The Burden of Emergency Department Use for Sickle Cell Disease: An Analysis of the National Emergency Department Sample Database

Abstract

It is estimated that there are 100,000 people living with sickle cell disease (SCD) in the United States.¹ The most common manifestation of sickle cell disease is vaso-occlusive crisis which is characterized by intermittent, unexpected episodes of excruciating pain. As these episodes often come on suddenly, much of the care for these crises occurs within emergency departments (EDs). Several studies have examined ED use and costs for certain groups of patients with sickle cell disease.^{2, 3;4} For example, in 1997 Woods et al found that 85.7% of 7202 hospital admissions for SCD in Illinois were for patients that came through the ED, and the total charges for sickle cell admissions in Illinois were found to be $30 million a year.² A recent study of healthcare use by children with sickle cell disease demonstrated that children insured by Medicaid had higher ED utilization than those with private insurance (57% v 45%).⁵ The purpose of our study was to provide national level estimates of ED utilization by SCD patients, which have not previously been available.

There were a total of 50,418 ED visits for patients with SCD sampled in the Nationwide Emergency Department Sample (NEDS) in 2006. This represents a national estimate of 232,381 ED visits. There were an estimated 188,194 (81%) adult ED visits and 44,188 (19%) pediatric (<18 years) visits. The characteristics of the sickle-related ED visits in 2006 are shown in Table I.

Table I.

National Estimates of SCD-related ED Visits in the US for the year 2006

Variable	Adult (n=188,194)* (≥18 yrs)	Pediatric (n=44,188)*
Female (%, 95% CI)	57.4 (55.1, 59.7)	49.1 (47, 51.2)
Mean Age (range)	32 (18–94)	8 (0–17)
Admitted (%, 95% CI)	38.4 (37, 41.8)	43.5 (38.7, 48.4)
Payor (%, 95% CI)
Medicaid	45.7 (43.2, 48.1)	67 (61.1,72.5)
Medicare	27.6 (25.8, 29.6)	0.07 (0.3,0.20)
Private	15.6 (14.1, 17.4)	26.2 (21.6, 31.4)
Self pay	8.3 (7.4, 9.3)	3.8 (3.1, 4.8)
Other	2.3 (1.4, 4)	1.9 (1.5, 2.6)
Metropolitan(%, 95% CI)	89.4 (85.8, 92.1)	91.7 (87, 94.8)
Diagnoses (%, 95% CI)
Crisis	73.7 (71.7, 75.6)	54 (51.6, 56.5)
Acute Chest	2.3 (1.9, 2.7)	4.7 (3.4, 6.3)
Pneumonia	5.2 (4.8, 5.6)	6.8 (5.6, 8.2)
Stroke	0.27 (0.19, 0.38)	0.17 (0.11, 0.26)
Asthma	6.6 (5.7, 7.6)	9.9 (8.3, 11.9)
Died during visit- (#)	1076 (858, 1293)	33 (−4, 69)
Admit (%)	38.3 (36.2, 40.7)	43.5 (38.7, 48.4)
Mean ED Charges (95% CI)	$1572(1495,1650)	$1351 (1225,1477)
Total ED Charges (95%CI)	$2.96 mill (2.56, 3.36)	$5.97 mill (3.86, 8.09)

^*Unweighted n-Adult=40,991, Pediatric=9427

Emergency Departments were categorized by SCD volume to determine relationships between experience and outcomes. There were 441 EDs that had between 0–50 sickle cell ED visits in 2006, 180 EDs that had between 51 and 250 sickle cell ED visits, 35 EDs that had between 251 and 500 sickle cell ED visits, and 14 EDs that had over 500 sickle cell ED visits in 2006. Over 23,000 ED visits for SCD occurred in EDs that had between 51–250 sickle cell visits in 2006. The EDs with the highest sickle cell volumes also had the largest total ED volumes. EDs that had over 500 sickle cell visits a year had a mean of 100,284 total ED visits in 2006 and those hospitals that had the fewest sickle visits had the lowest total ED volumes (mean 33,450). All of the hospitals with the highest sickle cell volumes were located in metropolitan areas while 27% of EDs with the smallest sickle cell volumes were located in non-metropolitan areas.

The total charges for all of the sickle cell ED visits were estimated to be $356 million, of which $238 million were for patients not subsequently admitted to the hospital. The combined ED and inpatient charges for those admitted were estimated to be $2.4 billion. In a multiple linear regression model that included age, whether the patient was admitted, payor, geographic region and sex, being female and increasing age independently predicted higher ED charges (Table II). Total charges per visit were significantly higher for those ED visits that occurred in the West (compared to the South) and for those visits that occurred in EDs that had more than 500 sickle cell patient visits in 2006 (compared to the lowest volume sickle cell EDs). The higher charges seen in females remained even after all visits with diagnoses associated with childbirth were removed from the analysis.

Table II.

