The ED as the primary source of hospital admission for older (but not younger) adults

a b s t r a c t

Introduction: The elderly population in the United States is growing. This age shift has important implications for emergency departments (EDs), which currently account for more than 50% of Inpatient hospitalizations. Our ob- jective was to compare the percentage of inpatient admissions starting in the ED between elderly and younger patients.

Methods: We conducted a retrospective analysis using the National Hospital Discharge Survey. Source of admis- sion to the hospital was evaluated for years 2003 to 2009. total admissions from the ED and trends over time were analyzed for the following age groups: 22 to 64, 65 to 74, 75 to 84, and 85+ years old. Likelihood of having been admitted from the ED was evaluated with logistic regression.

Results: A total of 1.7 million survey visits representing 216 million adult hospitalizations were analyzed. A total of 93 million (43.2%) were among patients 65 years and older. The ED was the source of admission for 57.3% of patients 65 years and older and 44.4% of patients 64 years and younger (95% confidence interval difference, 12.97%-13.00%). By 2009, more than 75% of nonelective admissions for patients 85 years and older were through the ED. There was a linear relationship between age and the ED as the source of admission, the odds increasing by 2.9% per year (95% confidence interval, 1.029-1.029) for each year beyond age 65 years.

Conclusion: Emergency departments are increasingly used as the gateway for hospital admission for older adults. An aging US population may increase the effect of this trend, a prospect that should be planned for. From the pa- tient perspective, barriers to care contributing to the age-based discrepancy in the use of the ED as source of ad- mission should be investigated.

Introduction

The role of emergency departments (EDs) within the US health care sys- tem is changing. More than half of patients admitted to the hospital in the United States now start their hospital stay in the ED. Between 2003 and 2009, hospital admissions originating in the ED increased by 17%, whereas admissions from physicians’ offices and clinics decreased by 10% [1].

Although the extent to which a patient’s age is a factor in the growing

use of the ED as a gateway to hospitalization has not been studied, there are several reasons to think that older patients may be disproportionately affected. The ED is a location where specialty consultation and advanced diagnostic technology are available at all hours. Given the medical com- plexity of older adults combined with the diagnostic and time constraints of primary care practices, older patients in need of these acute services are

? Sources of support: None.

?? Presentation: Preliminary data presented at Society of Academic Emergency Medicine Annual Meeting, May 2014.

* Corresponding author at: Emergency Medicine Residency, New York Presbyterian Hospital, 565 East 68th St, Box 301, New York, NY 10065. Tel.:+1 347 400 5844.

E-mail addresses: [email protected] (P.W. Greenwald), [email protected] (R.M. Estevez), [email protected] (S. Clark), [email protected]

(M.E. Stern), [email protected] (T. Rosen), [email protected] (N. Flomenbaum).

more likely to be referred to the ED [2-4]. In addition, patients themselves may be aware of the increased resources available in the ED or have expe- rienced their physician’s limitations, causing them to self-select the ED as their source of care. Once in the ED, older patients are more likely to be admitted than younger patients [5,6].

As is now well documented, there is a demographic change taking place in the United States; the population of older adults is growing. During the 2010 Census, more people were older than 65 years than at any previoUS time [7]. Between 2000 and 2010, the population 65 years and older increased at a faster rate (15.1%) than the total US pop- ulation (9.7%) [8]. The total number of adults 65 and older in the United States is projected to double to 70 million between 2000 and 2030, at which point older adults will comprise 20% of the US population. Cur- rently, the fastest growing older age group is people 85 years and older [2,9,10]. Older adults account for approximately 15% of annual US ED visits [10,11]. Based on current visit rates, this number is projected to increase to 25% by 2030 [11]. If older patients are more like- ly than younger patients to have their inpatient stays start in the ED, then these demographic changes will accelerate the growing use of the ED as the principal gateway to hospitalization

We used the 2003-2009 National Hospital Discharge Survey (NHDS) to investigate the relationship between increasing age and the

http://dx.doi.org/10.1016/j.ajem.2015.05.041

944 P.W. Greenwald et al. / American Journal of Emergency Medicine 34 (2016) 943–947

likelihood that hospitalization begins in the ED. Our analysis paralleled the methodology used by the RAND Corporation in their 2013 report “The Evolving Role of Emergency Departments in the United States,” with the addition of stratification by age.

