ABSTRACT
Background
The case‐mix and inpatient rehabilitation costs of traumatic cervical spinal cord injury (TCSCI) may have shifted with population aging and the coronavirus disease 2019 (COVID‐19) pandemic. We examined decade‐long trends in patient characteristics, injury mechanisms, and per‐diem rehabilitation costs for neurological‐level C4/C5 TCSCI.
Methods
All patients with C4/C5 TCSCI admitted to the China Rehabilitation Research Center were studied retrospectively, comparing 2013–2017 with 2018–2022, and, in an exploratory analysis, the pre‐pandemic (2018–2019) with the pandemic (2020–2022) period. Costs were adjusted for inflation using the medical services Consumer Price Index. Comparisons used the χ 2, Wilcoxon rank‐sum, and Kruskal–Wallis tests. Cost drivers were modeled using multivariable Gamma generalized linear models (log link).
Results
Among 1063 patients (male‐to‐female ratio 5.3:1), mean age rose from 43.9 to 49.5 years and the proportion injured by low‐level falls increased from 20.7% to 33.8% (χ 2 = 37.76, p < 0.001). Length of stay decreased (Z = −2.33, p = 0.02) while adjusted daily cost increased (Z = −10.07, p < 0.001; median 1330 RMB, interquartile range 908–1780, [USD 215]). In constant 2013 prices a real increase persisted (+12.5%, p < 0.001), and the later period remained independently associated with higher daily cost after case‐mix adjustment (adjusted cost ratio 1.15, 95% CI 1.07–1.24).
Conclusions
Admitted patients were increasingly older and frequently injured in low‐level falls. Per‐diem rehabilitation costs rose despite shorter hospital stays. Changes spanning the pandemic represent an exploratory temporal association rather than causal evidence. Fall prevention strategies for older adults and capacity planning for higher‐dependency rehabilitation are warranted.
Keywords: aging, cervical vertebrae, COVID‐19, health care costs, rehabilitation, spinal cord injuries
Trends in length of stay and inflation‐adjusted daily hospitalization cost in patients with C4/C5 TCSCI, China Rehabilitation Research Center, 2013–2022 (n = 1063).
Summary
A single‐center, 10‐year cohort (n = 1063) characterizes the changing case‐mix, not the population incidence, of C4/C5 traumatic cervical spinal cord injury in China.
Admitted patients aged progressively over the decade (mean 43.9 to 49.5 years), and low‐level falls rose (20.7% to 33.8%) as motor‐vehicle collisions declined.
Inflation‐adjusted daily rehabilitation cost increased despite shorter stays, and the increase persisted and strengthened after multivariable case‐mix adjustment (adjusted cost ratio 1.15). AIS grade was the dominant cost driver, and the increase began before the pandemic, which is consistent with a secular trend rather than a causal COVID‐19 effect.
The shift toward older, more dependent, fall‐related injuries argues for geriatric‐competent, respiratory‐focused rehabilitation capacity and targeted fall‐prevention.
1. Introduction
Traumatic spinal cord injury (TSCI) is a devastating global condition that imposes a substantial societal and familial burden [1]. In the United States, the estimated incidence of TSCI is 48.4 cases per million [2]. In a large Italian population‐based study, the reported incidence was approximately 26.5 cases per million [3]. In China, the largest developing country, the pooled TSCI incidence is estimated at 65.15 cases per million [4], with cervical spinal cord injury (CSCI) accounting for 37.4%–82.0% of these cases [5]. Notably, recent data indicates an increasing trend in its incidence [6]. As the most severe form of TSCI, the increase in cervical injuries necessitates urgent attention.
Injuries to the middle cervical spine (C4–C5) are the most prevalent form of CSCI [7], displaying distinct functional characteristics. While C4 injuries typically result in complete dependence on activities of daily living and often require ventilatory support, C5 injuries usually preserve elbow flexion, potentially allowing limited self‐care. This functional distinction differs substantially from that of C6 injuries, in which wrist extension allows for greater independence. However, comprehensive epidemiological data on C4/C5 traumatic cervical spinal cord injury (TCSCI) remain insufficient. Specifically, longitudinal analyses of the economic burden and epidemiological trends are rare [8].
