Presented at the Society for Academic Emergency Medicine Annual Meeting, Chicago, IL, May 16–19, 2007.
Emergency Department Case-finding for High-risk Older Adults: The Brief Risk Identification for Geriatric Health Tool (BRIGHT)
Version of Record online: 10 JUL 2008
© 2008 by the Society for Academic Emergency Medicine
Academic Emergency Medicine
Volume 15, Issue 7, pages 598–606, July 2008
How to Cite
Boyd, M., Koziol-McLain, J., Yates, K., Kerse, N., McLean, C., Pilcher, C. and Robb, G. (2008), Emergency Department Case-finding for High-risk Older Adults: The Brief Risk Identification for Geriatric Health Tool (BRIGHT). Academic Emergency Medicine, 15: 598–606. doi: 10.1111/j.1553-2712.2008.00157.x
Funding was provided by the Waitemata District Health Board, University of Auckland, and an Auckland University of Technology research grant (to MB).
- Issue online: 10 JUL 2008
- Version of Record online: 10 JUL 2008
- Received January 2, 2008; revisions received March 13 and April 2, 2008; accepted April 3, 2008.
- risk assessment;
- emergency services;
- case finding;
- geriatric assessment
Objective: The objective was to test the ability of the Brief Risk Identification for Geriatric Health Tool (BRIGHT) to identify older emergency department (ED) patients with functional and physical impairment.
Methods: This was a cross-sectional study in which 139 persons ≥75 years, who presented to an urban New Zealand ED over a 12-week period, completed the 11-item BRIGHT case-finding tool. Then, within 10 days of their index ED visit, 114 persons completed a comprehensive geriatric assessment. A “yes” response to at least 3 of the 11 BRIGHT items was considered “positive.” Primary outcome measures were instrumental activities of daily living (IADL), cognitive performance scale (CPS), and activities of daily living (ADL).
Results: The BRIGHT-identified IADL deficit (64% prevalence) with a sensitivity of 0.76, specificity of 0.79, and receiver operating characteristic (ROC) of 0.83 (95% confidence interval [CI] = 0.74 to 0.91, p < 0.01); cognitive deficit (35% prevalence) sensitivity of 0.78, specificity of 0.54, and ROC of 0.66 (95% CI = 0.55 to 0.76, p = 0.006); and ADL deficit (29% prevalence) sensitivity of 0.83, specificity of 0.53, and ROC of 0.64 (95% CI = 0.53 to 0.75, p = 0.020). Positive likelihood ratios (LR+) for the three outcomes of interest were 3.6, 1.7, and 1.8, respectively. Negative likelihood ratios (LR−) were 0.3, 0.4, and 0.3.
Conclusions: The 11-item BRIGHT successfully identifies older adults in the ED with decreased function and may be useful in differentiating elder patients in need of comprehensive assessment.
ACADEMIC EMERGENCY MEDICINE 2008; 15:598–606 © 2008 by the Society for Academic Emergency Medicine
Older adult patients commonly present to the emergency department (ED) with a complex mix of health and psychosocial issues. Functional dependence at the time of ED admission is a risk factor for subsequent health and functional decline, ED and hospital readmission, institutionalization, and death.1–4 Other risk factors for adverse outcomes include living alone, lack of social support, multiple comorbidity, polypharmacy, and diagnoses such as cardiovascular disease, diabetes, cognitive impairment, and depression.5,6 Research has demonstrated that systematic case-finding, assessment, and prompt intervention can forestall health and functional decline, decrease hospital length of stay and readmission, and decrease long-term care admission.7,8 In the Discharge of Elderly from the Emergency Department (DEED) II trial, ED-based comprehensive geriatric assessment followed by multidisciplinary community outreach resulted in a lower rate of repeat ED visits and a greater degree of physical and mental function compared to usual care.9 However, routine comprehensive geriatric assessment is not feasible in the ED setting due to the time and specialist training required. It is also not likely to be cost-efficient to follow-up all older ED patients after discharge. A relatively quick and easily administered case-finding tool is needed. Although there have been several reports of ED risk identification tools,10–12 they are not part of an integrated comprehensive geriatric assessment system, increasingly being called for in practice guidelines.13
The New Zealand Ministry of Health recently selected the internationally validated inter-Residential Assessment Instrument (interRAI) Minimum Data Set for Home Care Version 2.0 (MDS-HC)14–16 as the preferred assessment system. The interRAI MDS-HC assessment is extremely thorough, but resource-intensive, requiring 40 to 60 minutes to administer by specially trained staff. The Brief Risk Identification for Geriatric Health Tool (BRIGHT) was developed to provide a self-report tool compatible with the interRAI MDS-HC. The purpose of this study was to test the reliability and validity of the 11-item BRIGHT in an older adult ED population.
