Presented at the Society for Academic Emergency Medicine annual meeting, San Francisco, CA, May 2006.
Short-term Functional Decline and Service Use in Older Emergency Department Patients With Blunt Injuries
Article first published online: 6 JUL 2010
© 2010 by the Society for Academic Emergency Medicine
Academic Emergency Medicine
Volume 17, Issue 7, pages 679–686, July 2010
How to Cite
Wilber, S. T., Blanda, M., Gerson, L. W. and Allen, K. R. (2010), Short-term Functional Decline and Service Use in Older Emergency Department Patients With Blunt Injuries. Academic Emergency Medicine, 17: 679–686. doi: 10.1111/j.1553-2712.2010.00799.x
Dr. Wilber was funded by the American Geriatrics Society’s Dennis W. Jahnigen Career Development grant, supported by the John A. Hartford Foundation and the Atlantic Philanthropies. The Summa Foundation provides support for the Emergency Medicine Research Center.
Supervising Editor: Christopher R. Carpenter, MD, MSc.
- Issue published online: 6 JUL 2010
- Article first published online: 6 JUL 2010
- Received January 11, 2010; revision received February 23, 2010; accepted March 2, 2010.
- geriatric assessment;
- activities of daily living;
- wounds and injuries
ACADEMIC EMERGENCY MEDICINE 2010; 17:679–686 © 2010 by the Society for Academic Emergency Medicine
Background: Injuries are a common reason for emergency department (ED) visits by older patients. Although injuries in older patients can be serious, 75% of these patients are discharged home after their ED visit. These patients may be at risk for short-term functional decline related to their injuries or treatment.
Objectives: The objectives were to determine the incidence of functional decline in older ED patients with blunt injuries not requiring hospital admission for treatment, to describe their care needs, and to determine the predictors of short-term functional decline in these patients.
Methods: This institutional review board–approved, prospective, longitudinal study was conducted in two community teaching hospital EDs with a combined census of 97,000 adult visits. Eligible patients were ≥ 65 years old, with blunt injuries <48 hours old, who could answer questions or had a proxy. We excluded those too ill to participate; skilled nursing home patients; those admitted for surgery, major trauma, or acute medical conditions; patients with poor baseline function; and previously enrolled patients. Interviewers collected baseline data and the used the Older Americans Resources and Services (OARS) questionnaire to assess function and service use. Potential predictors of functional decline were derived from prior studies of functional decline after an ED visit and clinical experience. Follow-up occurred at 1 and 4 weeks, when the OARS questions were repeated. A three-point drop in activities of the daily living (ADL) score defined functional decline. Data are presented as means and proportions with 95% confidence intervals (CIs). Logistic regression was used to model potential predictors with functional decline at 1 week as the dependent variable.
Results: A total of 1,186 patients were evaluated for eligibility, 814 were excluded, 129 refused, and 13 were missed, leaving 230 enrolled patients. The mean (±SD) age was 77 (±7.5) years, and 70% were female. In the first week, 92 of 230 patients (40%, 95% CI = 34% to 47%) had functional decline, 114 of 230 (49%, 95% CI = 43% to 56%) had new services initiated, and 76 of 230 had an unscheduled medical contact (33%, 95% CI = 27% to 39%). At 4 weeks, 77 of 219 had functional decline (35%, 95% CI = 29% to 42%), 141 of 219 had new services (65%, 95% CI = 58% to 71%), and 123 of 219 had an unscheduled medical contact (56%, 95% CI = 49% to 63%), including 15% with a repeated ED visit and 11% with a hospital admission. Family members provided the majority of new services at both time periods. Significant predictors of functional decline at 1 week were female sex (odds ratio [OR] = 2.2, 95% CI = 1.1 to 4.5), instrumental ADL dependence (IADL; OR = 2.5, 95% CI = 1.3 to 4.8), upper extremity fracture or dislocation (OR = 5.5, 95% CI = 2.5 to 11.8), lower extremity fracture or dislocation (OR = 4.6, 95% CI = 1.4 to 15.4), trunk injury (OR = 2.4, 95% CI = 1.1 to 5.3), and head injury (OR = 0.48, 95% CI = 0.23 to 1.0).
Conclusions: Older patients have a significant risk of short-term functional decline and other adverse outcomes after ED visits for injuries not requiring hospitalization for treatment. The most significant predictors of functional decline are upper and lower extremity fractures.
