Emergency Department Visits and Neuroimaging for Concussion Patients from 2006–2011

Abstract

Objectives

Concussion is the most common type of mild traumatic brain injury for which patients present to the emergency department (ED). It is critical to understand the contemporary epidemiology of concussion and rates of head computed tomography (CT) use in head-injured patients in order to inform education of evidence-based clinical practice guidelines to emergency medicine providers.

Methods

This was a cross-sectional analysis of the Nationwide Emergency Department Sample (NEDS) for years 2006 to 2011, representing a stratified probability sample of all U.S. hospital-based EDs. Patients of all ages with concussion diagnoses were included, and those with intracranial hemorrhages or prolonged loss of consciousness were excluded. Descriptive and bivariate statistics were summarized for patient demographics, injury mechanism, Injury Severity Scores (ISS), and concussion incidence rates, based on U.S. Census Bureau population estimates.

Results

There were 756,214,762 (weighted) ED visits in the NEDS between 2006 and 2011, of which 0.5% received diagnoses of concussion. The national incidence rate of concussion visits increased 22.6% from 195 visits per 100,000 person-years in 2006, to 239 visits per 100,000 person-years in 2011. The incidence of concussion visits increased by 28.1% from 2006 to 2011 overall (580,573 to 743,994) and within all age group categories. The rate of head CT use in patients diagnosed with concussion increased 35.7% (range 34.5% to 46.8%) for the entire group from 2006 to 2011 (p < 0.0001). The injury severity of patients’ injuries decreased over time (66.4% ISS < 5 in 2006, and 75% in 2011; p < 0.0001), while the proportion of discharged patients increased over time (78.1% in 2006, 86.6% in 2011; p < 0.0001).

Conclusions

ED visits for concussions have increased over time, with a corresponding increase in head CT utilization despite a decrease in injury severity. Increased visits may be due to more concussion awareness and recognition of subtle injuries. Evidence-based clinical practice guidelines for neuroimaging in head-injured patients and management of concussion should be reinforced to emergency medicine providers.

INTRODUCTION

Concussion is a mild traumatic brain injury (TBI) induced by biomechanical forces, and has garnered increased attention by clinicians, scientists, and the lay public. Older estimates have demonstrated that there are approximately 1.3 million annual emergency department (ED) visits for TBI in all ages and of all severities,¹ and around 100,000 to 140,000 annual ED visits for concussion in children.^2,3 One recent analysis demonstrated a 29.1% increase in ED visits for TBI of any severity and in all age groups over a recent 5-year period, with over 2.5 million total visits in 2010.⁴ Given the media’s increased awareness of mild TBI, our aim was to quantify the frequency of ED visits specifically for concussion using a contemporary data source and a granular diagnosis inclusion criterion, and secondarily to describe the frequency of head computed tomography (CT) use for these patients.

METHODS

Study Design

This was a cross-sectional analysis of an existing government-sponsored administrative database. This study using de-identified administrative data, and was exempt from review by our institutional review board.

Study Setting and Population

The Nationwide Emergency Department Sample (NEDS) for years 2006 to 2011 was used. The NEDS is the largest all-payer hospital based emergency department (ED) database, and is part of the Healthcare Cost and Utilization Project (HCUP) sponsored by the Agency for Healthcare Research and Quality.⁵ The database is created using the HCUP State Emergency Department Databases and the State Inpatient Databases, and is a 20% stratified probability sample of all U.S. hospital-based EDs consisting of data provided by 33 unique hospitals. The NEDS stratifies on hospital characteristics in order to create a representative sample of hospital based EDs in the United States. National estimates of all hospital-based ED events are calculated by using the sample weights provided in the data set; 120 to 130 million estimated events with more than 100 data points are generated per year. Quality and equivalency checking of the variables was completed when the data sets were combined to create the analytics data set.