This was a cross-sectional study17 using a convenience sample of older adults in the ED comparing the results of the BRIGHT with a comprehensive interRAI geriatric assessment conducted within 10 days of their ED visit. The study was approved by the New Zealand Ministry of Health Northern Regional Ethics Committee.
Study Setting and Population
The study was conducted in a 463 bed acute care hospital in the Auckland, New Zealand, region. The hospital’s ED (known as the Emergency Care Centre) had approximately 44,000 patient-visits in 2005; 19% were aged 75 years or older, and of those, 84% were classified as nonurgent (triage 3–5). The ED is the point of entry for all hospital acute care admits. Persons eligible for enrollment included all those aged 75 years and older (65 years for Maori and Pasifika elders18) who presented to the ED with a nonurgent complaint (triage level 3–5) during a convenience sample of 4-hour time blocks (Monday–Friday, 8am–8pm) over a 12-week period between December 2005 and March 2006 (New Zealand summer season). The study protocol stated that patients should not be disturbed when sleeping, undergoing medical procedures, or in distress. Cognitively impaired patients were only enrolled if their family was available to assist in completing the BRIGHT. A separate proxy consent process was used for cognitively impaired patients.
Brief Risk Identification for Geriatric Health Tool. The 11-item BRIGHT was developed to be self-administered or administered with assistance by untrained caregivers. The items were based on a previously tested mailed questionnaire19 and the Self Reliance and Contact Assessment questionnaires developed by the interRAI group.20,21 The items address the following common geriatric issues: help with bathing, personal hygiene, dressing the lower body, getting around indoors, difficulty making decisions about everyday activity, shortness of breath, recent falls, perception of general health, memory problems, ability to do ordinary housework, and depression. The BRIGHT uses easily understood lay language and can be administered in a variety of settings or as a postal questionnaire. For each item, individuals are asked to consider “how you have been in the last three months” and reply “yes” or “no.” For example, “Do you usually need someone to help you dress your lower body?” and “Have you been bothered by feeling down, depressed, or hopeless?” The BRIGHT screen was designed to quickly identify those older adults who would benefit most from a comprehensive geriatric assessment.
Comprehensive Geriatric Assessment. Comprehensive geriatric assessments were conducted using the interRAI MDS-HC (Home Care) or the interRAI-AC (Acute Care).14,15,22–25 The interRAI-AC was chosen because it is tailored to assess people who are hospitalized and the interRAI-HC because it is specific to people residing in the community and therefore those who were discharged from the ED. Instrumental activities of daily living (IADL), activities of daily living (ADL), and cognitive performance scales (CPS) are comparable across the two interRAI instruments.
The interRAI assessment system is designed to inform an interdisciplinary intervention. The instruments are standardized and require 12–16 hours of staff training. The interRAI comprehensive geriatric assessment was selected as the “criterion standard” because it is the emerging system of choice in New Zealand for geriatric assessment. These instruments require 30–60 minutes to administer, which is clearly unfeasible in an ED. All assessors were health care professionals (four nurses and one dietician), experienced in comprehensive geriatric assessment and given full training in the use of the interRAI-HC and interRAI-AC.