The overall burden of injuries in the United States is significant. In 2004, injuries resulted in 31 million emergency department (ED) visits, which represented 32% of all ED visits, 1.9 million hospitalizations and 167,184 deaths.1 Mortality and hospitalization rates for injuries increase dramatically in older patients, who are at the highest risk for both fatal and nonfatal injuries.1,2 Falls are the most common mechanism of injury in older persons visiting the ED (62%) and are the most common cause of injury-related death.1,2 Although injuries in older patients can cause serious injury and death, three-quarters of older ED patients with injuries are treated and released from the ED.2
Older ED patients, in general, have high rates of functional decline and other adverse health outcomes following their ED visits.3–5 Much of the research on functional decline in older patients has focused on screening and case-finding programs to identify patients who are at risk for long-term functional decline.6–10 These studies treat the ED visit as a sentinel event and attempt to identify those individuals who could benefit from geriatric assessment and care management programs.6–10
Studies on functional outcomes of older, injured patients discharged from the ED have demonstrated that some patients will develop functional decline over time after the injury.11–14 However, these patients may also develop abrupt changes in their functional abilities, related to either their injuries or their subsequent treatments. Those with lower extremity injuries (ranging from sprains to pubic rami fractures) that may affect their mobility are often treated as outpatients. Upper extremity immobilization by slings or splints may affect the ability of patients to dress, feed, or transfer themselves. The functional decline that injured older patients experience is likely to begin immediately and be most severe during the immediate post-ED time period. Our objectives were to determine the incidence of functional decline in older ED patients not requiring hospital admission for treatment of their injuries, to describe their care needs, and to determine the predictors of immediate postinjury functional decline in these patients.
This was a prospective, longitudinal cohort study. This study was approved by the institutional review board of Summa Health System.
Study Setting and Population
This study was performed in two adult EDs of Summa Health System, a major teaching hospital of the Northeastern Ohio Universities College of Medicine. Both EDs are urban midwestern community tertiary care centers. The ED census mirrors the county’s sociodemographic distribution. Approximately 25% of the 97,000 visits per year are in patients age 65 and older.
We recruited subjects on weekdays between 08:00 and 20:00 hours for 20 months. We included patients 65 years old or older who presented with a blunt traumatic injury less than 48 hours old, were able to understand the questions or had a proxy available, were able to communicate in English, and gave their consent to participate. Patients were excluded if the treating physician determined that their medical condition prevented them from being interviewed and there was no available informant; they resided in a skilled nursing or rehabilitation facility; they were admitted to the hospital for surgical treatment of their injury, major trauma, or another acute medical condition; they had severe baseline functional limitations; or they were previously enrolled.
Trained full-time research associates performed all initial and follow-up interviews. Interviewers collected the following information from the patient or proxy at baseline: demographic information, prior medical history, medication use, visual deficits, history of recent falls, ED visits, and hospitalizations. Information regarding the site of injuries, type of injuries, and treatment were obtained from the treating emergency physician (EP), the medical record, and billing data.
Mental status was assessed using the Six-Item Screener (SIS).15 A score of 4 or less defined cognitive impairment.16,17 The Charlson Comorbidity Index score was calculated for each patient based on medical history.18
Research associates used the Older Americans Resources and Services (OARS) Multidimensional Functional Assessment of Older Adults Questionnaire (OMFAQ)19 to assess functional status, social resources, and service use. The OARS activities of daily living (ADL) questionnaire has been validated for assessing physical activities of daily living (PADLs) and instrumental activities of daily living (IADLs) in ED patients and has been successfully used in both in-person interviews and telephone follow-up in ED-based studies.20,21 Seven PADLs (eating, dressing, grooming, walking, transfer, bathing, and continence) and seven IADLs (telephone, transportation, shopping, meal preparation, housework, medications, and money management) were tested. Each was scored on a 0 to 2 scale, giving a range of 0 to 28 (28 represents complete independence in all and 0 complete dependence).19 Respondents were asked to answer these questions about the patients’ abilities prior to their injury.
Social resources and current service use were assessed using a subset of OMFAQ questions in these dimensions. Questions from the OMFAQ regarding mental health, economic functioning, and physical health were not used. The questions on service use include nursing care, physical therapy, continuous supervision, checking services (only if not under continuous supervision), homemaker services, meal preparation, and personal care. For each positive response to a service used, respondents were then asked who provided the service. As an example, a subject may have received nursing care, defined as treatments or medications prescribed by a doctor, by an unpaid family member, an unpaid friend, or someone hired to provide the service.