Study Protocol

Analytic sample selection criteria

Concussion was defined based on the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis code. The patient needed to have one or more ICD-9 codes for concussion (850.x), post-concussion syndrome (310.2), or skull fracture with concussion but without intracranial injury (800.09, 800.59, 801.09, 801.59, 803.09, 803.59, 804.09, 804.59); and not have any ICD-9-CM diagnosis code for TBI involving intracranial hemorrhage or injury (diagnostic description ICD9 TBI code keywords that included “contusion,” “hematoma,” “hemorrhage,” “laceration,” “with intracranial injury,” “with prolonged [more than 24 hours] loss of consciousness”). This latter group of patients with more severe TBI was considered separately for an analysis to compare their rates of head CT use to those of concussion patients. Additionally, in order to compare rates of concussion visits with those of other minor head injury without mild, moderate, or severe TBI, we also separately identified patients with ICD9 code 959.09 (head injury, unspecified).

E-code group assignment

The ICD-9-CM external cause code (E-codes) groups were created with the guidance of the Centers for Disease Control and Prevention’s recommended framework of E-code groupings for presenting injury mortality and morbidity data resource. Each event had up to four E-codes and the first valid E-code in the sequence was used for E-code group classification. Most injured patients only had one E-code assigned, and those with multiple codes had their primary mechanisms listed first. The final eleven E-code groups included: fall, motor vehicle traffic – occupant, motor vehicle traffic – motorcyclist, motor vehicle traffic – other, pedal cyclist, pedestrian, transport, struck by or against, sports, other, and unknown or no valid E-code.

Injury severity

Abbreviated Injury Scale (AID) scores were used to derive Injury Severity Scores (ISS).⁶ The first author, certified in AIS scoring, mapped each ICD-9-CM code associated with the visit to AIS98 codes using the ICDMAP-90 software, then manually re-mapped codes to the most recent AIS 2005/2008 versions using the AIS manual and the ICD-9-CM injury descriptions.⁷ AIS severity scores are on an ordinal scale and range from 1 (minor) to 6 (untreatable). ISS for each ED visit were calculated by summing the squares of the AIS scores of the three most severely injured body regions, and a maximum AIS 6 in any body region defaulted to an ISS of 75.⁶

CT scan

Utilization rates of CT scans of the head or brain were calculated by identifying the relevant codes in the procedure code fields, including 15 Current Procedure Terminology (CPT) procedure code fields (e.g., 70450, 70460, 70470) and nine ICD-9 procedure code fields (e.g., 8703, 8704) for each encounter.

Incidence rates

United States Census Bureau population estimates were used to calculate concussion incidence rates for the respective years of data analyzed.

Data Analysis

Descriptive statistics were calculated and Rao-Scott chi-square tests were conducted to summarize and compare patient and hospital characteristics by year. Weighted statistical techniques were used for all analyses to produce national event-level estimates with the NEDS hospital stratum as the stratification unit and the ED event as the unit of analysis. Analyses were conducted using SAS 9.3.

RESULTS

There were 756,214,762 (weighted) ED visits in the NEDS between 2006 and 2011, of which 0.5% received diagnoses of concussion (Table 1). The national incidence rate of all ED visits was 41,291 visits per 100,000 person-years, with a rate increase of 4.5% from 2006 to 2011, based on weighted ED visits in the NEDS and U.S. Census Bureau population estimates. The national incidence rate of concussion visits increased 22.6% from 195 visits per 100,000 person-years in 2006 to 239 visits per 100,000 person-years in 2011.

Table 1.

Characteristics of ED Visits Attributed to Concussions in the United States, 2006–2011^*