Three scales were calculated from the interRAI MDS-AC and the interRAI-HC assessments. The IADL Involvement Scale (maximum score of 21) has been validated against other established measures, such as Lawton’s IADL scale.15,26 The CPS (maximum score of 6) has been found to correlate with the Mini-Mental Status Exam (MMSE) and provides a measure of overall cognitive function.15,27 The ADL Short Form (maximum score of 16) has been validated with the Barthel’s ADL assessment.15,28
The primary investigator (MB) trained screening research assistants (RAs; n = 6) regarding sensitivity in communicating with older adults and the research protocol (including recruitment, participant and proxy informed consent, use of the brief screen, and documentation). During selected shifts, a list of older adults in the ED was obtained from the electronic hospital census. RAs contacted older patients and their families (if present) either while waiting for treatment or while in the observation area. The BRIGHT could be completed either by the older person herself or himself or with assistance by their family or the RA. ED visit and demographic data were also collected. After the participant’s ED visit, trained assessors (n = 6) conducted a comprehensive geriatric assessment either in the hospital or at the older adult's home. The assessors were blinded to the participant’s BRIGHT score. The interRAI-AC assessment was used for those admitted to hospital and as suggested by the interRAI developers, conducted 24 to 72 hours after admission. The interRAI MDS-HC was used for those discharged home, at a time convenient to the older adult within 10 days of the ED visit. Per study protocol, older adults with significant identified needs were referred to the appropriate interdisciplinary and multidisciplinary team for further assessment and intervention.
ED visit, demographics, and BRIGHT data were entered into SPSS (Version 14.0 for Windows, SPSS Inc., Chicago, IL). Descriptive statistics and internal reliability (Cronbach’s alpha coefficient = 0.73) for the BRIGHT were examined. The BRIGHT score was the total of all positive item responses (possible range 0 to 11), with higher scores indicating greater impairment. Assessors entered the interRAI MDS-HC data into an approved interRAI computer software package (AIS Canada, Toronto, Ontario, Canada). The interRAI-AC teleform hard copies were delivered to the Hirdes research group at the University of Waterloo, Ontario, Canada, for importing and scoring. Descriptive statistics were examined for the IADL Involvement Scale, CPS, and ADL Short Form scores. Persons with a score of 1 or more were considered to have functional or cognitive impairment.15,29 A CPS score of 1 or more corresponds to a MMSE score of 25/30 or below.15,27 The ability of the BRIGHT to identify elders with cognitive or functional decline was evaluated using test statistics, including specificity and sensitivity across BRIGHT scores with 95% confidence intervals (CIs) calculated by the Wilson method.30 Receiver operating characteristic (ROC) curves31,32 and likelihood ratios (LRs)33 were also used to identify an optimal BRIGHT cutoff score.
A total of 338 eligible older adults entered the ED during study shifts; 139 (41%) were recruited for the study. RAs often reported missing people who were out of the department for tests or not wanting to disturb those who were sleeping. Participants and nonparticipants were similar in age (82.5 and 82.6 years, respectively) and gender (67% and 63% females, respectively; see Table 1). Mean hospital length of stay was somewhat longer for participants (5.9 days) compared to nonparticipants (5.0 days). Participants were more likely to be New Zealanders of European descent (81%) compared to nonparticipants (66%). Most participants (71%) were admitted into hospital acute care rather than discharged back to the community (29%). Among the 139 participants, 138 had a completed BRIGHT.
|Ethnicity||Participants (n = 139)||Nonparticipants (n = 199)|
|Percent||Total Number||Percent||Total Number|
|Age, yr (mean ± sd)||82.5 (±5.4)||82.6 (±5.4)|
|Age range, yr†||65–102||68–101|
|Length of stay, days (mean ± SD)||5.9 (±8.8)||5.0 (±7.3)|
The majority (75%) of participants had assistance from a visitor or the RA to complete the BRIGHT; 25% completed the BRIGHT independently. Comprehensive geriatric assessment was completed for 114 (82%) participants (see Figure 1); the average time between the BRIGHT screen and the interRAI full assessment was 3.98 days (SD ± 4.23 days).
The BRIGHT score ranged from 0 to 10; the median was 3 (mean ± SD = 3.3 ± 2.5). Of the 11 items, the 3 most commonly reported positive items were needing help with housekeeping (61%), recent falls (46%), and depression (44%, see Figure 2). The least often reported positive response was difficulty with personal hygiene (8%).