Using 10 years of hospital records (2013–2022) from the CRRC, China's largest specialist SCI rehabilitation center, we conducted a retrospective cohort study with three objectives: (1) to describe demographic and injury‐mechanism characteristics of patients admitted with neurological level of injury (NLI) at C4 or C5; (2) to quantify changes in inpatient rehabilitation length of stay (LOS) and inflation‐adjusted costs over two 5‐year periods (2013–2017 vs. 2018–2022); and (3) to compare, in an exploratory temporal analysis, the pandemic period (2020–2022) with the immediately preceding pre‐pandemic period (2018–2019). We hypothesized that the cohort would show progressive aging, a shift from motor vehicle collisions (MVCs) toward low‐level falls, and rising per‐diem costs.
2. Materials and Methods
2.1. Study Population
This retrospective cohort study reviewed the electronic medical records of all patients with TCSCI treated at the CRRC (affiliated with Beijing Bo'ai Hospital, a national‐level tertiary rehabilitation center) between January 1, 2013, and December 31, 2022. The reporting followed the STROBE checklist for cohort studies. No a priori sample size calculation was performed and no random sampling was applied; all patients meeting the eligibility criteria and admitted during the 10‐year window were enrolled consecutively, so the analysis is a complete enumeration of eligible admissions at this center and the achieved sample size of 1063 was determined by the size of the source population. There was no prospective follow‐up, and all variables were obtained from the index rehabilitation admission records. Initially, individuals diagnosed with CSCI (ICD‐10: S14.101) were identified from discharge records. Patients whose NLI on admission to CRRC—corresponding to their neurological status on discharge from acute care—was C4 or C5, were selected for inclusion. The NLI was determined based on International Standard for Neurological Classification of Spinal Cord Injury. NLI was reassessed at rehabilitation discharge to capture neurological recovery, although the admission NLI was used for the cohort definition.
Exclusion criteria were: (1) non‐traumatic injuries, (2) Hospital LOS less than 7 days or longer than 1 year, (3) age ≤ 12 years (owing to difficulties in accurately determining NLI in this age group), (4) Incomplete or unclear records, and (5) concomitant brain injury or consciousness disorders. LOS thresholds were operationalized as follows: (1) LOS < 7 days, reflecting truncated rehabilitation engagement or premature care discontinuation, and (2) LOS > 1 year, capturing outliers with compounded medical complexities (e.g., pressure injury‐related admissions) or systemic care coordination failures. These exclusions mitigated bias in estimating TCSCI‐specific rehabilitation efficiency metrics and direct medical costs.
This study was approved by the CRRC Ethics Committee (No. CRRC‐IEC‐RF‐SC‐002‐01) and was conducted in accordance with the Declaration of Helsinki. Because of its retrospective design and the use of fully anonymized records, the requirement for written informed consent was waived by the CRRC Ethics Committee, and all personal identifiers were removed before analysis.
2.2. Data Collection and Grouping
Demographic data (sex, age, occupation, and etiology) were extracted from the International SCI Core Dataset. Hospital data comprised the NLI and AIS grade as determined at admission to the CRRC; these admission values, and not the discharge reassessment, were used for cohort definition, all descriptive analyses, and for the multivariable cost models. LOS and daily and total costs were also recorded. To analyze temporal trends, 10‐year data (January 1, 2013, to December 31, 2022) were stratified into two 5‐year cohorts: the Early Group (2013–2017) and the Late Group (2018–2022). Age was categorized into six groups: ≤ 15, 16–30, 31–45, 46–60, 61–75, and ≥ 76 years. Occupations were classified into categories, including manual laborers, students, government employees/civil servants, self‐employed, office clerks, and retirees. Etiology was classified as: Motor vehicle collisions, Falls (subdivided into high falls [≥ 1 m] and low‐level falls [< 1 m]), sports/leisure, being struck by objects, assault, and machinery‐related injuries. NLI was determined according to the 2011 AIS standards [9] and assessed by three professionals.