A research associate contacted patients by telephone at 1 and 4 weeks after ED evaluation. At this time, the OARS ADL scale and the OMFAQ items on service use were repeated. Additionally, patients were asked about ED visits, physician visits, rehabilitation or extended care facility (ECF) admissions, and hospitalizations.
The primary outcome was functional decline at one week. Functional decline was defined as a reduction of 3 points on the 28-point OARS ADL scale from their ED visit to follow-up. This definition has been used when change in OARS ADL scale is an outcome.21 Secondary outcomes included functional decline at 4 weeks, unscheduled medical contacts, new service use, and unmet service needs at both 1 and 4 weeks. Unscheduled medical contacts included unscheduled visits to private physicians or the ED or admission to the hospital, skilled nursing facility, or inpatient rehabilitation. New services were defined as services being provided at follow-up that were not being provided at enrollment. Unmet service needs were defined as services for which the patient perceived a need, but were not being provided at follow-up. Potential predictors of these outcomes were derived from prior studies of functional decline after an ED visit6,11,21 and from clinical experience.
Descriptive data are presented as means and proportions with 95% confidence intervals (CIs). Potential predictors of functional decline at 1 week were first evaluated for an association with the primary outcome using Fisher’s exact test or t-tests as appropriate. Those variables with a univariate p-value of <0.15 were evaluated for an independent association with the primary outcome using logistic regression. Initially, all variables with a univariate p-value of <0.15 were added to the logistic regression model. Nonsignificant predictors were then manually and sequentially removed while examining the likelihood ratio test between models. For the final model, diagnostic testing was performed by evaluating the pattern of the deviance residuals and for influential clusters of observations. We also calculated the Hosmer and Lemeshow goodness-of-fit test to evaluate the model fit and the area under the receiver operating characteristic (ROC) curve for the model.
We anticipated a maximum number of significant predictors to be six. Using the rule-of-thumb of 10 times the number expected in the smaller criteria group per variable in the logistic model, an adequate sample size would be 60 patients with functional decline. We estimated that 25% of patients would have functional decline, which would require 240 patients.
During the enrollment period, 1,186 people were approached, 129 refused, 13 were missed, 814 were excluded, and 230 were enrolled. Reasons for exclusion included surgical admission (n = 220), medical admission (n = 196), skilled nursing facility (SNF) patient (n = 191), injury >48 hours old (n = 187), previously enrolled (n = 11), unable to follow up (n = 6), and severely limited baseline function (n = 3).
The mean (±SD) age of enrolled patients was 77 years (±7.5), and 162 of 230 (70%) were female. The majority of patients (210 of 230) were Caucasian (93%, 95% CI = 87% to 95%). The patient was the informant in 190 of 230 cases (83%); in the remainder, a proxy was the informant. Falls were the cause of injury in 196 of 230 (85%, 95% CI = 80% to 90%), motor vehicle crashes in 20 of 230 (9%), and other in 14 of 230 (6%). The median number of prescriptions on arrival to the ED was 4 (interquartile range = 2–8). Other baseline data are listed in Table 1.
|Numerator (n = 230)||Percent||95% CI|
|Fall in prior 6 months||69||30||24–36|
|ED visit in prior 6 months||71||31||25–37|
|ED visit in prior 30 days||28||12||8–17|
|Hospitalized in prior 6 months||48||21||16–27|
|Hospitalized in prior 90 days||23||10||6–15|
|Nursing home in prior 6 months||13||6||3–9|
|Charlson Comorbidity Score|
|Services in the past 6 months|
|Body part injured|
Outcomes at 1 and 4 weeks are listed in Table 2. Ninety-three of 230 of patients (40%, 95% CI = 34% to 47%) had functional decline at 1 week. The magnitude of the functional decline is shown in Figure 1. Within the first week, 76 of 230 of patients had an unscheduled medical contact (33%, 95% CI = 27% to 39%), 114 of 230 had new services (49%, 95% CI = 43% to 56%), and 46 of 230 had an unmet need (20%, 95% CI = 15% to 26%). Family members provided new services to 98 of 114 of those patients who had new services (86%, 95% CI = 78% to 92%). Friends provided new services to 28 of 114 of patients (25%, 95% CI = 17% to 34%), and hired help provided new services to 42 of 114 patients (37%, 95% CI = 28% to 46%). The proportions of patients with decline in individual ADLs is shown in Figure 2.