The IADL Involvement Scale score ranged from 0 to 21; the median score was 3. Sixty-four percent reported difficulty with one or more IADL. The CPS score ranged from 0 to 3; the median score was 0. Thirty-five percent scored 1 or more. The ADL Short Form score ranged from 0 to 12; the median score was 0. Twenty-nine percent reported difficulty with one or more ADL.
Judged against the outcome variables, the BRIGHT performed best across cutoff scores of 3 or more or 4 or more (see Table 2 and Figures 3 and 4). Predicting IADL deficit, the sensitivity and specificity of the BRIGHT were 0.76 and 0.79 with a cutoff of 3 or more, and the area under the ROC was 0.83 (95% CI = 0.74 to 0.91), indicating moderate accuracy. Predicting cognitive performance, the sensitivity and specificity of the BRIGHT were 0.70 and 0.74 with a cutoff of 4 or more; and the area under the ROC was 0.73 (95% CI = 0.62 to 0.84), again indicating moderate accuracy. Predicting ADLs, the BRIGHT performed the poorest, the sensitivity and specificity were 0.69 and 0.70 with a cutoff of 4 or more, and the area under the ROC was 0.66 (95% CI = 0.54 to 0.78), indicating low accuracy.
|BRIGHT Cutoff Score (n = 114) (95% CIs30)||Two or more (71%)||Three or more (57%)||Four or more (43%)|
|IADL deficit prevalence, 0.64 (n = 106)|
|Sensitivity||0.85 (0.75, 0.92)||0.76 (0.65, 0.85)||0.59(0.47, 0.70)|
|Specificity||0.58 (0.42, 0.72)||0.79 (0.64, 0.90)||0.89(0.76, 0.96)|
|Cognitive deficit prevalence, 0.35 (n = 105)|
|Sensitivity||0.81 (0.66, 0.91)||0.78 (0.63, 0.89)||0.70 (0.54, 0.83)|
|Specificity||0.34 (0.24, 0.46)||0.54 (0.43, 0.66)||0.74 (0.62, 0.83)|
|ADL deficit prevalence, 0.29 (n = 99)|
|Sensitivity||0.90 (0.74, 0.96)||0.83 (0.65, 0.92)||0.69 (0.51, 0.83)|
|Specificity||0.37 (0.27, 0.49)||0.53 (0.41, 0.64)||0.70 (0.58, 0.79)|
The positive likelihood ratio (LR+) for a BRIGHT cutoff score of 3 or more to identify IADL, cognitive, and ADL deficits were 3.6, 1.7, and 1.8, respectively, and for a score of 4 or more, the LR+ were 5.6, 2.7, and 2.3 respectively. The negative likelihood ratio (LR−) ranged from 0.5 to 0.3 for all indicators with a cutoff of 3 or 4.34–36
Finally, test statistics for the interRAI MDS-HC and the interRAI MDS-AC were stratified by admission status (and thus MDS-HC or MDS-AC instrument) for the BRIGHT cutoff value of 3. However, the sample for discharged patients was small (n = 29); thus, the results should be interpreted with caution. In general, there were modest improvements in LRs for a positive BRIGHT (using a cutoff of 3) among discharged patients for ADL, IADL, and cognitive deficit (see Table 3). LRs for a negative BRIGHT were somewhat improved (lower) in the cohort of admitted patients for IADL and cognitive deficit. There were no false-negatives for discharged patients regarding ADL, resulting in improved ADL LR− for discharged patients.
|Admitted (n = 76)||0.79||0.75||0.56||0.90||3.14||0.29|
|Discharged (n = 28)||0.64||0.82||0.78||0.70||3.61||0.44|
|Admitted (n = 76)||0.83||0.46||0.81||0.50||1.53||0.37|
|Discharged (n = 27)||0.50||0.71||0.83||0.33||1.75||0.70|
|Admitted (n = 68)||0.80||0.40||0.77||0.44||1.32||0.51|
|Discharged (n = 29)||1.00||0.27||1.00||0.30||1.37||0.00|
We found the BRIGHT to have good internal reliability and reasonable validity when compared to comprehensive geriatric assessment in a sample of older adult ED patients. Older adult functional ability at the time of ED admission has been found to be the most predictive measure of subsequent health care service use, future functional decline, and death.4–6 Therefore, a systematic method to identify functional ability is needed. We found the BRIGHT to have reasonable ability to identify older adults in the ED who have functional deficits. In this study, 64% of older adults answering the BRIGHT were found to have difficulties with IADL. The tool had good ability to detect IADL deficit with a cut off of 3 (ROC area under the curve [AUC] = 0.83) or 4 (AUC = 0.81). At a cutoff point of 4, the tool demonstrated a reasonable ability to identify cognitive deficit (AUC = 0.73) and ADL deficit (AUC = 0.66).