2.3. Economic Standardization
To eliminate the confounding factor of inflation over the 10‐year period, hospitalization costs were adjusted using the Consumer Price Index for Medical Services obtained from the National Bureau of Statistics of China. This index was chosen over the general Consumer Price Index because it tracks the administered prices for hospital services, medicines, and consumables, which are the components that dominate inpatient rehabilitation charges. It is published as a year‐on‐year series, so the formula below corrects for differences in annual price movement relative to the 2013 index year rather than for the full cumulative change in the price level; the values reported in Tables 2 and 3, therefore, remain close to nominal charges. To quantify the cumulative price effect, we additionally chained the annual indices to express costs in constant 2013 prices, which we report as a sensitivity analysis and on which the multivariable cost models were fitted. Hospitalization costs comprised all itemized inpatient charges recorded in the hospital information system: bed, consultation, and registration fees; nursing and nursing treatment fees; general, special, physical, rehabilitation, traditional Chinese medicine, interventional, surgical, and anesthesia treatment fees; monitoring and assisted ventilation, oxygen, laboratory, pathology, radiology, ultrasound, nuclide, and other examination fees; Western, patent and herbal medicines, antibiotics, blood products, and biologicals; and single‐use materials for examination, treatment, intervention, and surgery, together with the patient out‐of‐pocket portion. Privately hired caregiver costs and post‐discharge expenditures were not captured. Costs are reported primarily in RMB, with USD shown for reference only at the fixed 2013 rate (1 USD = 6.1932 RMB). This benchmark conversion is not intended to imply current USD equivalence. We standardized all RMB‐denominated hospitalization costs to the 2013 benchmark value (1 USD = 6.1932 RMB) using the China Consumer Price Index for Medical Services (standardized cost = raw cost × (medical service price index for the benchmark year ÷ medical service price index for the current year)).
Table 2.
Trends in Hospitalization Data of Patients with TCSCI over a 10‐year period.
| Items | Overall | 2013–2017 | 2018–2022 | χ 2/t/Z | p‐value |
|---|---|---|---|---|---|
| Neurological level of injury | 0.66 | 0.23 | |||
| C4 | 725 (68.2) | 451 (67.3) | 274 (69.7) | ||
| C5 | 338 (31.8) | 219 (32.7) | 119 (30.3) | ||
| AIS grade | 13.09 | 0.004 | |||
| A | 377 (35.5) | 261 (39.0) | 116 (29.5) | ||
| B | 202 (19.0) | 131 (19.6) | 71 (18.1) | ||
| C | 250 (23.5) | 142 (21.2) | 108 (27.5) | ||
| D | 234 (22.0) | 136 (20.3) | 98 (24.9) | ||
| LOS, day | 64 (34, 96) | 68.8 (36, 99) | 60 (33, 92) | −2.33 | 0.02 |
| Adjust daily cost (RMB, thousand yuan) | 1.33 (0.91, 1.78) | 1.18 (0.75, 1.59) | 1.57 (1.20, 2.02) | −10.07 | < 0.001 |
| Adjust total cost (RMB, thousand yuan) | 80.01 (37.87, 134.62) | 73.19 (34.71, 125.016) | 90.95 (47.03, 152.69) | −4.03 | < 0.001 |
Note: Data are presented as n (%) or mean (IQR).
Abbreviations: AIS, American Spinal Injury Association Impairment Scale; LOS, length of stay.
Table 3.
Exploratory temporal comparison of case‐mix and hospitalization economics across pre‐pandemic and pandemic periods in patients with C4/C5 traumatic cervical spinal cord injury.