|Outcome||1 Week||4 Weeks|
|Numerator (n = 230 Unless Noted)||Percent (95% CI)||Numerator (n = 219 Unless Noted)||Percent (95% CI)|
|Functional decline||93||40 (34–47)||77||35 (29–42)|
|Unscheduled contact||76||33 (27–40)||123||56 (49–63)|
|ED visit||18||8 (5–12)||32||15 (10–20)|
|Unscheduled doctor visit||52||23 (17–29)||91||42 (35–48)|
|Inpatient rehabilitation||11||5 (2–8)||16||7 (4–12)|
|ECF admission||10||4 (2–8)||14||6 (4–10)|
|Hospital admit||15||7 (4–11)||24||11 (7–16)|
|Nursing||18||8 (5–12)||31||14 (10–19)|
|Physical therapy||18/229*||8 (5–12)||46||21 (16–27)|
|Continuous supervision||33||14 (10–20)||43/218*||20 (15–26)|
|Checking||19/179†||11 (7–16)||25/165†||15 (10–22)|
|Homemaker||50/224*||22 (17–28)||65/214*||30 (24–37)|
|Meals||57/229*||25 (19–31)||70||32 (26–38)|
|Personal care||53||23 (18–29)||59||27 (21–33)|
|Any new service||114||50 (43–56)||141||64 (58–71)|
|Perceived service need||46||20 (15–26)||34||16 (11–21)|
|Nursing||9||4 (2–7)||7||3 (1–6)|
|Physical therapy||35||15 (11–21)||29||13 (9–18)|
|Continuous supervision||0||0 (0–1.3)||2||0.9 (0.1–3)|
|Checking||0||0 (0–1.3)||0||0 (0–1.4)|
|Homemaker||4||2 (0.5–4)||1||0.5 (0.01–3)|
|Meals||1||0.4 (0.01–2)||0||0 (0–1.4)|
|Personal care||4||2 (0.5–4)||2||0.9 (0.1–3)|
At 4 weeks, 29 patients with functional decline had recovered, 17 had new functional decline, and 11 patients were lost to follow-up (four of whom had functional decline at 1 week). This left 77 of 219 with functional decline (35%, 95% CI = 29% to 42%). By 4 weeks, 123 of 219 of patients had an unscheduled medical contact (56%, 95% CI = 49% to 63%), 141 of 219 patients had new services (64%, 95% CI = 58% to 71%), and 34 of 219 had an unmet need (16%, 95% CI = 11% to 21%). Family members provided new services to 114 of 141 of those patients who had new services (81%, 95% CI = 73% to 87%). Friends provided new services to 30 of 141 of patients (21%, 95% CI = 15% to 29%) and hired help provided new services to 71 of 141 (50%, 95% CI = 42% to 59%).
Table 3 shows the univariate analysis of potential predictors of functional decline at 1 week. Female sex, hospitalization within the prior 6 months, nursing home admission within 6 months, dependence in any IADL prior to the injury, upper and lower extremity fractures or dislocations, and trunk injury had p-values of <0.15 and were associated with increased rates of 1-week functional decline; head-injured patients were less likely to have functional decline at 1 week (p < 0.0001).
