Shifting Trends in the Epidemiology of Cervical Spine Injuries: An Analysis of 11,822 Patients from the National Electronic Injury Surveillance System over Two Decades

Abstract

Cervical spine injuries (CSIs) are heterogeneous in nature and often lead to long-term disability and morbidity. However, there are few recent and comprehensive epidemiological studies on CSI. The objective of this study was to characterize recent trends in CSI patient demographics, incidence, etiology, and injury level. The National Electronic Injury Surveillance System was used to extract data on CSIs from 2002 to 2022. Weighted national estimates of CSI incidence were computed using yearly population estimates interpolated from U.S. census data. Data analysis involved extracting additional information from patient narratives to categorize injury etiology (i.e., fall) and identify CSI level. K-means clustering was performed on cervical levels to define upper versus lower cervical injuries. Appropriate summary statistics including mean with 95% confidence intervals and frequency were reported for age, sex, race, ethnicity, etiology, and disposition. Age between groups was compared using an independent weighted Z-test. All categorical variables were compared using Pearson chi-squared tests with Bonferroni correction for multiple comparisons. Ordinary least squares linear regression was used to quantify the rate of change of various metrics with time. A total of 11,822 patient records met the study criteria. The mean age of patients was 62.4 ± 22.7 years, 52.4% of whom were male and 61.4% of whom were White, 7.4% were Black, 27.8% were not specified, and the remaining comprised a variety of ethnicities. The most common mechanism of CSI was a fall (67.3%). There was a significant increase in the incidence of cervical injuries between 2003 and 2022 (p < 0.001). Unbiased K-means clustering defined upper cervical injuries as C1–C3 and lower cervical injuries as C4–C7. The mean age of patients with upper CSIs was 72.3 ± 19.6, significantly greater than the age of those with lower CSIs (57.1 ± 23.1, p < 0.001). Compared with lower CSI, White patients were more likely to have an upper CSI (67.4% vs. 73.7%; p < 0.001). While Black/African American (7.5% vs. 3.8%) and Hispanic (2.5% vs. 1.0%) patients were more likely to have a lower CSI (p < 0.001). Our study identified a significant increase in the incidence of CSIs over time, which was associated with increasing patient age. Our study detected a pragmatic demarcation of classifying upper injuries as C1–C3 and lower cervical injuries as C4–C7. Upper injuries were seen more often in older, White females who were treated and admitted, and lower injuries were seen more often in young, Black male patients who were released without admission.

Introduction

Spinal injuries, particularly cervical spine injuries (CSIs), represent a significant public health concern in the United States, with notable associations with falls among the aging population.^1,2 Spine injury (both neurological and musculoskeletal) continues to be a major cause of morbidity in America and around the world.^1,3 Specifically, injury to the cervical spine, or CSI, has been estimated to occur in approximately 50% of spine injuries.⁴ Research has also proposed a bimodal age distribution for CSIs, with such injuries being common at mean ages of 30 and 74, respectively.⁵

CSIs may occur anywhere in the cervical vertebral region, between C1 through C7, and may often occur at the atlantoaxial complex between C1 and C2.⁶ The atlantoaxial region is a very mobile joint surrounded by smaller vertebral bodies and traversed by several crucial neurovascular structures.⁷ This anatomy and unique biomechanics contribute to the likelihood of hyperflexion, hyperextension, and axial loading injuries.⁸ In fact, there is an increasing incidence of geriatric odontoid fractures and central cord injuries reported in recent literature.^9,10

Notably, traditional definitions of upper cervical injuries have been restricted to the C1–C2 levels. CSIs occurring in the subaxial region (C3–C7) tend to be due to high-energy accidents.⁶ Injury to this area may lead to severe impairment of surrounding structures, cardiopulmonary effects, and loss of motor function.^7,11,12 Although associated with severe morbidity, there is a paucity of research into CSI patient demographics in large population datasets.¹³ The lack of current population-based epidemiological studies highlighting the etiologies and personalized characteristics of CSI patients in the United States hampers efforts to prevent and address the injury. Consequently, elucidating the epidemiology of CSIs in the United States using a national representative database is of preeminent interest.

