Background and Importance

Patients aged 65 years or older account for more than 15% of all emergency department (ED) visits each year in the United States, and most of these patients are discharged home. Among older persons, disability and functional decline are associated with increased mortality, institutionalization, and costs. The estimated additional cost of medical and long-term care for newly disabled older persons in the United States is $26 billion per year.

Previous work has shown that illnesses and injuries leading to hospitalization are associated with functional decline. Prior studies have also suggested that older patients discharged from the ED may experience some functional decline, but these studies were limited by the absence of suitable comparison groups and by retrospective reports of preillness function.

Goals of This Investigation

The objective of this study was to evaluate the burden of disability during a 6-month period in older persons who were discharged from the ED (ED only) by comparing them with a matched control group (control) and with a group that was admitted to the ED and hospitalized (ED-hospitalized). We hypothesized that older persons who visited the ED and were discharged would experience a greater burden of disability during the following 6 months compared with those who did not visit the ED, but a lower burden of disability compared with those who were hospitalized. Admission to a nursing home and mortality were evaluated as secondary outcomes.

Study Design and Setting

This study is part of the Yale Precipitating Events Project, an ongoing prospective, longitudinal study of 754 initially nondisabled, community-living persons aged 70 years or older. The Precipitating Events Project was designed to elucidate the epidemiology of disability, with the goal of informing the development of effective interventions to maintain and restore independent function. Methods of this longitudinal study have been described in detail elsewhere. Briefly, the cohort was assembled between March 1998 and October 1999 from a computerized list of 3,157 age-eligible members of a large health plan in greater New Haven. Eligibility was determined during a screening telephone interview and was confirmed during an in-home assessment. Of the eligible members, 75.2% agreed to participate in the project, and persons who declined to participate did not significantly differ in age or sex from those who were enrolled. The Yale Human Investigation Committee approved the study protocol, and all participants provided informed consent.

Data Collection and Processing

From 1998 to 2012, participants completed comprehensive, home-based assessments at 18-month intervals and were interviewed monthly by telephone to reassess their functional status, ascertain intervening illnesses and injuries leading to ED visits and hospitalizations, and identify nursing home admissions and deaths. For participants with significant cognitive impairment, a proxy informant was interviewed, using a rigorous protocol with demonstrated reliability and validity, as described elsewhere. During the comprehensive assessments, data were collected on demographic characteristics, chronic conditions, body mass index, cognitive impairment, depressive symptoms, and physical frailty.

Age was measured in years at the time of the index ED visit. Nine self-reported, physician-diagnosed chronic conditions were assessed: hypertension, myocardial infarction, congestive heart failure, stroke, diabetes, fractures, arthritis, chronic lung disease, and cancer (excluding minor skin cancers). Cognitive impairment was defined as a score less than 24 on the Folstein Mini-Mental State Examination. Depressive symptoms were defined as a score greater than or equal to 20 on the Center for Epidemiological Studies–Depression scale. Body mass index was assessed as self-reported weight in kilograms divided by height in meters squared. Physical frailty was defined on the basis of slow gait speed, as previously described.

Complete details about the assessment of disability are provided elsewhere. Briefly, during the monthly interviews, participants were asked, “At the present time, do you need help from another person to [complete the task]?” for each of the 4 basic activities (bathing, dressing, walking across a room, and transferring from a chair), 5 instrumental activities (shopping, housework, meal preparation, receiving medications, and managing finances), and 3 mobility activities (walk one quarter mile [0.40 km], climb a flight of stairs, and lift and carry 10 pounds [4.5 kg]). For each of these 12 activities, disability was defined as the need for personal assistance or being unable to perform the activity. Participants were also asked about a fourth mobility activity, “Have you driven a car during the past month?” Participants who responded no were deemed to have stopped driving. To maintain consistency with the other activities, these participants were classified as being “disabled” in driving that month. The number of disabilities overall and for each functional domain (basic, instrumental, and mobility) was summed.

The primary source of information on ED visits and hospitalizations was linked Medicare claims data, which were available for nearly all hospitalizations and for ED visits among fee-for-service participants. Of the ED-only observations and ED-hospitalized observations, 605 (74.4%) and 619 (98.1%) were identified from the Medicare claims, respectively. For participants in managed Medicare, information on ED visits and some hospitalizations (ie, those without a Medicare claim) was obtained during the monthly interviews. Participants were asked whether they had visited an ED or stayed at least overnight in a hospital since the last interview. Among a subgroup of 191 participants, we found that the accuracy of self-reported ED visits, compared with Medicare claims data, was high (ĸ=0.80; 95% confidence interval [CI] 0.78 to 0.82). The raw agreement was 98.4%. The accuracy of self-reported hospitalization, based on an independent review of hospital records, was also high (ĸ=0.92; 95% CI 0.90 to 0.95). The raw agreement was 98.8%.