| Variables | Total (n = 1063) | E (n = 670) | L1 (n = 175) | L2 (n = 218) | Statistic | p |
|---|---|---|---|---|---|---|
| Age group, years | χ 2 = 38.41 | < 0.001 | ||||
| ≤ 15 | 7 (0.66) | 5 (0.75)a | 0 (0.00)a | 2 (0.92)a | ||
| 16–30 | 216 (20.32) | 155 (23.13)a | 30 (17.14)a,b | 31 (14.22)b | ||
| 31–45 | 251 (23.61) | 177 (26.42)a | 29 (16.57)b | 45 (20.64)a,b | ||
| 46–60 | 394 (37.06) | 241 (35.97)a | 73 (41.71)a | 80 (36.70)a | ||
| 61–75 | 169 (15.90) | 80 (11.94)a | 37 (21.14)b | 52 (23.85)b | ||
| ≥ 76 | 26 (2.45) | 12 (1.79)a | 6 (3.43)a | 8 (3.67)a | ||
| Occupation | χ 2 = 57.89 | < 0.001 | ||||
| Workers | 198 (18.63) | 139 (20.75)a | 25 (14.29)a | 34 (15.60)a | ||
| Students | 40 (3.76) | 26 (3.88)a | 8 (4.57)a | 6 (2.75)a | ||
| National civil servants | 45 (4.23) | 32 (4.78)a | 5 (2.86)a | 8 (3.67)a | ||
| Self‐employed | 49 (4.61) | 36 (5.37)a | 6 (3.43)a | 7 (3.21)a | ||
| Office clerks | 84 (7.90) | 45 (6.72)a | 23 (13.14)b | 16 (7.34)a,b | ||
| Retired | 122 (11.48) | 54 (8.06)a | 27 (15.43)b | 41 (18.81)b | ||
| The unemployed | 178 (16.75) | 118 (17.61)a | 34 (19.43)a | 26 (11.93)a | ||
| Freelance | 23 (2.16) | 12 (1.79)a | 3 (1.71)a | 8 (3.67)a | ||
| Enterprise managers | 15 (1.41) | 12 (1.79)a | 1 (0.57)a | 2 (0.92)a | ||
| Professional and technical personnel | 50 (4.70) | 27 (4.03)a | 12 (6.86)a | 11 (5.05)a | ||
| Farmers | 130 (12.23) | 85 (12.69)a | 21 (12.00)a | 24 (11.01)a | ||
| Soldiers | 9 (0.85) | 7 (1.04)a | 1 (0.57)b | 1 (0.46)a | ||
| Unrecorded | 120 (11.29) | 77 (11.49)a | 9 (5.14)a | 34 (15.60)a | ||
| Aetiology | χ 2 = 71.33 | < 0.001 | ||||
| Motor vehicle collision | 455 (42.80) | 310 (46.27)a | 66 (37.71)a,b | 79 (36.24)b | ||
| High fall | 170 (15.99) | 120 (17.91)a | 25 (14.29)a | 25 (11.47)a | ||
| Low fall | 272 (25.59) | 139 (20.75)a | 53 (30.29)b | 80 (36.70)b | ||
| Sports and leisure | 44 (4.14) | 30 (4.48)a | 12 (6.86)a | 2 (0.92)b | ||
| Hit by other object | 68 (6.40) | 46 (6.87)a | 10 (5.71)a | 12 (5.50)a | ||
| Assault | 37 (3.48) | 14 (2.09)a | 3 (1.71)a | 20 (9.17)b | ||
| Machine induced injury | 8 (0.75) | 6 (0.90)a | 2 (1.14)a | 0 (0.00)a | ||
| Unrecorded | 9 (0.85) | 5 (0.75)a | 4 (2.29)a | 0 (0.00)a | ||
| AIS grade | 14.34 | 0.026 | ||||
| A | 377 (35.47) | 261 (38.96)a | 55 (31.43)a,b | 61 (27.98)b | ||
| B | 202 (19.00) | 131 (19.55)a | 30 (17.14)a | 41 (18.81)a | ||
| C | 250 (23.52) | 142 (21.19)a | 50 (28.57)a | 58 (26.61)a | ||
| D | 234 (22.01) | 136 (20.30)a | 40 (22.86)a | 58 (26.61)a | ||
| LOS, day | 64.00 (34.00, 96.00) | 68.00 (36.00, 98.75)a | 63.00 (32.50, 86.50)a | 59.00 (33.25, 98.00)a | 6.14 | 0.046 |
| Daily cost, (RMB, yuan) | 1344.85 (912.16, 1794.24) | 1201.16 (758.85, 1621.19)a | 1539.16 (1150.32, 1933.93)b | 1587.93 (1230.15, 2041.74)b | 92.26 | < 0.001 |
Note: Data are presented as n (%) or mean (Q1, Q3). Each superscript letter denotes a subset of period categories (E, L1, L2) whose column proportions do not differ significantly from each other at the 0.05 level.
Abbreviations: AIS, American Spinal Injury Association Impairment Scale; LOS, length of stay.