|Variable||Functional Decline||No Functional Decline||Univariate p-value|
|Age (mean yr)||77.5||76.8||0.434|
|Sex (% female)||78||65||0.028|
|Cognitive impairment (%)||24||19||0.412|
|Fall within 6 months (%)||28||31||0.661|
|ED visit within 6 months (%)||31||31||1.0|
|ED visit within 30 days (%)||9||15||0.219|
|Hospitalization within 6 months (%)||27||17||0.071|
|Hospitalization within 90 days (%)||13||8||0.265|
|SNF admission within 6 months (%)||9||4||0.146|
|≥3 medications (%)||78||72||0.282|
|≥5 medications (%)||51||47||0.688|
|Vision problem (%)||86||90||0.406|
|Living alone (%)||38||35||0.780|
|Comorbidity (CCMS ≥ 1) (%)||44||49||0.503|
|Dependent in any IADL (%)||65||44||0.003|
|Dependent in any PADL (%)||42||40||0.787|
|Upper extremity fracture or dislocation (%)||37||12||<0.0001|
|Lower extremity fracture or dislocation (%)||15||4||0.003|
|Head injury (%)||16||39||0.0001|
|Thoracic injury (%)||31||10||0.091|
The final multivariate model of predictors of 1-week functional decline is shown in Table 4. Immobilization was dropped from the model due to collinearity. The Hosmer and Lemeshow goodness-of-fit test had a chi-square test statistic of 2.89 and a p-value of 0.94, showing a good model fit. The area under the ROC curve for the model was 0.77, showing acceptable discrimination.22 We found no pattern to the deviance residuals, and no cluster of observations that might have had a large effect on the model. The p-value for head injury was 0.051; however, the likelihood ratio test for the model without head injury was significant, suggesting that it should be retained in the model. In addition, we believe that the inclusion of head injury as a “protective” predictor made clinical sense. These were all patients who did not require admission for treatment; therefore, the patients had minor head injuries. The finding that these patients were less likely to have functional decline has face validity.
|Variable||OR (95% CI)|
|Female sex||2.2 (1.1–4.5)|
|Dependent in any IADL||2.5 (1.3–4.8)|
|Upper extremity fracture or dislocation||5.5 (2.5–11.8)|
|Lower extremity fracture or dislocation||4.6 (1.4–15.4)|
|Trunk injury||2.4 (1.1–5.3)|
|Head injury||0.48 (0.23–1.0)|
In this study, we found a significant proportion (40%) of older ED patients with blunt injuries not requiring hospital admission for treatment will have functional decline in the first week after their injury. Although many of these patients have new services initiated, most commonly family and friends provide the services. Additionally, one in five patients perceive the need for additional services, most commonly physical therapy (15%). One-third of patients had an unscheduled medical contact, although only 23% had a visit to an outpatient physician. There is some improvement in function and reduction in perceived service needs by 4 weeks. More patients have new services at 4 weeks, including a more than 2.5-fold increase in physical therapy, and more patients have had a visit with an outpatient physician.
In three-fourths of cases, older patients with falls and other injuries are discharged to home.2 The patient’s ability to function after discharge from the ED may be affected by the underlying condition that caused the fall, the injury itself, or the treatment of the injury (e.g., immobilization of an extremity). These findings highlight the importance of thoughtfulness during the discharge planning for older ED patients with blunt injuries. In the standard emergency care model, EPs focus on the treatment of acute illness and injury. The geriatric emergency care model suggests that a broader, biopsychosocial approach is necessary for older patients.23 As our findings demonstrate that older ED patients with injuries are unlikely to have a visit to an outpatient physician in the week after their ED visit, EPs need to be aware of the high incidence of functional decline and increased service needs, and take this into consideration during discharge planning.
Our intent in using logistic regression modeling to determine independent predictors of functional decline was to identify predictors of functional decline that may be useful for both clinicians and researchers interested in patient outcomes after an acute injury, not to create a clinical prediction rule or to suggest a threshold for enhanced care. For instance, clinicians may consider assessing social supports in those patients with upper or lower extremity fractures who are discharged home. Conversely, researchers studying interventions to reduce functional decline after an acute injury may consider excluding patients with an isolated minor head injury, as they have a lower probability of decline that will make it more difficult to detect a clinically important and statistically significant difference. We used brief screening tools such as the SIS that are practical for ED use and have been studied in this population, even though they may not be as accurate as more complex and lengthy tools.15–17 We evaluated potential predictors that have been associated with adverse outcomes in the general population of older ED patients. 5,6,21,24 However, given the difference in our population, it is not surprising that these factors were not significantly associated with our primary outcome.