Two recent studies from 2023 reported the epidemiology of cervical spinal cord injuries (CSCIs) specifically (excluding other types of injuries that may occur to the cervical spine) using the North American Clinical Trials Network (NACTN) for Spinal Cord Injury registry.^14,15 The NACTN registry was established in 2004 to promote clinical translational research, better care protocols, and improved outcomes relating to spinal cord injury (SCI).¹⁶ Interestingly, one of the major findings from the NACTN data analysis was the suggestion that upper cervical injuries should be defined as C1–C3 and lower as C4–C7. Despite this interesting finding, the NACTN database only contains data on 1200 patients.¹⁶ Although these collected data within the NACTN database are of high quality, the data pool is not wide enough to facilitate analyses that may be truly representative on a national scale. Other sources of data on CSIs include the National Trauma Data Bank (NTDB) and the Kids’ Inpatient Databases (KID). The NTDB is a larger data repository than NACTN, but it is generated from only patients at Level 1 and 2 trauma centers, making it nonrepresentative of all CSI patients.¹⁷ In addition, the KID database is limited by its nature as a pediatric inpatient database that only represents children who are admitted and lacks information on patients who died or were never admitted for a hospital stay.¹⁸

One other source to supplement existing research is the National Electronic Injury Surveillance System (NEISS). NEISS is a system started in 1971 and maintained by the Consumer Product Safety Commission (CPSC) that monitors emergency department (ED) visits for injuries involving consumer products. Previous research identified that most of the injuries contained within the NEISS database were primarily due to falls, though other causes include blunt force trauma, penetrating injury, and motor vehicle accidents.¹⁹ Using the NEISS database, it is possible to provide additional population data on CSIs. Therefore, the purpose of this descriptive epidemiological study is to explicate recent statistics on CSIs while expanding on current knowledge of the mechanistic and demographic characteristics of CSI patients presenting to U.S. EDs between 2003 and 2022 through the NEISS database. Specifically, this study focused on (1) describing the demographic data such as age, sex, race, ethnicity, and cervical level with appropriate summary statistics, (2) examining trends in the incidence of cervical injuries over the last 20 years stratified by gender, age, and percent due to falls, and (3) comparing the frequency of upper and lower cervical injuries based on demographic variables (age, sex, race, ethnicity, the proportion of injuries due to falls, and disposition) as done in prior work.

Materials and Methods

Data source and case selection

The NEISS was used to extract information on CSIs from 2002 to 2023. NEISS is operated by the CPSC and collects data on injuries related to consumer products from over 100 representative EDs across the United States and its territories. Each of these EDs is assigned a statistical sample weight to allow for the estimation of nation-wide epidemiological trends using their individual data.^20,21 Data variables are collected by trained, onsite coders at each NEISS hospital and include basic demographic information, injury diagnosis, and a brief narrative description of the incident. Data derived from the NEISS datasets have been used to describe and analyze a wide variety of injuries and their association with specific products.^22–25 As the datasets are publicly available and deidentified, the study did not require institutional review board approval. In 2000, NEISS was expanded to collect data on all injuries for the Centers for Disease Control and Prevention through an interagency agreement.

Data were extracted from NEISS archive data from 2003 to 2022. A CSI was defined as any neck injury (code 89) with any of the following associated injury designations: fracture (code 57), dislocation (55), or nerve damage (61). In this manner, we were able to isolate all potential CSIs and potential associated instances of neurological compromise.

In order to compute national estimates of weighted data from the NEISS data, we also obtained U.S. population data from the United Nations World Population Prospects.²⁶ Mid-year population estimates were used from 2003 to 2021, and the medium-variant projection of population was utilized for 2022 (this corresponds to the median projection of thousands of demographic subgroups).

Data collation and variable generation

Once the data were collected, several steps were taken to extract additional information, primarily from patient narratives. Injury etiology was binarized into whether patients were injured due to a fall. Any patient narrative mentioning a patient fall was tagged using a wildcard search query within Microsoft Excel. The CSI level was identified using the narrative and a robust search of narratives, including a list of possible identifiers. For example, an injury at the C3 vertebra could be identified whether it was mentioned as “C3,” “C-3,” “C1-3,” “C2-5,” or otherwise. K-means clustering was performed on cervical levels to define upper versus lower cervical injuries. Patients were grouped into these groups based on their injury location, and individuals with injuries across both groups were excluded. Kernel density estimates (KDEs) were used to visualize and compare different patient subgroups. KDEs are nonparametric and smoothed continuous probability density estimates of random variables.