2.4. Statistical Analysis
Statistical analyses and data visualization were primarily performed using R software version 4.5.3 (R Foundation for Statistical Computing, Vienna, Austria). The Joinpoint Regression Program version 5.4.0 (Statistical Research and Applications Branch, National Cancer Institute) was specifically employed to analyze temporal trends and identify inflection points in injury etiology (motor vehicle collisions and falls). Normality of the data distribution was assessed using the Kolmogorov–Smirnov test. Continuous variables are presented as means ± standard deviations (SD) for normally distributed data, or as medians with interquartile ranges (IQR) for non‐normally distributed variables (specifically length of stay and hospitalization costs). Categorical variables are expressed as frequencies and percentages. To compare differences between the two 5‐year cohorts (2013–2017 vs. 2018–2022), the independent samples t‐test was used for normally distributed continuous variables, whereas the Wilcoxon rank‐sum test was applied to non‐normally distributed data. Differences in categorical variables were analyzed using Pearson's χ 2 test. Furthermore, to compare the three time periods spanning the pandemic in an exploratory analysis, the non‐parametric Kruskal–Wallis test was used for continuous comparisons and the χ 2 test for categorical variables. For three‐group comparisons (Early Period (E group): January 1, 2013–December 31, 2017; Late Pre‐COVID Period (L1 group): January 1, 2018–December 31, 2019; Pandemic Period (L2 group): January 1, 2020–December 31, 2022), post‐hoc pairwise comparisons were performed using Dunn's test with Bonferroni correction when the Kruskal–Wallis test showed significance (p < 0.05). Effect sizes were calculated using eta‐squared (η 2) for continuous variables. All economic data were analyzed using inflation‐adjusted values. Statistical significance was defined as a two‐tailed p < 0.05 throughout. Missing data: Patients with incomplete records (n = 299) were excluded without imputation (Figure 1). Sensitivity analysis: Primary two‐group comparisons for LOS and daily costs were repeated, excluding LOS > 180 days, and the results were consistent. The drivers of inpatient costs were examined with multivariable generalized linear models (Gamma family, log link; glm() function, stats package, R version 4.5.3) for inflation‐adjusted daily cost and total cost, adjusting for age, sex, AIS grade, neurological level (C4 vs. C5), etiology, occupation, and study period, with length of stay additionally entered for total cost. A secondary model replaced the two‐period variable with the three pandemic‐defined phases (E, L1, L2). Log‐transformed linear regression and a model excluding stays longer than 180 days were used as sensitivity analyses, and adjusted effects are reported as cost ratios (exp[β]) with 95% confidence intervals. De‐identified aggregate annual data, a data dictionary, and the analysis code have been deposited in a public repository (DOI to be added on acceptance), and individual‐level data are available from the corresponding author upon reasonable request, subject to institutional data‐governance approval.
Figure 1.
Selection of patients with C4/C5 traumatic cervical spinal cord injury (TCSCI).
3. Results
3.1. Patient Selection
A total of 5732 inpatients diagnosed with CSCI were identified between January 1, 2013, and December 31, 2022. Among them, 2929 patients were excluded because of repeated hospitalizations (for patients with multiple admissions, only the first [index] rehabilitation admission was retained). An additional 542 patients were excluded because of non‐traumatic injuries. Furthermore, 147 patients were excluded because of an inappropriate hospital length of stay. Twenty‐seven patients aged ≤ 12 years were excluded. A total of 299 patients with incomplete medical records were excluded. Additionally, 116 patients were excluded because of consciousness disorders. Of the 1672 patients with traumatic CSCI (C1–C8) who remained after these exclusions, 609 had a neurological level other than C4 or C5 and were therefore not eligible for the present analysis (Figure 1). Ultimately, 1063 patients with NLI at C4 or C5 were included in our analysis (Figure 1).
3.2. Demographic Characteristics
Among the 1063 included patients with neurological‐level C4/C5 TCSCI, the overall mean age was 46.0 ± 15.5 years. Significant differences were observed between the two time periods (χ 2 = 34.49, p < 0.001), indicating a gradual aging trend (Table 1). Occupational distribution also differed significantly between groups (χ 2 = 36.45, p < 0.001) (Table 1). As shown in Figure 2, the proportion of patients aged 61–75 years increased, while the proportion of younger patients (16–30 and 31–45 years) decreased (p < 0.001). Notably, after 2020, the curves for the 46–60 and 61–75 age groups demonstrated an upward trend, whereas the 16–30 age group showed a downward trend.
Table 1.