Participants who reported visiting the ED were asked the primary reason for their visit. These self-reports were complemented by Medicare claims data when needed. The list of the visit reasons was independently reviewed by 2 physicians (J.M.N. and W.F.), and the reasons were grouped into distinct diagnostic categories with a revised version of a previously published protocol, as shown in Table 1 .

Participants or their proxies were also asked whether they had been admitted to a nursing home during the past month. The accuracy of these reports was high when compared with Medicare claims data, with a sensitivity of 96% and specificity of 100%. Deaths were ascertained by review of the local obituaries or from a proxy during a subsequent telephone interview, with a completion rate of 100%.

To test our hypotheses, we used a matched cohort design. This design reduces bias with little loss of precision and permits the use of generalized estimating equations, which accounts for the correlation of observations within each cluster of matched observations. We compared the course of disability during 6 months among 3 groups: those who had an ED visit but were not hospitalized (ED only), a matched group who did not visit an ED (control), and an unmatched group who were hospitalized after an ED visit (ED-hospitalized). The pool of potential observations was not sufficiently large to permit concurrent matching for the ED-hospitalized group in addition to the control group.

Participants were included only if they had been living in the community immediately before their ED visit (or corresponding point for the control group). To make full use of our longitudinal data, participants were allowed to contribute more than one qualifying ED-only or ED-hospitalized event, but only the first event was included from a specific 18-month interval, which was the period between the comprehensive assessments. This combination of a participant and his or her event within an 18-month interval defined an “observation” and was our unit of analysis. Similarly, only one control observation per participant was permitted within a specific 18-month interval.

To assemble the control group, we used a SAS (version 9.4; SAS Institute, Inc., Cary, NC) macro to sequentially match each ED-only observation with an unexposed, or control, observation on the following 4 features: sex, participant age (±4 years) at the ED visit, number of disabilities (±1) out of the 13 possible in the month before the ED visit, and number of months since the previous comprehensive assessment. The analytic sample included 813 observations (from 430 participants) in the ED-only group, 813 observations (from 442 participants) in the control group, and 631 observations (from 390 participants) in the ED-hospitalized group. One hundred twenty-one participants had observations in both the ED-only and control groups, 97 in both the ED-only and ED-hospitalized groups, and 139 in all 3 groups.

Outcome Measures

The primary outcome was the number of disabilities in the 13 basic, instrumental, and mobility activities during the 6 months after the ED visit, hospitalization, or corresponding time point for the control group. Hereafter, we refer to this time point as the index month. To determine whether our findings were consistent across these 3 functional domains, we also evaluated the number of disabilities in the 4 basic, 5 instrumental, and 4 mobility activities. Although there is some debate about whether these activities should be considered on an ordinal or interval scale, we have chosen to analyze them on an interval scale as in a previous study. We chose the 6-month period because it has previously been used to evaluate disability after hospitalization. As secondary outcomes, we evaluated nursing home admissions and deaths during the 6-month period after the index month.

Primary Data Analysis

The reasons for the ED visits were tabulated. Demographic and clinical characteristics were summarized by means with SDs and frequencies with proportions for the ED-only group, the matched control group, and the unmatched ED-hospitalized group.

Because each outcome represents a count, we fit Poisson generalized estimating equation models to evaluate associations between the 2 primary comparison groups (ED only and control) and disability scores during the 6 months of follow-up. The generalized estimating equation Poisson models were adjusted for the 4 matching criteria, race, education, living situation (alone versus with others), number of chronic conditions, body mass index, cognitive impairment (Mini-Mental State Examination score <24), depressive symptoms (Center for Epidemiological Studies–Depression scale score ≥20), physical frailty, and calendar year. In accordance with previous recommendations, the model adjusts for matching variables to address residual confounding and censoring. A compound symmetry covariance structure accounted for correlations among multiple intervals from the same participant and matching between the ED-only and control groups. These models yielded adjusted relative risks, which denote the increase in disability burden during the 6-month follow-up period for the ED-only group relative to the control group. These analyses were repeated for comparisons between the ED-only and unmatched ED-hospitalized groups. All models were checked for fit with the quasi-information criterion.

For the secondary outcomes, we plotted the percentage of observations with a nursing home stay and the percentage of reported deaths in the 6 months after the index month. We then used multivariable Cox regression models to assess the independent associations between exposure to ED only (versus matched control) and ED-hospitalized (versus ED only) and time to nursing home admissions and death, respectively, while adjusting for the previously described set of covariates. The correlation among multiple intervals from the same participant for the nursing home outcome was accounted for as a cluster, whereas matching between the ED-only and control groups was accounted for using a conditional model.