Statistical analysis

The incidence of CSI was quantified by first taking national estimates of disease within a year and dividing by the U.S. population that same year. The average patient age for the entire sample as well as each individual year was computed as a weighted mean within each group. Comparisons between upper and lower injuries were performed for patient age and fall occurrence using weighted independent z- and t-tests, respectively. Additional comparisons between categorical patient demographics were performed using Pearson chi-squared tests. Records that did not include demographic data were not included in the statistical testing across these categorical groups. Ordinary least squares linear regression was used to quantify the rate of change of various metrics with time. All statistical analysis was performed in Microsoft Excel and Python 3.9.16, with the following packages: numpy 1.22.3, pandas 1.4.2, matplotlib 3.5.1, seaborn 0.11.2, and statsmodels 0.13.5.^27–31 Multiple comparison testing was corrected via the Bonferroni correction.

Results

Baseline data

The initial query resulted in n = 11,822 unweighted records, representing a total weighted estimate of n = 453,069 ED visits related to CSIs (Table 1). The mean age of all patients with a CSI was 62.4 ± 22.7 years, with a majority of these patients being male (52.4%). Most patients identified as White (61.4%), with 7.4% identifying as Black or African American. Of note, 27.8% of records did not specify the patients’ race. Of those records that included data on ethnicity (25.6% of all records), a majority identified as non-Hispanic. Most injuries were the result of falls (67.3%).

Table 1.

Demographics and Injury Etiology and Levels of Included Patients

Age	62.4 ± 22.7
Sex
Female	215,808 (47.6%)
Male	237,261 (52.4%)
Race
White	278,212 (61.4%)
Black/African American	33,348 (7.4%)
Asian	5019 (1.1%)
American Indian/Alaska Native	896 (0.2%)
Native Hawaiian/Pacific Islander	352 (0.1%)
Other	9106 (2%)
Not Specified	126,136 (27.8)
Ethnicity
Hispanic	11,686 (2.6%)
Not Hispanic	104,283 (23.0%)
Not Specified	337,100 (74.4%)
Etiology
Fall	305,092 (67.3%)
Not fall	147,977 (32.7%)
C-spinea level
C1	35,783 (7.9%)
C2	54,312 (12.0%)
C3	12,639 (2.8%)
C4	15,612 (3.4%)
C5	133,573 (29.5%)
C6	26,200 (5.8%)
C7	28,707 (6.3%)
No level specified	299,019 (66%)

^aCervical spine.

Trends in CSI across age and sex

There has been a significant increase in the incidence of CSIs from 2003 to 2022, both across the total population and across male and female subpopulations (p < 0.001 for linear fits to data, Fig. 1A). The average age of the population of patients with CSIs has also significantly increased for both the total population and the male population (p < 0.001), whereas the female population’s age has not significantly changed since 2003 (p = 0.111, Fig. 1B). Of note, the injured male subpopulation is significantly younger than the female subpopulation across all years in this study (p < 0.001, weighted independent z-test, Fig. 2). The percent of injuries due to falls is decreasing slightly among the female population (p = 0.037) and increasing in the male population (p = 0.058), as illustrated in Figure 1C. Female CSI patients are significantly more likely to have suffered a fall compared with male CSI patients across all years (p < 0.001, independent t-test).

FIG. 1.

Figure showing change of several metrics in all patients and subdivided by patient sex from 2003 to 2022: (A) incidence of CSI, (B) average patient age, and (C) percentage of injuries due to falls. Legends include p-values of linear fits of each subgroup over time, α = 0.05.

FIG. 2.

Kernel density estimates of weighted (A) female and (B) male CSI subpopulations plotted across age and CSI level. Vertical histograms correspond to the weighted counts of each injury level, and horizontal histograms correspond to the weighted counts of each age.

