE, Zekry D, Giannelli S, Michel JP, Chevalley T. Comprehensive geriatric assessment in the emergency department. J Am Geriatr Soc 2010;58:2032–3. 7. Hoogerduijn JG, Schuurmans MJ, Korevaar JC, Buurman BM, de Rooij SE. Identification of older hospitalised patients at risk for functional decline, a study to compare the predictive values of three screening instruments. J Clin Nurs 2010;19:1219–25. 8. Hoogerduijn JG, Schuurmans MJ, Duijnstee MS, de Rooij SE, Grypdonck MF. A systematic review of predictors and screening instruments to identify older hospitalized patients at risk for functional decline. J Clin Nurs 2007;16:46–57.
9. Reuben DB. Medical care for the final years of life: ‘‘When you’re 83, it’s not going to be 20 years’’ JAMA 2009;302:2686–94. 10. Reuben DB. Better care for older people with chronic diseases: an emerging vision. JAMA 2007;298:2673–4. 11. Covinsky KE, Palmer RM, Fortinsky RH, et al. Loss of independence in activities of daily living in older adults hospitalized with medical illness: increased vulnerability with age. J Am Geriatr Soc 2003;51:451–8. 12. Zanocchi M, Maero B, Maina P, et al. Factor predicting a prolonged hospital stay in elderly patients. Minerva Med 2002;93:135–43. 13. Hirsch CH, Sommers L, Olsen A. The natural history of functional morbidity in hospitalized older patients. J Am Geriatr Soc 1996;38: 1296–303. 14. Aminzadeh F, Dalziel WB. Older adults in the emergency department: a systematic review of patterns of use, adverse outcomes, and effectiveness of interventions. Ann Emerg Med 2002;39:238–47. 15. McCusker J, Cardin S, Bellavance F, et al. Return to the emergency department among elders: patterns and predictors. Acad Emerg Med 2000;7:249–59. 16. Hastings SN, Schmader KE, Sloane RJ, et al. Adverse health outcomes after discharge from the emergency department—incidence and risk factors in a veteran population. J Gen Intern Med 2007;22: 1527–31. 17. Folstein MF, Folstein SE, McHugh PR. ‘‘Mini-mental state’’. A practical method for grading the cognitive state of the patient for the clinician. J Psychiatr Res 1975;12:189–98. 18. American Geriatrics Society, British Geriatrics Society, and American Academy of Orthopedic Surgeons Panel on Falls Prevention. Guideline for the prevention of falls in older persons. J Am Geriatr Soc 2001;49:664–772. 19. Oliver D, Daly F, Martin FC, McMurdo ME. Risk factors and risk assessment tools for falls in hospital in-patients: a systematic review. Age Ageing 2004;33:122–30. 20. Bloem BR, Steijns JA, Smits-Engelsman BC. An update on falls. Curr Opin Neurol 2003;16:15–26. 21. Fried LP, Bandeen-Roche K, Kasper JD, Guralnik JM. Association of comorbidity with disability in older women: the Women’s Health and Aging Study. J Clin Epidemiol 1999;52:27–37. 22. Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56: M146–56. 23. Brown CJ, Redden DT, Flood KL, Allman RM. The underrecognized epidemic of low mobility during hospitalization of older adults. J Am Geriatr Soc 2009;57:1660–5. 24. Rozzini R, Sleiman I, Maggi S, Noale M, Trabucchi M. Gender differences and health status in old and very old patients. J Am Med Dir Assoc 2009;10:554–8. 25. Pitt B. Male gender, diabetes, COPD, anemia, and creatinine clearance < 30 mL/min predicted hospitalization after heart failure diagnosis. Ann Intern Med 2010;152:JC4–2, JC4-3. 26. Lee TC, Wang HP, Chiu HM, et al. Male gender and renal dysfunction are predictors of adverse outcome in nonpostoperative ischemic colitis patients. J Clin Gastroenterol 2010;44:e96–100. 27. Zekry D, Herrmann FR, Grandjean R, et al. Does dementia predict adverse hospitalization outcomes? A prospective study in aged inpatients. Int J Geriatr Psychiatry 2009;24:283–91. 28. Ehlenbach WJ, Hough CL, Crane PK, et al. Association between acute care and critical illness hospitalization and cognitive function in older adults. JAMA 2010;303:763–70. 29. Boyd CM, Landefeld CS, Counsell SR, et al. Recovery of activities of daily living in older adults after hospitalization for acute medical illness. J Am Geriatr Soc 2008;56:2171–9. 30. Cummings SR, Nevitt MC, Kidd S. forgetting falls: the limited accuracy of recall of falls in the elderly. J Am Geriatr Soc 1988; 36:613–6.
Screening Tool for Long Hospital Stay
ARTICLE SUMMARY 1. Why is this topic important? Because of the increased number of frail older adults hospitalized in the Emergency Department, current hospital care systems are driven by the high burden of chronic diseases and their adverse consequences, worsening the patient’s quality of life and healthcare professional practice. 2. What does this study attempt to show? This study attempts to show that a brief geriatric assessment carried out in older patients admitted in the ED may predict the risk of a long hospital stay in a geriatric acute care unit. 3. What are the key findings? The combination of a history of falls, male gender, cognitive impairment and age under 85 years, identified elderly patients admitted and at highest risk of long hospital stay. 4. How is patient care impacted? Use of a brief geriatric assessment (BGA) to identify frail older patients early could improve their discharge planning.
745