Methods

We conducted an analysis of data from the NHDS, an annual survey conducted by the National Center for Health Statistics at the Centers for Disease Control and Prevention. The NHDS is a complex, stratified, multi- stage probability design survey that draws on hospital discharge records from more than 250000 patients at short-stay nonfederal hospitals. The data sets and probability design are available for public use (http:// www.cdc.gov/nchs/about/major/hdasd/nhdsdes.htm; accessed May 9, 2013). Inclusion eligibility is restricted to hospitals having 6 or more beds and an average length of stay for all patients less than 30 days [12]. The NHDS provides detailed information on patient demographics, type of admission, source of admission, geographic region, expected source of payment, hospital size, and type of hospital ownership [12].

“Observation” status hospital stays are not included in the NHDS.

The NHDS data set includes “source of hospital admission” data starting in 2001 and ending in 2010. To allow for comparisons between our analysis and RAND’s 2013 report, we evaluated data from 2003 to 2009. RAND limited their report to these years because of high rates of missing data in the source of admission field for years 2001 and 2002. Data for 2010 were not included by RAND because the coding guidelines changed, making comparison to earlier years unreliable [1].

The NHDS lists the following 10 categories for source of admission:

(1) ED, (2) physician referral, (3) clinical referral, (4) HMO referral,

(5) transfer from a hospital, (6) transfer from a skilled nursing facility,

(7) transfer from other health facility, (8) court/law enforcement,

All analyses were conducted using Stata version 12.0 (StataCorp, LP, College Station, TX).

The study was reviewed by the Weill Cornell Medical College Institu- tional Review Board and classified as exempt.

Results

Between 2003 and 2009, there were 1.7 million NHDS survey visits, representing 216 million adult hospitalizations. Of the 216 million hospital- izations, 93 million (43.2%) were among patients 65 years and older. Of all hospitalizations, 60.1% were for women. The mean age of patients at the time of admission was 58 years old. Nonelective admissions were 73.4% of total admissions. Slightly more than 50% of admissions were from the ED. The ED was the source of admission for 57.3% of patients 65 years and older (95% confidence interval [CI], 57.3%-57.4%) and for 44.4% of patients younger than 65 years (95% CI, 44.4%-44.4%), with a difference of 12.99% (95% CI difference, 12.97%-13.00%). For patients age 65 to 74 years, 51.8% of admissions started in the ED (95% CI, 51.8%-51.9%); for patients age 75 to 84 years, 58.1% of admissions started in the ED (95% CI, 58.1%-58.2%); and for patients 85 years and older, 64.9% of their hos-

pitalizations began in the ED (95% CI, 64.9%-64.9%).

There were differences in percentage of admissions starting in the ED by sex, region, hospital size, hospital type, race, and patient insur- ance type. These differences are summarized in Table 1.

The plot of percentage of admission from the ED by age demonstrates age-based differences in the use of the ED as the source of

Table 1

NHDS admissions and percentage of those admissions coming from the ED: by sex, race, primary payer type, hospital size, hospital ownership, and region

(9) other, and (10) not available [12]. Our analysis compared the ED as the source of hospital admission to all other sources of admission. The NHDS also codes the acuity of admission as emergency, urgent, elective,

Patient sex

No. of admissions (% total)