Defining upper and lower cervical injuries

A total of 154,050 total patients (34% of the extracted records) included information on the level of CSI. KDEs for the entire cervical spinal injury population (CSI) plotted across age, and CSI level demonstrated an increased density of CSI occurring in two separate distributions. There is a dense distribution of injuries occurring between C1 and C3 for patients centered at 80 years of age. There is a second dense distribution occurring between C5 and C7 for patients centered at 65 years of age (Fig. 3). K-means clustering across cervical spine levels likewise indicated that the optimal division of cervical spine levels is C1–C3 and C4–C7 (Fig. 4). Furthermore, when performing KDEs with cervical level and age across sex, race, and etiology, there is a single density of injuries within C1–C3 and again from C4–C7 (Fig. 2, Fig. 5, Fig. 6). Of the 154,050 estimated patients, 81,333 (52.8%) had upper CSIs, 65,809 (42.7%) had lower CSIs based on the definition of upper being C1–C3 and lower as C4–C7, and 6908 (4.5%) had both upper and lower CSIs. These data are provided in Table 2.

FIG. 3.

Kernel density estimates of the entire cervical spinal injury population (CSI) plotted across age and CSI level. Vertical histograms correspond to the weighted counts of each injury level, and horizontal histograms correspond to the weighted counts of each age.

FIG. 4.

K-means clustering across cervical spine level indicates that the optimal division of cervical spine levels is C1–C3 and C4–C7.

FIG. 5.

Kernel density estimates of weighted (A) White and (B) non-White CSI subpopulations plotted across age and CSI level. Vertical histograms correspond to the weighted counts of each injury level, and horizontal histograms correspond to the weighted counts of each age.

FIG. 6.

Kernel density estimates of weighted (A) falls and (B) not falls CSI subpopulations plotted across age and CSI level. Vertical histograms correspond to the weighted counts of each injury level, and horizontal histograms correspond to the weighted counts of each age.

Table 2.

Comparison of Patient Demographics and Injury Etiology between Upper and Lower CSIs^a from 2003 to 2022 (Critical Alpha = 0.0125)

	C1–C3b	C4–C7c	p value
Age	72.3 ± 19.6	57.1 ± 23.1	p < 0.001
Sex			p < 0.001
Female	46344, (57.0%)	24061, (36.6%)
Male	34989, (43.0%)	41748, (63.4%)
Race			p < 0.001
White	59958, (73.7%)	44333, (67.4%)
Black/African American	3117, (3.8%)	4938, (7.5%)
Asian	918, (1.1%)	1044, (1.6%)
American Indian/Alaska Native	48, (0.1%)	240, (0.4%)
Native Hawaiian/Pacific Islander	69, (0.1%)	95, (0.1%)
Other	571, (0.7%)	1273, (1.9%)
Not specified	16653, (20.5%)	13886, (21.1%)
Ethnicity			p < 0.001
Hispanic	805, (1.0%)	1641, (2.5%)
Not Hispanic	19762, (24.3%)	16460, (25.0%)
Not specified	60766, (74.7%)	47708, (72.5%)
Etiology			p < 0.001
Fall	67862, (83.4%)	44535, (67.7%)
Not fall	13471, (16.6%)	21275, (32.3%)
Total count	81333	65809

^aCervical spine injuries.

^bCervical spine levels 1–3.

^cCervical spine levels 4–7.

Differences in demographics, etiology, and disposition based on CSI level

The mean age of patients with upper CSIs (C1–C3) was 72.3 ± 19.6, significantly greater than the age of patients with lower CSIs (C4–C7) (57.1 ± 23.1, p < 0.001 from weighted independent z-test, Table 2). A majority of upper CSI patients were female (57.0%), whereas a majority of lower CSI patients were male (63.4%, Pearson chi-squared p < 0.001). This is observed in Figure 2, where the majority of female patients have C1- or C2-level injuries, compared with the bimodal distribution of injury levels in male patients. There were significant differences in patient race and ethnicity between upper and lower CSI as per Pearson chi-squared (p < 0.001 for both), with the majority of White patients suffering from upper cervical injuries 59,958/104,291 (57.5%) and the majority of Black patients suffering from lower cervical injuries 4938/8055 (61.3%). The distribution across other racial and ethnic groups differed as well, but the numbers were too low to identify clear trends. The etiology of injury differed significantly between upper and lower CSIs (Pearson chi-squared p < 0.001), with 83.4% of upper CSIs due to falls compared with 67.7% of lower CSIs.