Predictors of Total ED Charges and Adjusted Odds of Hospital Admission from the ED

Variable	Difference in Total ED Charges ($)	AOR† for admission
Female	88.0 (30.5,145.3)***	1.07 (0.98, 1.17)
Age	6.8 (3.1, 10.5)***	1.00 (0.998, 1.01)
Pneumonia	160.5 (97.4,223.6) ***	16.2 (13.91, 18.74)***
Stroke	573.2 (208.4, 938) ***	5.24 (2.18, 12.61)***
Asthma	34.8 (−49.6, 119.2)	2.67(2.20, 3.24)***
Acute chest	49.8 (−53.7,153.3)	5.74 (3.00, 11.04)***
Crisis	173.6 (86.0,261.2)***	2.86 (2.52,3.23)***
Payor
Medicare	60.2 (−16.8,137.4)	1.00 (0.880,1.15)
Private	57.0 (−50.1,164.2)	1.08 (0.97, 1.21)
Self-Pay	−27.5 (−139.5,84.5)	0.52 (0.44, 0.61)***
Medicaid	Ref	Ref
Region
Northeast	115.5 (−183.7, 414.6)	1.28 (0.98, 1.69)
Midwest	−137.3 (−377.6, 103)	0.99 (0.73,1.35)
West	474.3 (244,704.6)***	1.31 (1.04,1.67)**
South	Ref	Ref
SCD Hospital Volume
0–50	Ref	Ref
51–250	120.9 ((−24.3,266.1)	1.00 (0.84, 1.19)
251–500	46.9 (−189.5,283.4)	0.84 (0.65,1.10)
501+	243.9 (−2.2, 490)*	1.45 (1.02,2.05)****

^†Adjusted Odds Ratio

^****p = .04,

^***p<0.01,

^**p = .02,

^*p = .05

In a multiple logistic regression predicting admission from the ED, having a diagnosis code for vaso-occlusive crisis or for asthma independently increased the odds of hospital admission (Table II). If a patient was seen in an ED in the West their odds of being admitted was 1.3 times that of those patients seen in an ED in the South. Those patients that were seen in an ED that had over 500 ED visits for SCD in 2006 had a significantly higher odds of being admitted when compared to patients that were seen in EDs that had fewer than 51 ED visits for SCD in 2006. (Adjusted Odds Ratio [AOR] 1.45, p=0.04)

Table III shows a comparison of the number of admissions from the ED for SCD, asthma, HIV and CHF. In 2006, there were an estimated 68,420 hospital admissions from ED visits where SCD was listed as the first diagnosis. The mean charge for these visits was $21,679, giving a total estimated charge for SCD-related hospital admissions from the ED in 2006 of $1.5 billion. Using the estimated prevalence of 100,000 people with SCD in the US, this equates to 68.4 hospital admissions per 100 estimated people with the disease and $1.5 million in charges per 100 people with SCD. In comparison, there were 830,613 hospital admissions from ED visits where CHF was listed as the first diagnosis in 2006. The mean charge for these visits was $29,317, and the total estimated charges for hospital admissions from the ED in 2006 were $ 24.4 billion. Using the estimated prevalence of 4.8 million people with CHF in the US this equates to 17.3 hospital admissions per 100 estimated people with the disease and $ 500,000 in charges per 100 people with CHF. Comparable values for HIV and asthma can be seen in Table III.

Table III.

Charges Associated with Admission from the ED for Sickle Cell Compared to More Common Disorders

Diagnosis	Total admissions from ED	Estimated size of population in US	Mean charges ($)	Total estimate charges ($)	Admits/100 patients	Charges/100 patients ($)
SCD	68,420	100,000	21,679	1.5 bill	68.4	1.5 mil
Asthma	294,557	34 mill	16,485	4.9 bill	1.1	14,411
HIV	61,966	1.1 million	50,743	3.1 bill	5.1	281,818
CHF	830,613	4.8 million	29,317	24.4 bill	17.3	500,000

This paper is the first to look at national estimates of ED utilization by patients with SCD in the US. We have shown that most ED visits for SCD occur in EDs that see on average less than one person with SCD a day. Multivariable analysis shows that females have higher ED charges than males.

Another finding was the higher charges observed for EDs in the West and the higher charges seen in EDs that had the highest sickle cell volumes. Only California, Arizona, and Utah comprise the western states that contributed data to the NEDS, and it is likely that the vast majority of data on sickle cell patients is from California. As California has the fourth highest expenses per inpatient day in the US and may also have more inflated charges relative to cost this may explain the higher charges seen in EDs in the West observed in our study.⁶

The overall combined charges of ED and inpatient visits of $2.4 billion dollars appears to be a significant burden. We found it useful to compare this value with that of other more common diseases and describe the charges on a per 100 person basis. When we did this we found that those with SCD had three times the charges for hospitalizations associated with an ED visit per 100 people estimated to have the disease when compared to those with congestive heart failure. Similarly, the charges associated with ED use leading to hospitalization for SCD was higher per 100 persons than charges for HIV and asthma. The difference was due to a greater frequency of SCD-related ED visits, rather than higher mean charges per hospitalization. . Of note, when comparing the different medical conditions we were limited to the online Healthcare Cost and Utilization Project (HCUP) system that uses only the first listed diagnosis. It is possible that admissions for diseases such as CHF or AIDS were underestimated if other comorbid conditions were listed as primary. We suspect this to be true for all diagnoses examined, but may have been more likely in CHF where comorbidity is high. More concerning, however is that this data only reflects those charges associated with an ED visit. Steiner et al showed that 75% of all hospital admissions for SCD came through the ED in 2004⁷, thus suggesting that the overall burden of sickle care is higher than the $2.4 billion described in this paper.