All analyses were performed with SAS (version 9.4), and P <.05 (2-tailed) was used to indicate statistical significance.

Study Design and Setting

This study is part of the Yale Precipitating Events Project, an ongoing prospective, longitudinal study of 754 initially nondisabled, community-living persons aged 70 years or older. The Precipitating Events Project was designed to elucidate the epidemiology of disability, with the goal of informing the development of effective interventions to maintain and restore independent function. Methods of this longitudinal study have been described in detail elsewhere. Briefly, the cohort was assembled between March 1998 and October 1999 from a computerized list of 3,157 age-eligible members of a large health plan in greater New Haven. Eligibility was determined during a screening telephone interview and was confirmed during an in-home assessment. Of the eligible members, 75.2% agreed to participate in the project, and persons who declined to participate did not significantly differ in age or sex from those who were enrolled. The Yale Human Investigation Committee approved the study protocol, and all participants provided informed consent.

Data Collection and Processing

From 1998 to 2012, participants completed comprehensive, home-based assessments at 18-month intervals and were interviewed monthly by telephone to reassess their functional status, ascertain intervening illnesses and injuries leading to ED visits and hospitalizations, and identify nursing home admissions and deaths. For participants with significant cognitive impairment, a proxy informant was interviewed, using a rigorous protocol with demonstrated reliability and validity, as described elsewhere. During the comprehensive assessments, data were collected on demographic characteristics, chronic conditions, body mass index, cognitive impairment, depressive symptoms, and physical frailty.

Age was measured in years at the time of the index ED visit. Nine self-reported, physician-diagnosed chronic conditions were assessed: hypertension, myocardial infarction, congestive heart failure, stroke, diabetes, fractures, arthritis, chronic lung disease, and cancer (excluding minor skin cancers). Cognitive impairment was defined as a score less than 24 on the Folstein Mini-Mental State Examination. Depressive symptoms were defined as a score greater than or equal to 20 on the Center for Epidemiological Studies–Depression scale. Body mass index was assessed as self-reported weight in kilograms divided by height in meters squared. Physical frailty was defined on the basis of slow gait speed, as previously described.

Complete details about the assessment of disability are provided elsewhere. Briefly, during the monthly interviews, participants were asked, “At the present time, do you need help from another person to [complete the task]?” for each of the 4 basic activities (bathing, dressing, walking across a room, and transferring from a chair), 5 instrumental activities (shopping, housework, meal preparation, receiving medications, and managing finances), and 3 mobility activities (walk one quarter mile [0.40 km], climb a flight of stairs, and lift and carry 10 pounds [4.5 kg]). For each of these 12 activities, disability was defined as the need for personal assistance or being unable to perform the activity. Participants were also asked about a fourth mobility activity, “Have you driven a car during the past month?” Participants who responded no were deemed to have stopped driving. To maintain consistency with the other activities, these participants were classified as being “disabled” in driving that month. The number of disabilities overall and for each functional domain (basic, instrumental, and mobility) was summed.

The primary source of information on ED visits and hospitalizations was linked Medicare claims data, which were available for nearly all hospitalizations and for ED visits among fee-for-service participants. Of the ED-only observations and ED-hospitalized observations, 605 (74.4%) and 619 (98.1%) were identified from the Medicare claims, respectively. For participants in managed Medicare, information on ED visits and some hospitalizations (ie, those without a Medicare claim) was obtained during the monthly interviews. Participants were asked whether they had visited an ED or stayed at least overnight in a hospital since the last interview. Among a subgroup of 191 participants, we found that the accuracy of self-reported ED visits, compared with Medicare claims data, was high (ĸ=0.80; 95% confidence interval [CI] 0.78 to 0.82). The raw agreement was 98.4%. The accuracy of self-reported hospitalization, based on an independent review of hospital records, was also high (ĸ=0.92; 95% CI 0.90 to 0.95). The raw agreement was 98.8%.

Participants who reported visiting the ED were asked the primary reason for their visit. These self-reports were complemented by Medicare claims data when needed. The list of the visit reasons was independently reviewed by 2 physicians (J.M.N. and W.F.), and the reasons were grouped into distinct diagnostic categories with a revised version of a previously published protocol, as shown in Table 1 .

Table 1

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

Adolescent Male With Right Shoulder Pain After Football Injury Association Between the Opening of Retail Clinics and Low-Acuity Emergency Department Visits Bending the Curve of Health Trajectories for Older Adults Discharged From the Emergency Department Yet Another Sedating the Agitated Patient: A Moving Target? Classified 2017 Advertising Rates & Information