Percentage of admissions from ED (95% CI)

and not available. We considered emergency and urgent admissions as

“non-elective” admissions and compared them to elective admissions. Non- Female elective admissions are defined by their inability to be postponed and are Patient race dictated by the patient’s medical condition and their treating physician’s de- White	129933308 (60.1) 131917987 (61.1)	45.5 (45.5-45.5) 49.3 (49.3-49.3)
termination that hospitalization is required to address the problem. Elective ^{Black/African} ^American	25902002 (12.0)	59.0 (58.9-60.0)
American Indian/Alaskan Native	764633 (0.4)	46.1 (46.0-46.0)
admissions are not urgent and are chosen by the patient or their physician _Asian	3251301 (1.5)	38.6 (38.6-38.7)
for reasons that are perceived to be beneficial to the patient [1]. Native Hawaiian or other Pacific	343530 (0.2)	45.4 (45.2-45.5)
The proportion of missing data for the category “source of admis- Other	4286380 (2.0)	52.2 (52.2-52.3)
sion” decreased steadily from 12% in 2003 to 4% in 2009. There were ^Multiple ^race	246661 (0.1)	51.7 (51.5-51.9)
also changes in missing data for “admission type” by year. To address Race not stated	49329068 (22.8)	47.8 (47.8-47.8)

Male 86108254 (39.9) 56.8 (56.8-56.8)

Hospital region

missing data for both source of admission and hospital admission	Northeast	46815827 (21.7)	57.7 (57.6-57.7)
type, we used the same methodology described by RAND: multivariable	Midwest	49944954 (23.1)	48.6 (48.6-48.6)

imputation using year, region, sex, and age group was used when source of admission or “hospital admission type” were coded as “not available”

South	80078411 (37.1)	49.8 (49.8-49.8)
West	39202370 (18.1)	42.9 (42.8-42.9)
Hospital bed size

[1]. Multivariable imputation allows estimation missing values for	6-99	47933427 (22.2)	48.6 (48.5-48.6)
source of admission and hospital admission type within strata of year,	100-199	46507930 (21.5)	53.3 (53.3-43.4)
region, sex, and age group. Results of analysis of unimputed data and de-	200-299	45196783 (20.9)	51.5 (51.4-41.5)
tails regarding distribution of not available data are provided in the ap- pendix. Our analysis included admissions for adults 22 years and older.	300-499 500+ Hospital ownership	48224016 (22.3) 28179406 (13.0)	52.7 (52.7-52.7) 47.3 (47.3-47.3)
We evaluated age both as a linear variable and by grouping patients as	Government hospital	26888809 (12.4)	51.4 (51.3-51.4)
younger (age b 65 years) vs older (age >= 65 years). In addition, we cate-	Proprietary hospital	24564504 (11.4)	38.8 (38.8-38.8)
gorized age into 4 bins: age 22 to 64 years, age 65 to 74 years, age 75 to	Nonprofit hospital	164588249 (76.2)	51.4 (51.4-51.4)
84 years, and age >= 85 years. Percentage of admissions starting in the ED was calculated for grouped data. The relationship of age as a linear pre-	Primary payer Worker’s compensation Medicare	1088400 (0.5) 97033222 (44.9)	39.8 (39.7-39.8) 57.1 (57.1-57.1)
dictor for source of hospital admission was evaluated with logistic re-	Medicaid	27183916 (12.6)	45.5 (45.4-45.5)
gression. Models were evaluated with and without adjustment for the	Other government	3214053 (1.5)	47.0 (47.0-47.1)
effect of sex, ethnicity, Insurance type, hospital size, hospital ownership, and geographic region. Additional models were constructed that were	BlueCross/BS HMO/PPO Other private insurance	19000761 (8.8) 31092180 (14.4) 19157928 (8.9)	39.0 (39.0-39.1) 39.9 (39.3-39.3) 41.6 (41.6-41.7)
limited to patients older than 64 years based on the presence of a strong	Self-pay	9943651 (4.6)	66.8 (66.8-66.8)
linear relationship between age and source of admission for those 65	No charge	437675 (0.2)	62.1 (61.9-62.2)
years and older. Trends over time were evaluated by calculating per- centage of patients admitted from the ED by year and age group.	Other Payer not stated	3964798 (1.8) 3924978 (1.8)	43.9 (43.8-43.9) 51.2 (51.2-51.3)

