The authors have declared that no competing interests exist.
Conceived and designed the experiments: NC TW TS RYH. Performed the experiments: TS RYH. Analyzed the data: TS NC RYH. Contributed reagents/materials/analysis tools: RYH. Wrote the paper: NC TW RYH TS.
We examined the charges, their variability, and respective payer group for diagnosis and treatment of the ten most common outpatient conditions presenting to the Emergency department (ED).
We conducted a cross-sectional study of the 2006–2008 Medical Expenditure Panel Survey. Analysis was limited to outpatient visits with non-elderly, adult (years 18–64) patients with a single discharge diagnosis.
We studied 8,303 ED encounters, representing 76.6 million visits. Median charges ranged from $740 (95% CI $651–$817) for an upper respiratory infection to $3437 (95% CI $2917–$3877) for a kidney stone. The median charge for all ten outpatient conditions in the ED was $1233 (95% CI $1199– $1268), with a high degree of charge variability. All diagnoses had an interquartile range (IQR) greater than $800 with 60% of IQRs greater than $1550.
Emergency department charges for common conditions are expensive with high charge variability. Greater acute care charge transparency will at least allow patients and providers to be aware of the emergency department charges patients may face in the current health care system.
Emergency Departments (EDs) play a key role in the delivery of health care services for a wide variety of acute medical needs.
Rising healthcare charges and associated system cost control have been at the forefront of recent economic, political, and medical discussion.
This is a cross-sectional study of the 2006–2008 Medical Expenditure Panel Survey (MEPS), a public data source from the US Agency for Healthcare Research and Quality. The MEPS uses a complex sampling design based on the National Health Interview survey framework, and then applies survey weights to the absolute results to create representative estimates of the United States medical diseases profile, patient demographics, healthcare utilization and charges.
The MEPS uses multiple panels of households to create an overlapping data set from year to year. Each household serves for two full years on a panel, with the last year of data used to create overlap with new households on the panel. The designated household representative submits information for each individual household member. The MEPS collects information from the respondent on inpatient and outpatient medical usage from patient interviews and medical diaries. This data is then cross-referenced with provider and insurer records to ensure validity.
Our study was exempt from the Institutional Review Board at the University of California, San Francisco because we used a public data source that was masked for identifiers.
We gathered data on ED use along with associated ED charges from the 2006–2008 MEPS. We merged two MEPS data files, one with ED visits and the other with population characteristics, using a unique patient identifier. Patient demographics, insurance status and medical comorbidities were gathered from the MEPS population characteristics file, while the clinical characteristics for each visit were taken from the ED event file.
For each ED patient encounter, MEPS reports up to three 3-digit International Classification of Disease, Ninth Revision (ICD-9) condition codes along with an associated Clinical Classification Software (CCS) code (Agency for Healthcare Research and Quality, Washington, DC). The CCS condition code was used as the listed encounter diagnosis for our analysis.
Each ED encounter in MEPS between 2006 and 2008 was used as a separate unit of analysis regardless of patient identifier (n = 18,315). Our patient selection process is outlined in
Our primary outcome measure was total charge. The total charge recorded in the MEPS includes the sum of medical care, tests, and treatment (facility and physician fee). These charges do not represent the amount patients or insurers reimburse providers, but rather the total charge that patients or their insurance providers are billed.
We began by analyzing the demographic breakdown of the absolute and weighted number of visits (
Characteristic | Observations(n = 8,303) | Weighted visits (In millions)(n = 76.6 million) |
|
||
18–24 | 1556 (18.7%) | 13.9 (18.2%) |
25–29 | 1110 (13.4%) | 10.5 (13.8%) |
30–39 | 1892 (22.8%) | 16.3 (21.4%) |
40–49 | 1703 (20.5%) | 16.0 (21.0%) |
50–59 | 1545 (18.6%) | 14.5 (18.9%) |
60–64 | 497 (6.0%) | 5.2 (6.8%) |
|
||
Male | 3089 (37.2%) | 31.5 (41.0%) |
Female | 5214 (62.8%) | 45.2 (59.0%) |
|
||
Private | 4068 (48.9%) | 44.7 (58.4%) |
Medicaid | 1822 (21.9%) | 12.7 (16.5%) |
Uninsured | 3048 (36.7%) | 24.7 (32.2%) |
|
||
Asian/Pacific Islander | 166 (2.0%) | 1.4 (1.8%) |
Black | 1899 (22.9%) | 12.5 (16.3%) |
Non White Hispanic | 1778 (21.4%) | 9.6 (12.6%) |
White | 4196 (50.5%) | 50.9 (66.4%) |
Other/multiple race | 264 (3.2%) | 2.3 (3.0%) |
|
||
Hypertension | 2429 (29.3%) | 21.3 (27.8%) |
Hypercholesterolemia | 1886 (22.7%) | 18.2 (23.7%) |
Asthma | 1481 (17.8%) | 13.3 (17.3%) |
Diabetes | 786 (9.5%) | 6.0 (7.8%) |
Our sample consisted of 8,303 observations, representing 76.6 million ED visits. About 22.8% of the observations were from patients between 30–39 years old. Of ED visits, 62.8% were female patients, 48.9% privately insured, and 50.5% self-reported as white. As shown in
Disease | Observations(n = 2,717) | Weighted Visits(n = 25.3 million) | Rank (Observations, Weighted Visits) |
Sprains and strains | 415 (15.3%) | 4.4 million (17.4%) | 1, 1 |
Other injury | 354 (13.0%) | 3.3 million (13.0%) | 2, 3 |
Open wounds of extremities | 312 (11.5%) | 3.4 million (13.4%) | 3, 2 |
Pregnancy | 298 (11.0%) | 2.1 million (8.3%) | 4, 6 |
Headache | 287 (10.6%) | 3.1 million (6.7%) | 5, 4 |
Back problems | 250 (9.2%) | 2.1 million (7.9%) | 6, 5 |
Upper respiratory infection | 215 (7.9%) | 1.7 million (5.9%) | 7, 8 |
Kidney stone | 204 (7.5%) | 2.0 million (7.9%) | 8, 7 |
Urinary tract infection | 192 (7.1%) | 1.5 million (5.9%) | 9, 10 |
Intestinal infection | 190 (7.0%) | 1.7 million (6.7%) | 10, 9 |
During our study period, the median charge for outpatient conditions in the emergency room was $1233 (95% CI: $1199– $1268). Upper respiratory infections had the lowest median charge of $740 (95% CI: $651– $817), while a urinary tract calculus (kidney stone) was charged the highest median price of $3437 (95% CI: $2917– $3877).