Patient disposition differed significantly between upper and lower cervical injuries (Pearson chi-squared p < 0.001, Table 3). Lower CSIs were examined and treated in the ED and released (27.5%), whereas upper CSIs were more commonly treated and transferred to another institution (22.9%). Aside from this difference, the most common disposition across both levels was treatment and admission from the ED (56.3% for upper CSI, 53.9% for lower CSI).

Table 3.

Comparison of Disposition of Patients with Upper and Lower CSIs^a from 2003 to 2022

	C1–C3b	C4–C7c	p value
Disposition			p < 0.001
Treated and transferred	18,612 (22.9%)	10,607 (16.1%)
Treated and admitted/hospitalized	45,820 (56.3%)	35,503 (53.9%)
Treated/examined and released	15,836 (19.5%)	18,080 (27.5%)
Held for observation	880 (1.1%)	1374 (2.1%)
Left without being seen	63 (0.1%)	213 (0.3%)
Fatality, Incl. DOA, died in ER	121 (0.1%)	32 (0.0%)
Total count	81,333	65,809

^aCervical spine injuries.

^bCervical spine levels 1–3.

^cCervical spine levels 4–7.

Discussion

This epidemiological study confirms several previously described trends in CSIs and introduces key new findings requiring further research. We observe a notable increase in cervical injuries with time, particularly among older women due to falls, with upper cervical injuries being more prevalent than lower cervical injuries. Using K-means clustering, we have also confirmed the demarcation between upper cervical injuries (C1–C3) and lower cervical injuries (C4–C7) proposed in previous research.¹⁴ Unlike previous studies, this work illustrated a bimodal age distribution in male patients as compared with a unimodal distribution among female patients with CSIs. Patients with upper cervical injuries tended to be older, White, and female, whereas patients with lower cervical injuries tended to be younger, Black, and male. Furthermore, those patients with upper CSIs were more likely to be evaluated, admitted, or transferred when seen in the emergency room, whereas those with lower injuries were significantly more likely to be discharged from the ED. These findings introduce several new questions for researchers and outline areas for improvement in delivering equitable care across different demographic groups.

Women with cervical injuries are older than their male counterparts

Women with cervical injuries were significantly older than men. Two explanations for such results are that either women live longer or older women tend to have cervical injuries more than men. A study that examined a cohort of patients 65 years or older found a higher incidence of females sustaining CSI compared with males.²¹ However, this contrasts with Lomoschitz et al. who also examined a cohort of patients 65 years and older, finding an overall incidence between females and males to be 48% and 52%, respectively.³² This is perhaps due to the difference in cohort sizes between Asemota and Lomoschitz (167,278 vs. 149).^21,32 Hu et al. found two peak incidence rates in their cohort, one in the second and third decade of life in males and another in elderly females.³³ This is consistent with previous studies that included cohorts of all ages, in which males predominated in the overall incidence of CSI, whereas the studies that focused on patients aged 65 and older indicate females have a higher or equivalent incidence of CSI.^{21,22,25,32,34}

A bimodal age distribution exists for men only

We observed an interesting bimodal age distribution for cervical level when stratified by gender. Most cervical injuries occurred solely in the upper cervical spine for women. For older men, cervical injuries occurred in the upper spine, but in the lower cervical spine for younger men. This aligns with previous research that demonstrated upper injuries occurring more frequently in older patients and lower injuries occurring more often in younger patients.¹⁶ The CSIs in younger patients are often due to high-impact injuries such as motor vehicle collisions.

It is unclear, however, why previous research studies found a bimodal age distribution among all CSI (regardless of sex). It is possible that in decades past, there was once a bimodal age distribution among the general population. A study suggests that this allowed for a bimodal age distribution among cervical injuries in both middle-aged and elderly groups.⁵ As time has progressed, however, there may be a more recent shift with a now unimodal distribution in the elderly. This shift to a unimodal age distribution, however, appears to be affecting women more than men. A 30-year study from the University of Alabama may support such theories for changing epidemiology.³⁵ They investigated the epidemiology of spinal cord injuries (irrespective of vertebral level and cooccurrence of fracture) from the years 1973 to 2003 and found that, over time, the mean age of injury was increasing. Between 1973 and 1979, the mean age was 28.9 years. However, between 2000 and 2003, the mean age of injury was 38.0 years. A more recent 2014 study investigating trends in incidence from 2007 to 2009 demonstrated a decreasing incidence in those aged 18–64 years and an increasing incidence in those aged 65 years and older.³⁶ Based on these trends, it is not surprising that the mean age of our population was 62.4 ± 22.7 years and did not have a large pediatric or young adult contribution.