There are several limitations of this study; the first is that data was analyzed based on ICD-9 codes. The validity of these codes is based on the quality of coders at each institution and therefore is subject to error. For this reason we did not examine whether there were significant differences in our outcomes for different sickle genotypes. Another important limitation is that the NEDS does not have unique identifiers and therefore it is impossible to know how many individuals with SCD are represented in this data. It is likely that a small proportion of patients disproportionally contributes to the number of ED visits, admissions and charges^4;8;9. Nevertheless, this still represents the experience of care for this population. In addition to these concerns it is important to note that there are a limited number of States involved in the NEDS. This may not allow a clear estimate of all patients with SCD who are not evenly geographically distributed throughout the US.¹ For example, New York did not contribute data to the NEDS and therefore a large number of patient visits expected to have occurred in New York is not represented. These geographic issues are likely to have led to an underestimate in charges. We did not convert the charges to costs. As discussed by Hsia et al ¹⁰ this is a complex process and computing the actual costs is difficult. In their study of ED charges from 1996–2004 they found the average paid charges in 2004 was 42% of the total charges. There was variability however in the reimbursement rates by insurer with Medicaid having the lowest and private insurers the highest rates of reimbursement. Although actual costs of caring for those with SCD in the ED are significantly lower than the charges, as Hsia points out, it is likely that ED’s are not recovering their costs.

The extent to which the enormous expense related to sickle care is due to inefficient care or due to the severity of the disease is currently unknown. However, prior investigators have demonstrated that use of other models of care such as day treatment centers^11–13 reduces ED use, subsequent admissions, and overall costs. It remains unclear how many patients are using the ED for their primary source of care. Further research into how those with SCD use the ED and evaluation of models of care that decrease their ED utilization need to be done.

Methods

We used the Nationwide Emergency Department Sample (NEDS) for the year 2006 as the source of data for our primary analyses. The NEDS is provided by the Healthcare Cost and Utilization Project (HCUP), and is sponsored by the Agency for Healthcare Research and Quality. The NEDS contains information from 950 hospitals and approximates a 20% stratified sample of US hospital based emergency departments. States that contribute data to the NEDS include: AZ, CA, CT, FL, GA, HI, IA, IN, KS, MA, MD, ME, MN, MO, NE, NH, NJ, OH, SC, SD, TN, UT, VT, WI. The discharge data represent total ED visits and hospital discharges and no individual patient level data is available. Using this dataset we identified all ED visits for adults (≥18 yrs) and children with the following SCD related ICD9 codes listed anywhere in the list of diagnoses: 28260, 28261, 28262, 28263, 28264, 28268, 28269, 28241, and 28242. . Of these 68% were coded as SS (282.61–62), 2% were coded as SC disease (282.63–64), 2% were coded as sickle beta thalassemia (282.41–42) and 28% were coded as other (282.60 &282.68–69). ICD9 code 517.3 was used to identify discharges with acute chest syndrome and 282.62, 282.64, 282.69, 282.42 were used to identify discharges with crises codes. CCS codes were used to pull discharges for pneumonia (122), stroke (109) and asthma (128). Bivariate and multivariate analyses were conducted using t-tests, oneway ANOVAs, and multiple linear or logistic regressions¹⁴ as appropriate. Sampling weights are included in the NEDS to allow for the generation of national estimates. As recommended by HCUP, who note that 23%% of ED visits are missing data on charges, estimates for the sum of charges were made by using the mean charges for missing data. Charges that are not included in the dataset include professional fees and non-covered charges. As the distribution of charges was skewed we did analysis using log transformed charges. This analysis did not change the outcome of the analysis and so for simplicity the non-transformed data is reported here. All analysis were done using Stata 10.0 (StataCorp LP College Station, TX).

In addition to the full NEDS dataset, the Agency for Healthcare Research and Quality makes freely available on-line a query system based on HCUP data¹⁵ We used this online query system to analyze data on total ED visits that ended in hospital admission for other common disorders in order to compare ED use for those disorders with use for sickle cell disease. In order to analyze charge data from the online system, we were required to restrict this analysis to those ED visits where our diagnoses of interest were the first listed diagnoses. We analyzed data on sickle cell disease, asthma, HIV and congestive heart failure (CHF) using the clinical classification software codes 61, 128, 5, and 108 (respectively) as defined in the HCUP system. Data about the prevalence of the diseases was collected from the websites of the Centers for Disease Control and the National Heart, Lung, and Blood Institute.