Trends in epidemiological characteristics of traumatic cervical spinal cord injury patients over a 10‐year period.
| Characteristics | Overall | 2013–2017 | 2018–2022 | χ 2/t/Z | p‐value |
|---|---|---|---|---|---|
| Sex | |||||
| Male | 895 (84.2) | 572 (85.4) | 323 (82.2) | 1.88 | 0.169 |
| Female | 168 (15.8) | 98 (14.6) | 70 (17.8) | ||
| Age, years | 46.0 ± 15.5 | 43.9 ± 15.0 | 49.5 ± 15.8 | −5.76 | < 0.001 |
| < 15 | 7 (0.7) | 5 (0.7) | 2 (0.5) | 34.49 | < 0.001 |
| 16–30 | 216 (20.3) | 155 (23.1) | 61 (15.5) | ||
| 31–45 | 251 (23.6) | 177 (26.4) | 74 (18.8) | ||
| 46–60 | 394 (37.1) | 241 (36.1) | 153 (38.9) | ||
| 61–75 | 169 (15.9) | 80 (11.9) | 89 (22.7) | ||
| 76+ | 26 (2.4) | 12 (1.8) | 14 (3.6) | ||
| Occupation | 36.45 | < 0.001 | |||
| Workers | 198 (18.6) | 139 (20.7) | 59 (15.0) | ||
| Students | 40 (3.8) | 26 (3.9) | 14 (3.6) | ||
| National civil servants | 45 (4.2) | 32 (4.8) | 13 (3.3) | ||
| Self‐employed | 49 (4.6) | 36 (5.4) | 13 (3.3) | ||
| Office clerks | 84 (7.9) | 45 (6.7) | 39 (9.9) | ||
| Retired | 122 (11.5) | 54 (8.1) | 68 (17.3) | ||
| The unemployed | 178 (16.8) | 118 (17.6) | 60 (15.3) | ||
| Freelance | 23 (2.2) | 12 (1.8) | 11 (2.8) | ||
| Enterprise managers | 15 (1.4) | 12 (1.8) | 3 (0.8) | ||
| Professional and technical personnel | 50 (4.7) | 27 (4.0) | 23 (5.9) | ||
| Farmers | 130 (12.2) | 85 (12.7) | 45 (11.5) | ||
| Soldiers | 9 (0.8) | 7 (1.0) | 2 (0.5) | ||
| Unrecorded | 120 (11.3) | 77 (11.5) | 43 (10.8) | ||
| Etiology | 37.76 | < 0.001 | |||
| Motor vehicle collision | 455 (42.8) | 310 (46.3) | 145 (36.9) | ||
| High fall | 170 (16.0) | 120 (17.9) | 50 (12.7) | ||
| Low fall | 272 (25.6) | 139 (20.7) | 133 (33.8) | ||
| Sports and leisure | 44 (4.1) | 30 (4.5) | 14 (3.6) | ||
| Hit by other object | 68 (6.4) | 46 (6.9) | 22 (5.6) | ||
| Assault | 37 (3.5) | 14 (2.1) | 23 (5.9) | ||
| Machine induced injury | 8 (0.8) | 6 (0.9) | 2 (0.5) | ||
| Unrecorded | 9 (0.8) | 5 (0.7) | 4 (1.0) |
Note: Data are presented as n (%) or mean (SD).
Figure 2.
Annual percentage distribution of admissions by age group in patients with C4/C5 TCSCI, China Rehabilitation Research Center, 2013–2022 (n = 1063). Bars show the percentage of each year's admissions contributed by each age group. TCSCI, traumatic cervical spinal cord injury.