Regarding variability of charges, all diagnoses had an interquartile range (IQR) greater than $800 with 60% of IQRs greater than $1550. The diagnoses with the largest IQRs were: “calculus of urinary tract” (kidney stone) ($3742), “normal pregnancy and delivery” ($2008), and “urinary tract infection” ($1975).
Diagnosis | Median charge ($) (95% CI) | Mean charge ($) (95% CI) | Inter-quartile range (IQR) | Minimum charge | Maximum Charge |
Sprains & strains | 1051 (982–1110) | 1498 (1304–1692) | 1018 | 4 | 24110 |
Other injury | 1151 (1003–1281) | 2103 (1770–2437) | 1594 | 46 | 27238 |
Open wounds of extremities | 979 (864–1090) | 1650 (1341–1959) | 924 | 29 | 25863 |
Normal pregnancy and/or delivery | 1204 (1027–1384) | 2008 (1701–2315) | 2008 | 19 | 18320 |
Headache | 1210 (1093–1344) | 1727 (1510–1943) | 1572 | 15 | 17797 |
Back problems | 871 (741–984) | 1476 (1265–1687) | 1189 | 66 | 10403 |
Upper respiratory infection | 740 (651–817) | 1101 (891–1312) | 827 | 19 | 17421 |
Kidney stone | 3437 (2917–3877) | 4247 (3642–4852) | 3742 | 128 | 39408 |
Urinary tract infection | 1312 (1025–1580) | 2598 (1780–3416) | 1975 | 50 | 73002 |
Intestinal infection | 1354 (1114–1524) | 2398 (1870–2927) | 1960 | 29 | 29551 |
Total outpatient conditions | 1233 (1199–1268) | 2168 (2103–2233) | 1957 | 3.5 | 73,002 |
All diagnoses have an IQR of greater than $800. The diagnoses with the largest IQRs were kidney stone ($3742), normal pregnancy and delivery ($2008), and urinary tract infection (UTI) ($1975).
Analysis by insurance group (
Vertical bars indicate median charge for each of the ten conditions by insurance type: uninsured (black), Medicaid (dark grey), and private insurance (light grey). Medicaid patients were charged the most overall (median $1305), followed by private insurance ($1245), and uninsured patients ($1178).
Using the MEPS 2006–2008, we find previously undocumented patterns in emergency department charges for the ten most frequent outpatient diagnoses. The most frequent outpatient diagnoses were sprains, other injuries and open wounds of extremities. The median charge for outpatient conditions in the emergency department was $1233, which is 40% more than the average American pays in rent each month ($871).
While overall in hospital charge burden and variation by diagnosis has been well studied,
We report ED charges as they would appear on a patient’s bill, revealing the discrepancies in charges for the same diagnoses that patients are generally unaware of. Providers are often at a loss when their patient questions them about the charges for a certain procedure or treatment.
It is important to note that while these ten most common conditions comprised 32% of outpatient ED visits, these are most likely not the most costly conditions. Had we chosen a methodology to isolate the most costly conditions, the median and mean charges would be much higher. Our goal was to provide a representation of the burden of the most common conditions, rather than the most expensive conditions.
Efforts to increase price transparency have been proposed by over 30 states and are being pursued by the public and private sector as the next phase in medical care.
Further research should examine the sources of this variation in care within diagnoses in the emergency department, as well as how charge transparency could work to reduce this variability and increase healthcare efficiency.
Our study has several limitations given its retrospective design and information available for analysis in the Medical Expenditure Panel Survey. First, MEPS relies on survey responses and therefore could be subject to recall bias. However, MEPS charge information is based on responses from both provider and patient, and therefore charge variations between diseases should not be affected.
Second, diagnoses were reported using the Clinical Classification Software (CCS). While we did describe patient-level clinical comorbidities present in the data, we did not investigate how variation in charges could be due to differences in patient condition severity or other factors unable to be captured from these administrative datasets. For instance, the diagnosis “normal pregnancy and delivery” included a wide spectrum of presentations ranging from woman in active labor needing admission to otherwise healthy woman during the course of their pregnancy. Further research should try to elucidate the variation in costs controlling for such clinical severity factors.
Finally, we did not adjust for the facility where treatment was received. Differences in baseline charges between hospitals have been well documented and are in part due to factors including geographical differences, provider reimbursement variation, and health care monopolies.
Emergency departments play a valuable role in healthcare delivery, yet consumers know little concerning their ED charges before they receive the bill. In this context, we have identified a high charge burden and charge variation for those that seek outpatient care in the ED. Whether or not acute care charge transparency will aid in mitigating costs still needs to be investigated, however, better information for patients and providers on consumer cost of medical care going forward will allow patients to be aware of the charges they face in the ED.
We especially thank Julia Brownell, BA, for editorial and technical assistance, and Ralph Gonzales, MD, MSPH, for his comments on the manuscript.