The BRIGHT has similar discrimination ability as other tools designed for case-finding in the ED. McCusker and colleagues12 implemented the six item Identification of Seniors at Risk (ISAR) with older ED patients. Patients who scored 2 or more on the 6-item screen were found to adequately predict one or more adverse health outcome, such as institutionalization, functional decline, or death; the ISAR identified 72% of older adults with functional decline during the 6 months after the index visit (AUC = 0.71, 95% CI = 0.68 to 0.74).37,38 Another example is the Triage Risk Screening Tool (TRST), based on five risk factors determined by a panel of gerontology clinical experts.39 Patients who had two or more positive risk factors were found to be more likely to have subsequent ED visits, hospitalizations, and institutionalization.40 The Blaylock Risk Assessment Screening Score (BRASS) is a 10-item index of older adult risk.41 A study by Mistiaen et al.10 found that patients identified as at risk by the BRASS index were more often admitted to hospital from the ED and were more likely to have a longer hospital length of stay, but performed more poorly in predicting problems after discharge (sensitivities 0.16–0.30).
Unfortunately, across available screening tools for ED use (including the BRIGHT, TRST, ISAR), prognostic test characteristics are less than ideal. Although adding items and sophistication to the available screening tools would likely increase prognostic test characteristics, the tools would become cumbersome and require specialist assessment skills, transitioning from a brief screen to a comprehensive assessment. The BRIGHT screen exhibits a better ability to identify those with an IADL deficit compared to other screens. The LR+ for IADL deficits was 3.6 for a cutoff of 3, compared to the ISAR LR+ of 2.2 for a cutoff of 2,42 and the TRST LR+ of 1.4 for a cutoff of 233 (see Table 4). A “strong” prognostic/diagnostic test would have LR+ >10 and LR− <0.1, and a medium prognostic/diagnostic test would have LR+ of 5 to 10 and a LR− of 0.1 to 0.2.43
|Current study||114, aged 75+ ED visit||BRIGHT ≥3 (of 10)||1- to 10-day functional and physical deficit determined by comprehensive geriatric assessment using the interRAI MDS|
|Dendukuri et al.42||1,122 (study 1) and 520 (study 2) aged 65+ discharged home from ED visit||ISAR ≥2 (of 6 items)||Baseline (ED visit) severe functional impairment determined by OARS ADL scale (14 items; partial or complete dependence ≥3 ADLs)||2.2||0.1|
|Hustey et al.33||650, aged 65+ discharged home from ED visit||TRST cognitive impairment OR ≥2 (of 6 items)||30-day decline determined by telephone interview|
|ADL (decline in ≥1 ADL)||1.6||0.6|
|IADL (decline in ≥1 IADL)||1.4||0.8|
|Fan et al.45||120, aged 64+ discharged home from ED||TRST ≥2 (of 5 items)||30-day composite (ED revisit, hospital admission, long-term care placement) determined by database review||1.4||0.7|
|McCusker et al.38||676 (validation sample), aged 65+ nonacute ED visit||ISAR ≥2 (of 6 items)||6-month adverse outcome (death, long-term care placement) or ≥3-point decline in OARS ADL scale (of 28) by telephone interview||2.0||0.3|
The BRIGHT was designed to quickly identify older adults that would benefit from a full geriatric assessment. The 11-item instrument can be easily self-administered or administered by untrained caregivers or staff. It was envisaged that the BRIGHT would be used as part of a two-stage assessment process. The data from this study support the BRIGHT’s ability to accurately identifying older adults who would likely benefit from further evaluation. This type of two-tiered assessment system in the ED has been described by others.12,39
More studies are needed to provide information about the BRIGHT’s ability to predict outcomes, such as short- and long-term ED recidivism and mortality, as well as further investigation with a larger sample of older adults discharged directly back into the community. Although it is important to identify high-risk older people being directly discharged to the community, it is also important to identify high-risk older people being admitted. Admitted older people are at high risk for delirium and loss of ADL ability during admission. Early interventions could be implemented to prevent complications during the inpatient stay.44