Cervical injuries are increasing in the elderly

Our study demonstrated that cervical injuries are increasing over time. This is consistent with the current body of literature, which states that increasing age is associated with an increased risk of cervical spine fracture.²¹ For example, census data report that in the last 100 years (from 1920 to 2020), there has been an almost 1000% increase in the world population.³⁷ In addition, in 2020 alone, one-sixth of the people were reported to be over the age of 65 years old, consistent with findings from the R. Adams Cowley Shock Trauma Center illustrating an increase in patient age among individuals with CSIs.³⁸ Suspected reasons for fractures in this population are likely due to diminishing bone quality and increased falls risk due to multiple medical comorbidities such as poor vision, loss of proprioception, diabetes, syncope, and medication side effects within the elderly population. Understanding this trend opens the door for changes both in the prevention of CSIs and in the rapid management of CSIs. These could take the form of changing safety standards to prevent falls in the elderly or rapid triage of these patients in an emergent setting to foster early detection of possible spinal cord injuries.

Current evidence supports new definitions of upper versus lower injury

Currently, there is a lack of consensus regarding the classification of cervical injuries as upper versus lower. Traditional grading systems considered upper injuries to include the C1–C2 vertebral levels.³⁹ This classification is based on the structural anatomy of the C1 and C2 vertebrae being strikingly different from the morphology of C3–C7. This classification, however, is not based on differing mechanisms or long-term outcomes. Therefore, the utility of older classification systems has been called into question. More recently, there have been suggestions that upper and cervical levels should be redefined as upper as C1–C3 and lower as C4–C7.¹⁴ This is based on differences in demographics, mechanisms of injury, and outcomes seen across vertebral levels. The current study found evidence to support recent suggestions to include C3 in the classification of upper cervical injuries based on statistical methods using K-means clustering. Such findings have implications regarding our current understanding of the biomechanical forces occurring in the cervical spine. Although the C1 and C2 vertebrae are traditionally believed to be the most mobile vertebrae in the cervical spine, research that demonstrates flexion and extension is often limited at C4 and below.³⁹

Being older, White, female, and being admitted correlate with upper injuries, but being younger, Black, male, and being discharged correlate with lower injuries

When upper cervical levels were defined as C1–C3 and lower injuries were defined as C4–C7, our results demonstrated there was a clear demarcation related to age, gender, race, and disposition. Upper-level injuries were more commonly seen in older groups, whereas lower-level injuries were more associated with younger age. These results confirm our previous findings.¹⁴ Both the aforementioned study and the current study are at odds with some prior work that suggested upper-level injuries result solely from C1 to C2 injuries.⁴⁰

Upper injuries were also more associated with being female, White, and being admitted or transferred to another facility. Lower cervical injuries conversely were more associated with non-White men and being discharged from the ED. Given currently available research findings, it is challenging to determine whether these differences in disposition were related to injury characteristics or racial disparities. This is due to the lack of representation of marginalized populations in spinal trauma research as well as in the current database. For example, only 7.4% of patients with CSI in the current database were classified as Black, which is much lower than prior national averages. Additionally, a retrospective cohort study involving 753 facilities in the year 2017 for pediatric patients found the highest rates of spine injuries were in Black patients (49.93% of those with a cervical/thoracic spine injury). Non-Hispanic White patients were second to that at 37.39%. About 34.29% and 38.71% rates were seen among Asian and Hispanic groups, respectively.⁴¹ This is especially relevant that recent research has demonstrated pediatric Black patients tend to have more severe injuries as measured by injury severity scores.⁴² Such racial disparities are not only seen regarding representation in epidemiological studies focused on cervical trauma but in short- and long-term outcomes as well for spine surgery overall. A review encompassing 22 articles found that African American patients were found to have higher odds of nonoperative treatment for SCI and have higher odds of postoperative complications (mortality, cerebrospinal fluid leak, nervous system complications, bleeding, and infection) after spinal surgery.⁴³