P.W. Greenwald et al. / American Journal of Emergency Medicine 34 (2016) 943–947 945

Figure.

hospitalization. (Figure) Adjusting for geographic region, race, sex, pa- tient insurance type, hospital ownership, and hospital size, the odds of admission with each additional year of age was 1.021 (95% CI, 1.021- 1.021). Based on the relationship demonstrated in the graph, a separate analysis was conducted to evaluate the relationship between age and the ED as the source of admission for patients 65 years and older. The odds of the ED being the source of hospital admission limited to evaluation of pa- tients 65 years and older after adjustment for geographic region, race, sex, patient insurance type, hospital ownership, and hospital size in- creased by 1.029 or 2.9% increase for each additional year of age (95% CI, 1.029-1.029). Both analyses were similar before and after adjustment. The percentage of admissions starting in the ED increased over time for all age groups (Table 2). Patients 65 years and older were more likely to be admitted from the ED than were younger patients for every year studied. Increases in the percentage of admissions starting in the ED were primarily associated with nonelective admissions. Only 7.5% of elective admissions began in the ED, and there was no clear change in this number over time. The ED was the source of admission for patients with nonelective hospitalizations 63% of time in 2002 and increased each year, becoming the source for 71% of nonelective admissions by 2009. Older patients were more likely to be admitted through the ED for each year studied. By 2009, 78% of nonelective admissions for pa- tients 85 years and older started in the ED. Analysis of data without im-

putation demonstrates similar trends and is available in the appendix.

Discussion

Patients 65 years and older were more likely than younger patients to have their hospitalizations start in the ED. Emergency departments are the entry point for more than half of all hospital admissions, but only 44% of admitted patients younger than 65 years had the ED as the source of their admission as compared to 57% of patients 65 years and older. The percentage of inpatient stays that started in the ED increased consistent- ly for each year of older than 65 years, with each additional year of age

Table 2

Percentage of hospitalizations starting in the ED by age and calendar year

Percentage of nonelective admissions from ED

Age (y)	2003	2004	2005	2006	2007	2008	2009	Total
22-65	56.8	57.8	58.5	58.3	59.8	64.7	66.1	60.4
65-74	67.0	66.7	66.7	67.4	68.5	72.3	74.1	69.1
75-84	69.0	66.7	66.7	67.4	68.5	75.4	76.0	71.7
N 84	72.3	66.1	69.3	74.2	73.7	77.5	78.2	73.2
Total	62.6	62.2	63.2	63.8	64.9	69.4	70.5	65.3
Percentage of elective admissions from ED
22-65	7.4	9.2	9.3	6.5	5.7	4.4	5.1	6.7
65-74	10.2	10.7	9.9	8.3	7.2	5.8	5.3	8.1
75-84	13.3	11.1	10.4	10.2	7.9	4.6	7.5	9.3
N 84	15.1	8.1	7.2	11.9	11.2	6.0	9.6	9.9
Total	8.1	8.3	8.2	8.0	8.2	8.0	8.6	7.5

having a 2.9% increase in the likelihood that hospitalization started in the ED. For the oldest patients, those older than 85 years, the ED was the gateway to hospitalization approximately two-thirds of the time.

The trend to increased use of the ED as the gateway to hospitalization may be due to several factors. primary care providers (PCPs) might be more likely to direct elderly patients to the ED rather than seeing them in the office. Primary care providers are increasingly under pressure to see a larger volume of patients, making it difficult for patients to arrange an acute care visit [3,4]. Cognizant that relatively mild complaints can rep- resent significant illness in the elderly, a PCP may be more likely to refer their elderly patient to the ED rather than suggest that they wait for the next available appointment [4]. This trend is probably more pronounced in the Off hours when the on-call provider might not know the patient di- rectly. In addition, if a provider does see an older patient in the office, if the provider feels that the patient requires advanced diagnostic modalities, specialty consultation, or social service evaluation, they may be referred to the ED [2,13]. Older patients may, therefore, be more likely to be sent to the ED for further evaluation, including evaluation regarding the need for acute hospitalization [10]. Once an older patient arrives in the ED, the same scheduling pressures that may have led to the ED evaluation in the first place may make the ED provider more likely to admit an older patient to the hospital, even if it is only for a brief stay [14]. Lack of availability of timely follow-up with either the PCP or specialist consultant is one of the main reasons emergency physicians err on the side of admission to the hospital vs discharge from the ED [1,15].