	2006			2011				All Years
	n	Weighted %	Weighted 95% CI	n	Weighted %	Weighted 95% CI	P- value†	n	Weighted %	Weighted 95% CI
Overall	580,573			743,994				4,038,762
Age, years
Mean for all	31.4		(30.8–31.9)	32		(31.7–33.1)	<0.0001	32.2		(31.9–32.5)
0–11	71,174	12.3	(11.3–13.2)	91,737	12.3	(11.3–13.4)	<0.0001	497,319	12.3	(11.8–12.8)
12–17	103,528	17.8	(17.0–18.7)	136,587	18.4	(17.3–19.5)		711,138	17.6	(17.1–18.1)
18–24	107,050	18.4	(17.9–19.0)	126,463	17.0	(16.4–17.6)		707,304	17.5	(17.2–17.8)
25–34	84,976	14.6	(14.1–15.1)	106,532	14.3	(13.8–14.9)		585,992	14.5	(14.3–14.8)
35–44	71,856	12.4	(11.9–12.8)	78,027	10.5	(10.1–10.9)		459,199	11.4	(11.2–11.6)
45–64	92,016	15.8	(15.3–16.4)	129,709	17.4	(16.8–18.1)		685,987	17.0	(16.7–17.3)
65+	49,949	8.6	(8.2–9.0)	74,897	10.1	(9.6–10.6)		391,484	9.7	(9.4–10.0)
Sex
Male	349,497	60.2	(59.6–60.8)	422,134	56.8	(56.2–57.4)	<0.0001	2,349,939	58.2	(57.9–58.5)
Female	230,961	39.8	(39.2–40.4)	321,296	43.2	(42.6–43.8)		1,686,329	41.8	(41.5–42.1)
Median household income of patient's zip code
$1–$40,999	139,100	24.7	(22.7–26.7)	181,632	25.0	(23.1–26.9)	0.0596	986,363	25.1	(24.0–26.2)
$41,000–$50,999	145,739	25.8	(24.3–27.3)	177,548	24.4	(22.7–26.2)		1,054,856	26.8	(25.8–27.8)
$51,000–$66,999	140,676	24.9	(23.6–26.3)	188,730	25.9	(24.4–27.5)		980,023	24.9	(24.0–25.8)
$67,000+	138,645	24.6	(22.3–26.8)	179,396	24.7	(22.2–27.1)		911,297	23.2	(21.8–24.6)
Primary payer
Medicare	52,677	9.1	(8.7–9.6)	85,290	11.5	(10.9–12.1)	<0.0001	424,082	10.6	(10.3–10.8)
Medicaid	77,844	13.5	(12.6–14.3)	129,946	17.6	(16.8–18.3)		609,953	15.2	(14.7–15.6)
Private, including HMO	284,447	49.2	(47.8–50.6)	351,269	47.4	(46.1–48.8)		1,945,986	48.4	(47.7–49.2)
Self	105,320	18.2	(16.9–19.5)	111,383	15.0	(14.1–16.0)		670,931	16.7	(16.1–17.3)
No charge	3,992	0.7	(0.5–0.9)	2,809	0.4	(0.2–0.5)		24,298	0.6	(0.4–0.8)
Other	53,869	9.3	(8.5–10.1)	59,706	8.1	(7.4–8.7)		342,300	8.5	(8.1–8.9)
Region of hospital
Northeast	116,282	20.0	(15.9–24.2)	139,986	18.8	(15.5–22.2)	0.9828	765,822	19.0	(17.1–20.8)
Midwest	148,539	25.6	(21.7–29.5)	179,973	24.2	(20.8–27.6)		1,018,791	25.2	(23.3–27.2)
South	183,204	31.6	(27.5–35.6)	272,244	36.6	(32.6–40.5)		1,397,786	34.6	(32.4–36.8)
West	132,548	22.8	(19.0–26.7)	151,791	20.4	(17.8–23.0)		856,363	21.2	(19.6–22.8)
Urban/rural location of the patient's residence
Metropolitan	467,964	81.4	(79.4–83.4)	595,150	80.4	(78.5–82.3)	0.6371	3,207,264	80.1	(79.1–81.1)
Micropolitan/rural	106,732	18.6	(16.6–20.6)	144,953	19.6	(17.7–21.5)		797,158	19.9	(18.9–20.9)
Type of institution
Metropolitan non-teaching	230,231	39.7	(35.4–43.9)	297,065	39.9	(36.3–43.5)	0.8135	1,683,423	41.7	(39.6–43.8)
Metropolitan teaching	247,623	42.7	(37.8–47.5)	311,631	41.9	(37.6–46.2)		1,614,979	40.0	(37.6–42.4)
Non-metropolitan	102,718	17.7	(15.3–20.1)	135,299	18.2	(15.9–20.5)		740,359	18.3	(17.1–19.6)
Disposition from ED
Routine discharge	453,207	78.1	(75.6–80.5)	644,358	86.6	(85.3–87.9)	<0.0001	3,354,843	83.1	(82.1–84.1)
Admitted to same hospital	88,795	15.3	(13.1–17.5)	74,659	10.0	(8.8–11.3)		507,913	12.6	(11.6–13.5)
Other‡	38,571	6.6	(5.1–8.2)	24,977	3.4	(2.9–3.8)		176,006	4.4	(3.9–4.8)
Mechanism of concussion‡
Fall	154,712	26.6	(25.7–27.6)	234,475	31.5	(30.6–32.5)	<0.0001	1,217,621	30.1	(29.6–30.7)
Motor vehicle traffic and pedestrian§	137,679	23.7	(21.9–25.5)	130,487	17.5	(16.1–19.0)		808,521	20.0	(19.2–20.8)
Pedal cyclist	18,125	3.1	(2.9–3.3)	21,461	2.9	(2.7–3.1)		121,687	3.0	(2.9–3.1)
Transport, other	22,783	3.9	(3.6–4.3)	21,172	2.8	(2.7–3.0)		140,587	3.5	(3.3–3.6)
Struck by, against	92,183	15.9	(15.3–16.5)	120,816	16.2	(15.6–16.8)		650,152	16.1	(15.8–16.4)
Sports	50,180	8.6	(7.9–9.4)	78,854	10.6	(9.8–11.4)		383,780	9.5	(9.0–10.0)
Other	40,897	7.0	(6.5–7.6)	59,284	8.0	(7.5–8.4)		305,217	7.6	(7.3–7.8)
No external cause code available	64,013	11.0	(9.4–12.6)	77,445	10.4	(8.5–12.3)		411,197	10.2	(9.3–11.1)
Injury Severity Score (ISS)
< 5	359,406	66.4	(64.4–68.4)	504,014	73.5	(71.4–75.7)	<0.0001	2,631,073	70.2	(69.2–71.2)
5–14	172,012	31.8	(30.0–33.5)	172,058	25.1	(23.2–27.0)		1,056,140	28.2	(27.3–29.0)
≥ 15	10,061	1.9	(1.5–2.2)	9,286	1.4	(1.1–1.6)		60,801	1.6	(1.4–1.8)
Neuroimaging in the ED
Head CT	200,078	34.5	(31.0–38.0)	348,232	46.8	(43.8–49.8)	<0.0001	1,629,921	40.4	(38.4–42.3)
Brain MRI	716	0.1	(0.1–0.2)	2,215	0.3	(0.3–0.3)	<0.0001	8,572	0.2	(0.2–0.2)