Limitations

The limitations of this study are that the NEISS database aggregates data regarding injuries in patients presenting to EDs in approximately 100 hospitals selected as a probability sample covering the United States. As mentioned, one must be cautious when interpreting the generalizability of these results to the general population, given the current database may not capture all patients with limited access to care. In addition, the current database excludes injuries involving alcohol, drugs, firearms, no mentioned product, or intentionally inflicted injuries.⁴⁴ Therefore, some CSIs may not be captured in the NEISS. Our search terms favored all cervical injuries and did not only focus on injuries to the spinal cord itself. The data may also be influenced by changes in reporting efficiency over time within the NEISS database. In our analysis, we opted to group patients into upper versus lower CSIs, excluding a population of individuals with injuries across this artificial boundary. While this approach helped simplify our analysis, it leaves room for future study. Finally, the methodology of the current study does not account for population migrations, and it cannot conclude causation, risk directionality, or temporality for its outcomes.

Conclusions

Although CSCIs specifically (and not CSIs in general) continue to be well studied, a substantial amount of the research corresponding generally to CSIs (which would include fractures, neurological injuries, and muscular injuries such as cervical strains) is from over a decade in the past and may already be outdated.^21,22,45 The current study sought to update CSI data regarding demographics and trends in frequency based on demographics and provide new evidence describing the frequency of different vertebral levels of involvement.

In support of prior work using the NACTN for Spinal Cord Injury registry, we found that not only are cervical injuries increasing over time, but they are increasingly observed in older women due to falls (especially upper cervical injuries).^14,15 In addition, using a more robust clustering methodology compared with prior studies using the NACTN database, we found a demarcation of upper cervical injuries being defined as C1–C3 and lower as C4–C7. Unlike in previous work, we also found that men have a more bimodal age distribution, with upper cervical injuries also occurring in older groups but lower cervical injuries occurring in younger patients. Finally, we observed an important new demographic trend that upper cervical injuries are correlated with being older, White, and female. These patients are more likely to be evaluated and then subsequently admitted or transferred, whereas lower cervical injuries seen in patients that are either younger, Black, male are more likely to be discharged from the ED. These results are especially relevant given that they may lead to research investigating improving treatment protocols to better recognize injuries in the aging population and to provide more just and equitable care across different demographic groups.

Transparency, Rigor, and Reproducibility Statement

This study was not formally registered as it is a retrospective analysis of an existing database. Data for this study were extracted from the National Electronic Injury Surveillance System (NEISS) database, covering the period from 2002 to 2022. NEISS is a publicly available dataset including anonymized patient records and data such as patient demographics, mechanism of injury, and disposition. The analysis plan was not formally preregistered, but all code and raw data can be found in this public GitHub repository: https://github.com/AndreasSeas/NEISS_CSI. As this was a retrospective analysis, no statistical power analysis was performed, and all data were analyzed at the same time. Relevant statistical tests and corrections were performed according to standard statistical practice and are all available in the GitHub repository. Multiple comparisons were corrected via the Bonferroni correction. The authors agree to provide the full content of the article on request by contacting the corresponding author.

Authors’ Contributions

B.G.F.: Conceptualization, methodology, formal analysis, data curation, and writing—original draft; A.S.: Conceptualization, methodology, software, validation, formal analysis, data curation, writing—original draft, and visualization; F.O.U.: Writing—original draft and writing—review and editing; S.K.: Writing—original draft and writing—review and editing; J.K.: Formal analysis and writing—review and editing; T.W.: Writing—review and editing; J.D.G.: Writing—review and editing; M.G.F.: Writing—review and editing; M.M.A.: Conceptualization, methodology, writing—review and editing, supervision, and project administration; NF: Conceptualization, methodology, writing—review and editing, supervision, and project administration.

Author Disclosure Statement

No authors report conflict of interest for this study.

Funding Information

No funding was used to perform this study.

Supplementary Material

Supplementary Table S1