Traditionally, EDs have been defined by their role of providing early diagnosis and treatment of life-threatening illness and injury. However, EDs have increasingly become the site of care for the uninsured, the lo- cation for expedited medical evaluation for patients who cannot get timely access to primary or specialty care, a place where complex diag- nostic evaluations can occur at all hours, and the coordination point for patients who need urgent but not necessarily emergency admission [16]. Elderly patients have needs that pose diagnostic and treatment challenges due to the complex interplay of comorbidities, atypical pre- sentations, polypharmacy, and the possibility of further complication caused by cognitive or functional impairment. The medical manage- ment of older patients is more time and resource intensive than for younger patients and more often requires a multidisciplinary approach. Interventions have been proposed and are beginning to be implement- ed to improve the care of older adults in the ED [17-19]. These include education in Geriatric EMergency medicine core competencies for all ED providers (physicians, midlevel providers, and nurses), specific geri- atric clinical protocols for common geriatric syndromes (eg, falls, delir- ium, and polypharmacy), and ED physical space modification [17].

Our study has several limitations. The results of our study are limited by the quality of data obtained by the NHDS and the ability of its design to be a representative national sample. In addition, the NHDS data set ex- cludes Observation status admissions, despite the increasing use of obser- vation hospital stays since their initial introduction in 2002. The omission of observation status admissions can lead to an underestimation of overall admissions, and it is possible that these observation status admissions have different trends, with respect to age and source of observation stay, than the admissions represented in the NHDS survey. It would be reasonable to conclude that the admissions captured by the NHDS are more complex, longer admissions than observation status admissions, which are not captured. Despite the absence of observation status admis- sions, we believe that the NHDS data demonstrate a disproportionate use of the ED as the gateway to admission for older patients when compared to younger patients. We feel that this has special importance in the con- text of the aging of the US population. Given the focus of the Affordable Care Act on efficient use of inpatient resources, attention must be turned to ED operations and the increasingly important role that the ED plays in facilitating needed admissions and avoiding those that are preventable, especially as it relates to the Geriatric population.

Further research is needed to understand if age-based differences in ED use for hospital access represent barriers to care or if, alternatively, it

946

P.W. Greenwald et al. / American Journal of Emergency Medicine 34 (2016) 943–947

Appendix A

Table A.1

Unimputed data. Number and percentage of NHDS hospitalizations for which the emergency department was the source of admission (2003-2009) by age group and admission type^a