HMO = health maintenance organization; CT = computed tomography; MRI = magnetic resonance imaging

^*All values are weighted. Visits with missing values were excluded from calculations except for mechanism of concussion. Data were missing for the following variables over all years: age, n=339; sex, n=2,494; median household income of patient's zip code, n=106,223; primary payer, n=21,212; urban-rural location of the patient's residence, n=34,340; mechanism of concussion, n=411,197; Injury Severity Score, n=290,748.

^†A t-test of association among means was used for continuous age and a Rao-Scott modified chi-square test, a design-adjusted test of association between the row and column variables, was used for all other categorical variables.

^‡Transfer to short-term hospital; other transfers, including skilled nursing facility, intermediate care, and another type of facility; home health care; against medical advice; died in ED; discharged/transferred to court/law enforcement; not admitted, destination unknown; discharged alive, destination unknown (but not admitted).

^§The motor vehicle traffic and pedestrian category includes: motor vehicle traffic - pedestrian, pedestrian other, motor vehicle traffic - occupant, motor vehicle traffic - motorcyclist, motor vehicle traffic - unspecified as well as all other codes in motor vehicle traffic that are not in a motor vehicle traffic subgroup; the pedal cyclist category includes: motor vehicle traffic - pedal cyclist and pedal cyclist other; the sports category is created by codes E0001 – E0010, E828.0, E885.0 – E885.4, E917.0, E917.5; the other category includes: cut/pierce, drowning/submersion, fire/burn, firearm, machinery, natural/environmental, overexertion, poisoning, suffocation, other specified and classifiable, other specified not elsewhere classifiable, unspecified, adverse effects.