Outcomes for ED case-finding and follow-up programs have demonstrated varied and inconclusive results.9,12,32,38,42,45 Variation may be explained by the population characteristics, instrumentation, or perhaps most importantly, by how the comprehensive assessment is defined and performed. Optimal benefit occurs with an interdisciplinary approach that coordinates and integrates treatment plans.7
The DEED II study by Caplan et al.9 is an example of ED case-finding and community intervention that showed a lower rate of admissions to hospital and ED within 30 days and 18 months after the initial ED visit. They found no difference in long-term care admissions or mortality. McCusker et al.12,46 found that identified patients were more likely to receive services in the community and to have decreased functional decline, but unexpectedly, were also more likely to return to the ED. The SIGNET model utilizes the TRST.11,40 This tool was found to identify those elders in the ED who were more likely to be readmitted to the hospital and ED. The subsequent follow-up intervention did not demonstrate decreased hospital readmission, but did show lower long-term care admissions at 30 days post–index ED visit. Participants were found to have increased patient satisfaction with care. It is clear that intervention, beyond referring older adults back to primary care, is needed and that it should incorporate specialized geriatric intervention and follow-up.47 More investigation is needed to identify what aspects of follow-up care can be of greatest benefit to high-risk older adults identified in the ED.
This study was conducted in a single ED with a predominantly New Zealand European (Caucasian) sample. Among eligible patients, 41% completed the BRIGHT. Eighteen percent of those who completed the BRIGHT were lost to follow-up for the comprehensive assessment, thus introducing the possibility of follow-up bias. In addition, the sample size was small; for example, the 95% confidence limits around the point estimate for a positive BRIGHT (57%, based on a cutoff score of 3) was ±9.
This is the first application of the BRIGHT. Further testing across settings and locations with larger samples and different risk profiles is needed. In populations with less (or more) morbidity, the BRIGHT might perform differently. For example, this study was designed to include only nonurgent older adults in the ED and was conducted in a setting with a low proportion of discharged patients (29%). Further research is needed to compare the utility of the BRIGHT screen to identify high-risk older people discharged from the ED and those who are admitted.
In addition to the incomplete cognitive geriatric assessment follow-up bias and the single-center, largely Caucasian sample limiting external validity, numerous other forms of bias are possible. For example, nonconsecutive patient recruitment possibly overestimates diagnostic accuracy.48 Regarding measurement error, the interRAI-AC is not widely used to date and has limited validation evidence.25
The BRIGHT demonstrated a reasonable ability to identify functional issues in older adults presenting to the ED. This case-finding tool was designed to be used in combination with the interRAI assessment system and to be able to be quickly and efficiently self-administered by older adults or their family caregivers. The tool compares favorably with other reported brief case-finding tools and could be used as a basis for early intervention for older adults at risk.
We thank Helen Frances and the staff of Waitemata District Health Board Home and Older Adult Services, Needs Assessment Service Coordination and Emergency Care Centre. We also acknowledge Karen Fredrickson, Tissa Fernandez, Mike Armour, Liz Sherpa, and Jenny Oettli who performed data collection and assessments. We also thank Professor John Hirdes from University of Waterloo, Ontario, for his assistance with this research.
- 1The discharge of elderly patients from an accident and emergency department: functional changes and risk of readmission. Age Ageing. 1990; 19:415–8., , , , .
- 14Integrated health information systems based on the RAI/MDS series of instruments. Healthc Manage Forum. 1999; 12(4):30–40., , , et al.
- 22A new assessment for elders admitted to acute care: reliability of the MDS-AC. Aging (Milano). 2001; 13:316–30., , , et al.