	2003	2004	2005	2006	2007	2008	2009	Total
Elective admissions
22-64	95433 (2.06%)	125145 (2.67%)	105558 (2.32%)	109269 (2.31%)	95614 (1.95%)	85488 (1.79%)	120873 (2.37%)	737380 (2.21%)
65-74	23654 (2.02%)	27870 (2.29%)	37034 (2.85%)	23741 (1.97%)	18936 (1.48%)	21790 (1.64%)	37642 (2.54%)	190667 (2.12%)
75-84	35796 (3.47%)	34745 (3.16%)	35403 (3.23%)	32490 (3.07%)	20251 (1.85%)	19771 (1.86%)	47095 (4.07%)	225551 (2.97%)
N 84	24945 (7.14%)	21632 (5.84%)	22211 (5.55%)	19099 (4.78%)	16560 (4.03%)	12090 (3.26%)	23871 (6.15%)	140408 (5.22%)
Total	179828 (2.50%)	209392 (2.84%)	200206 (2.72%)	184599 (2.50%)	151361 (1.97%)	139139 (1.85%)	229481 (2.82%)	1294006 (2.46%)
Total admissions	7192593	7371566	7354004	7398137	7686331	7534641	8128945	52666217
Nonelective admissions
22-64	5970089 (59.37%)	6074874 (58.42%)	6071949 (57.10%)	6457986 (58.36%)	6669298 (59.95%)	7909521 (67.09%)	8214349 (68.15%)	47368066 (61.42%)
65-74	2077398 (67.48%)	2029556 (66.53%)	2061918 (64.97%)	2110666 (66.01%)	2107210 (67.49%)	2583298 (73.42%)	2711625 (74.55%)	15681671 (68.85%)
75-84	2587512 (69.36%)	2593463 (67.15%)	2622409 (67.79%)	2605519 (68.53%)	2683327 (70.38%)	3193760 (76.14%)	3055886 (76.34%)	19341876 (70.92%)
N 84	1553822 (70.89%)	1506539 (68.52%)	1657266 (71.05%)	1713180 (71.02%)	1721067 (71.41%)	2197562 (77.19%)	2166929 (77.82%)	12516365 (72.87%)
Total	12188821 (63.96%)	12204432 (62.56%)	12413542 (62.04%)	12887351 (62.93%)	13180902 (64.39%)	15884141 (71.07%)	16148789 (71.84%)	94907978 (65.75%)
Total admissions	19056808	19508872	20007659	20478442	20469835	22349662	22478501	144349779
Admission type not available
22-64	200385 (8.80%)	178432 (8.31%)	164638 (8.47%)	237699 (13.40%)	197361 (14.16%)	213535 (15.72%)	206671 (16.59%)	1398721 (11.52%)
65-74	78050 (12.76%)	56041 (10.76%)	37409 (8.77%)	62622 (16.03%)	46539 (14.70%)	48353 (19.35%)	40733 (21.33%)	369747 (13.66%)
75-84	110353 (15.29%)	75048 (13.02%)	43433 (10.13%)	61714 (15.77%)	47168 (16.77%)	39214 (16.98%)	32692 (17.27%)	409622 (14.53%)
N 84	52412 (15.92%)	37141 (13.40%)	23470 (11.62%)	44208 (20.72%)	31005 (23.39%)	30859 (30.01%)	22122 (21.80%)	241217 (17.76%)
Total	441200 (11.20%)	346662 (9.85%)	268950 (8.96%)	406243 (14.67%)	322073 (15.16%)	331961 (17.09%)	302218 (17.49%)	2419307 (12.72%)
Total admissions	3940257	3521003	3001297	2768875	2124346	1942328	1727460	19025566
All admission types
22-64	6265907 (36.92%)	6378451 (37.02%)	6342145 (37.01%)	6804954 (38.72%)	6962273 (39.97%)	8208544 (45.79%)	8541893 (46.42%)	49504167 (40.36%)
65-74	2179102 (44.83%)	2113467 (44.16%)	2136361 (43.60%)	2197029 (45.84%)	2172685 (46.02%)	2653441 (52.09%)	2790000 (52.52%)	16242085 (47.12%)
75-84	2733661 (49.84%)	2703256 (48.81%)	2701245 (50.08%)	2699723 (51.40%)	2750746 (53.03%)	3252745 (59.27%)	3135673 (58.63%)	19977049 (53.00%)
N 84	1631179 (56.82%)	1565312 (55.00%)	1702947 (58.04%)	1776487 (58.72%)	1768632 (59.87%)	2240511 (67.48%)	2212922 (67.59%)	12897990 (60.77%)
Total	12809849 (42.43%)	12760486 (41.97%)	12882698 (42.43%)	13478193 (43.98%)	13654336 (45.09%)	16355241 (51.39%)	16680488 (51.59%)	98621291 (45.65%)
Total admissions	30189658	30401441	30362960	30645454	30280512	31826631	32334906	216041562

^a The denominator for percentage of total admissions is all admissions within same age group and admission type.