The incidence of concussion visits increased by 28.1% from 2006 to 2011 overall (580,573 to 743,994) and within each age group (0 to 11 years old: 71,174 to 91,737; 12 to 17 years old: 103,528 to 136,587; 18 to 24 years old: 107,050 to 126,463; 25 to 34 years old: 84,976 to 106,532; 35 to 44 years old: 71,856 to 78,027; 45 to 64 years old: 92,016 to 129,709; 65+ years old: 49,949 to 74,897). Similarly, rates of non-TBI head injury (ICD-9 code 959.09, ‘head injury unspecified’) increased by 31.5% from 2006 through 2011.

Most concussions were isolated injuries. Injury severity was low (70.2% of all cases had ISS < 5), and decreased over time (66.4% of patients had ISS < 5 in 2006, with 73.5% in 2011, p < 0.0001). Most patients (83.1%) were discharged from the ED, and the proportion of discharged patients increased over time (78.1% in 2006, 86.6% in 2011; p < 0.0001). The rate of head CT use for concussion patients increased 35.7% (range 34.5% to 46.8%) for all ages from 2006 to 2011 (Table 2). In order to evaluate whether this reflected an increase in true head CT use rather than an increase in CT coding in the dataset, we examined the rate of listed head CT use for TBI patients with intracranial hemorrhage (who presumably all received head CT). This rate (83.3%) remained relatively stable over time (85.1% in 2006, to 82.5% in 2011).

Table 2.

Head Computed Tomography by Age Group Among ED Visits for Concussions in the United States, 2006–2011^*

Age group, yrs	2006, n (%)	2011, n (%)	All Years, n (%)
0 – 11	23,051 (32.4)	34,351 (37.4)	174,233 (35.0)
12 – 17	37,837 (36.5)	60,506 (44.3)	290,573 (40.9)
18 – 24	37,268 (34.8)	62,287 (49.3)	295,061 (41.7)
25 – 34	29,445 (34.7)	52,465 (49.2)	242,797 (41.4)
35 – 44	25,628 (35.7)	39,396 (50.5)	191,036 (41.6)
45 – 64	31,976 (34.8)	64,479 (49.7)	285,575 (41.6)
65+	14,874 (29.8)	34,741 (46.4)	150,584 (38.5)

^*All values are weighted. 339 visits with missing values were excluded from calculations.

DISCUSSION

We observed a 28.1% increase in ED visit incidence for concussion in the 6-year study period overall and within each age group. These results align with a recent analysis of ED visits for TBI of all severities using the same dataset,⁴ although our study differed as it focused specifically on concussion using a more granular diagnosis criteria, included an additional recent year of analysis (2011), and described head CT use over time and in various age groups. Despite long-standing awareness about the harm of ionization radiation⁸ and prediction rules designed to minimize neuroimaging in minor head injury,⁹ head CT use for concussion patients also increased. This is particularly concerning since our results also demonstrated a decrease in injury severity over time. Increased concussion visits may be due to a greater awareness of the injury and recognition of more subtle injuries, which are less likely to require neuroimaging. Emergency physicians should continue to apply evidence-based guidelines for neuroimaging, and should consider strategies to maximize use of validated prediction rules in order to minimize unnecessary radiation and cost associated with head CT use.¹⁰

LIMITATIONS

Since the concussion diagnosis was based on ICD-9-CM codes from retrospective administrative data, there may have been misclassification, either from inappropriate diagnosis or missed diagnosis, with resultant errors in the incidence of ED visits. We attempted to minimize the misclassification by excluding patients with any intracranial hemorrhage, even if they also had concussion codes. We also attempted to compare rates of concussion visits with that of non-TBI mild head injury, and found increases in both, which demonstrates an overall increased trend in ED visits for head trauma during the study period. There may also have been misclassification of CT head utilization or changes in coding practices over time versus actual utilization. Although there may have been missing CT data for the entire sample, the rate of CT use in concussion patients increased, while the rate in more severe TBI was stable, suggesting a true increase in CT use for concussion.

CONCLUSIONS

Based on our analysis of an administrative database, ED visits for concussion and use of head computed tomography have both increased over time, while injury severity has decreased. Emergency medicine providers should continue to adhere to evidence-based clinical practice guidelines for diagnosis and management of patients with concussion.