Future studies of interventions to prevent or mitigate adverse outcomes in this population are warranted. Several studies have been performed evaluating interventions for improving outcomes of older patients discharged from the ED, although they do not address injury specifically.7–10,21,24–29 Results of these studies have been mixed. McCusker and colleagues25 found a reduction in functional decline at 4 months following the ED visit (although no change in other outcomes),8 which they attributed to the early provision of home care. An Australian study of comprehensive geriatric assessment and interventions did find fewer ED visits and hospital admissions and less decline in physical and mental functioning.26“Quick response” programs, characterized by a strong relationship between the hospital and home care agencies, assessment of patients and coordination of services during the ED visit (and therefore available beyond the traditional Monday through Friday daytime hours only), a rapid provision of in-home services that are more intense than traditional home care, and time-limited services have been described and may be a useful model for an intervention in this population.27
Other studies have evaluated fall prevention programs for patients seen in the ED after a fall.30–38 The goal of these programs has typically been the prevention of future falls, rather than the prevention or mitigation of adverse outcomes related to the index fall. In many of the studies, the interventions began well after the ED visit.32–35 A recent meta-analysis of multifactorial assessment and intervention for falls in older ED patients, including six with ED recruitment, found that no study reported outcomes related to quality of life or physical activity. No clear effect was found on reducing the number of recurrent falls or fall-related injuries.39
Our results add to the existing literature assessing longer-term outcomes of older ED patients with injuries, and the results of these studies also highlight the substantial risk of adverse outcomes in these patients. Grisso and colleagues12 studied older patients treated in one of 11 Philadelphia EDs after a fall. Eight weeks after the fall, 43% of patients reported continued pain or restriction in their usual activities due to the fall. Seven months after the injury, 41% reported continued pain or activity restriction. Only 7% had received home health aide care at 7 months, and 39% had received physical therapy. Only 14% said their physician or other health professionals had been helpful in their recovery. Shapiro and colleagues11 found that, at 3 months after an ED visit for a minor traumatic injury, 7% of patients had a decline in basic ADLs, and 23% had a decline in IADLs. Russell et al.14 assessed older patients seen in the ED for a fall in their homes a median of 20 days after the ED visit. Ten percent of subjects fell between discharge from the ED and the home assessment; the risk of recurrent falls increased with an increasing length of time between the ED visit and the home assessment. Similar rates of functional decline (35%) and risk factors for functional decline (female sex, fractures, depression, functional independence before the fall, and a slower Timed Up and Go test) were found.
Our study was performed in one community. Although the study subjects were typical of our older ED patients, differences in demographics and baseline health between our population and others may affect the generalizability of the findings. We enrolled patients only between 08:00 and 20:00 hours. Although most older patients present to our EDs during these hours, there may be differences in the population that present after hours which might bias our results. We used a validated self-reporting questionnaire rather than performance testing to determine functional decline. Older patients may overestimate their functional abilities,40 and this may affect the overall scores on the OARS ADL questionnaire. However, estimates of functional decline are less likely to be affected as we interviewed the same person at enrollment and at follow-up in nearly all cases. We also relied on self-report for the other outcomes (except repeat visits to our institution), so these numbers could be affected by patient or proxy recall. Our modeling of predictors of functional decline included variables that we thought were potential predictors based on prior research and clinical experience. However, our study was not powered to assess all potential associations. In addition, this is a derived model that has not been independently validated.
Older patients have a significant risk of short-term functional decline and other adverse outcomes after ED visits for injuries not requiring hospitalization. Although many of these patients have new services initiated after their ED visit, there is still a significant need for services perceived by patients. The most significant predictors of functional decline are upper and lower extremity fractures. Future studies of interventions to reduce adverse outcomes in this population are warranted.
The authors thank Dimitre Stepanov, PhD, for his assistance with the logistic regression modeling.
- 1Injury in the United States: 2007 Chartbook. Hyattsville, MD: National Center for Health Statistics, 2008., , , .
- 2Centers for Disease Control and Prevention. WISQARS (Web-based Injury Statistics Query and Reporting System). Available at: http://www.cdc.gov/injury/wisqars/index.html. Accessed Oct 1, 2009.
- 19Multidimensional Functional Assessment of Older Adults: The Duke Older Americans Resources and Services Procedures. Hillsdale, NJ: Lawrence Erlbaum Associates Inc., 1988..
- 22Applied Logistic Regression. 2nd ed. Hoboken, NJ: Wiley-Interscience Publication, 2000., .
- 23Emergency Care of the Elder Person. St. Louis, MO: Beverly Cracom Publications, 1996..
- 26A randomized, controlled trial of comprehensive geriatric assessment and multidisciplinary intervention after discharge of elderly from the emergency department--the DEED II study. J Am Geriatr Soc. 2004; 52:1417–23., , , .
- 33Older people and falls: a randomized controlled trial of a health visitor (HV) intervention. Rev Clin Gerontol. 2001; 11:209–14., , , .