P.W. Greenwald et al. / American Journal of Emergency Medicine 34 (2016) 943–947 947

Table A.2

Unimputed data. Percentage of NHDS hospitalizations for which source of admission was not available (2003-2009) by age group

	2003	2004	2005	2006	2007	2008	2009	Total
Elective admissions 22-64	5.05%	6.00%	7.04%	5.37%	5.13%	3.18%	2.50%	4.85%
65-74	4.98%	6.64%	6.14%	5.01%	5.37%	4.12%	3.00%	4.98%
75-84	4.79%	5.51%	6.57%	5.89%	5.64%	2.94%	2.30%	4.79%
N 84	3.79%	4.63%	4.99%	5.69%	5.13%	2.42%	2.32%	4.17%
Total	4.94%	5.96%	6.70%	5.40%	5.24%	3.28%	2.55%	4.83%
Nonelective admissions 22-64	3.58%	4.89%	7.91%	6.77%	5.84%	1.49%	0.85%	4.39%
65-74	2.87%	4.46%	5.72%	5.02%	4.66%	1.14%	0.86%	3.44%
75-84	2.82%	4.85%	5.65%	5.40%	4.26%	1.19%	0.77%	3.52%
N 84	2.94%	4.82%	5.25%	5.05%	4.38%	1.37%	1.01%	3.42%
Total	3.24%	4.80%	6.81%	6.04%	5.20%	1.36%	0.85%	3.96%
Admission type not available 22-64	70.47%	69.25%	69.76%	61.41%	56.51%	28.12%	35.39%	58.88%
65-74	68.81%	70.85%	73.40%	64.82%	60.12%	34.28%	44.38%	63.42%
75-84	65.14%	70.01%	71.52%	70.30%	63.50%	42.03%	53.35%	64.98%
N 84	70.32%	74.04%	75.87%	66.70%	59.61%	37.07%	60.11%	67.01%
Total All admission types	69.23%	69.99%	70.94%	63.56%	58.17%	31.04%	39.80%	61.01%
22-64	12.95%	13.21%	14.69%	11.91%	9.70%	3.96%	3.64%	9.91%
65-74	11.68%	12.23%	11.72%	9.90%	8.57%	3.54%	3.02%	8.55%
75-84	11.39%	11.77%	11.08%	10.34%	7.76%	3.25%	2.96%	8.37%
N 84	10.78%	11.53%	10.07%	9.48%	6.97%	2.59%	3.00%	7.58%
Total	12.26%	12.64%	13.13%	11.08%	8.92%	3.63%	3.36%	9.20%

Table A.3

Unimputed data. Percentage of NHDS hospitalizations starting in the ED by age and calendar year^a Percentage of nonelective admissions from ED

	Age (y)	2003	2004	2005	2006	2007	2008	2009	Total
	22-64	61.6	61.4	62.0	62.6	63.7	68.1	68.7	64.2
	65-74	69.5	69.6	68.9	69.5	70.8	74.3	75.2	71.3
	75-84	71.4	70.6	71.9	72.4	73.5	77.1	76.9	73.5
	N 84	73.0	72.0	75.0	74.8	74.7	78.3	78.6	75.5
	Total	66.1	65.7	66.6	67.0	67.9	72.1	72.5	68.5
Percentage of elective admissions from ED
22-64		2.2	2.8	2.5	2.4	2.1	1.9	2.4	2.3
65-74		2.1	2.5	3.0	2.1	1.6	1.7	2.6	2.2
75-84		3.6	3.3	3.5	3.3	2.0	1.9	4.2	3.1
N 84		7.4	6.1	5.9	5.1	4.2	3.3	6.3	5.5
Total		2.6	3.0	2.9	2.6	2.1	1.9	2.9	2.6

^a Observations where data are not available regarding source of admission or admission type have been dropped from analysis.

represents a change in preference among patients and providers. Similarly, research is needed to better delineate the reasons for admission of older pa- tients once they are in the ED and to determine the cost or value of emer- gency admissions for older adults. Finally, improved knowledge and understanding of the unique needs of older patients are essential, given the rapidly growing geriatric demographic.

References