3.3. Injury Mechanisms and Temporal Trends
MVC remained the leading cause of injury (42.8%), followed by low‐level falls (25.6%) and high falls (16.0%). However, longitudinal analysis revealed a statistically significant decrease in the proportion of admissions due to motor‐vehicle collisions over the decade (χ 2 = 71.33, p < 0.001, Table 1). Time‐trend analysis of MVC showed no significant changes (Figure 3). During 2013–2015, the annual percentage change (APC) was 6.72% (95% CI: −29.20 to 60.86, p = 0.700), which was not statistically significant. From 2015 to 2022, the APC was −6.06% (95% CI: −13.55 to 2.08, p = 0.110), showing a declining trend but not reaching statistical significance. The Average Annual Percentage Change (AAPC) for the entire study period was −3.36% (95% CI: −11.25 to 5.24; the 95% CI includes zero, indicating no statistically significant overall trend), suggesting that the overall admission proportion attributable to motor vehicle collisions remained relatively stable. Time‐trend analysis revealed a significant upward trend in the proportion of admissions due to low‐level falls. During 2013–2015, the APC was 33.42% (95% CI: −26.02 to 140.64, p = 0.264), failing to reach statistical significance. From 2015 to 2022, the APC was 13.27% (95% CI: 5.72 to 21.36, p = 0.006), indicating a significant upward trend. The AAPC over the entire study period was 17.47% (95% CI: 5.45 to 30.86, p = 0.004), indicating a sustained significant increase in the proportion of low‐level‐fall admissions. The apparent discrepancy between the significant categorical decline in motor vehicle collision admissions and the non‐significant Joinpoint trend reflects the different questions addressed by the two analyses: The χ 2 test compares aggregated proportions between multi‐year periods, whereas the Joinpoint model tests the year‐by‐year slope, which was estimated with wide confidence intervals because of small annual counts (see Figure 3).
Figure 3.
Annual proportion of admissions attributable to motor vehicle collisions and low‐level falls in patients with C4/C5 TCSCI, China Rehabilitation Research Center, 2013–2022 (n = 1063). Values are the percentage of each year's admissions. TCSCI, traumatic cervical spinal cord injury.
3.4. Length of Stay and Hospitalization Costs
Over the 10‐year period, the median LOS was 64.00 days (IQR: 34.00, 96.00) overall, with a statistically significant decrease between the two periods (68.00 vs. 60.00 days; Z = −2.33, p = 0.02) (Table 2). The inflation‐adjusted median daily cost was 1330 RMB (IQR: 908, 1780; USD 215). However, the inflation‐adjusted daily hospitalization cost showed a significant increasing trend (1179 vs. 1567 RMB; Z = −10.07, p < 0.001, Figure 4). Because the index used for the primary adjustment is published as a year‐on‐year series, we repeated the comparison after chaining the annual indices to express all costs in constant 2013 prices. Cumulative medical‐service price inflation over the decade was 26.2%; after removing it, the median daily cost still rose from 1124 RMB in 2013–2017 to 1265 RMB in 2018–2022, a real increase of 12.5% (Z = −4.86, p < 0.001) (Table S3). Therefore, the increase in per‐diem costs exceeds what would be expected from medical service price inflation alone.
Figure 4.
Trends in length of stay and inflation‐adjusted daily hospitalization cost in patients with C4/C5 TCSCI, China Rehabilitation Research Center, 2013–2022 (n = 1063). Lines show the annual median and shaded areas the interquartile range: length of stay in days (left axis) and adjusted daily cost (right axis). Costs were standardized to 2013 values using the Consumer Price Index for Medical Services. Monotonic trends were tested with the Mann–Kendall test: length of stay τ = −0.84, Z = −3.31, p < 0.001; daily cost τ = 0.69, Z = 2.68, p = 0.007. IQR, interquartile range; TCSCI, traumatic cervical spinal cord injury; USD, United States Dollar.
In multivariable generalized linear models (Gamma family, log link) adjusting for age, sex, AIS grade, neurological level, etiology and occupation, admission in the later period (2018–2022) remained independently associated with a higher inflation‐adjusted daily cost (adjusted cost ratio [aCR] 1.15, 95% CI 1.07–1.24, p < 0.001); this exceeded the unadjusted estimate (cost ratio 1.10), indicating that case‐mix differences did not account for the increase. For total cost, the later period did not differ before adjustment (cost ratio 1.01, p = 0.83) but became significantly higher once length of stay was included (aCR 1.15, 95% CI 1.06–1.25, p < 0.001), showing that the shorter admissions of the later period had masked a higher per‐admission cost intensity. AIS grade was the dominant case‐mix driver of cost (grade D vs. A: aCR 0.63 for daily and 0.57 for total cost, both p < 0.001), and length of stay strongly predicted the total cost (aCR 1.012 per additional day, p < 0.001) (Table S3). When the two‐period variable was replaced by the three pandemic‐defined phases, daily cost was elevated in both the pre‐pandemic phase (L1, 2018–2019: aCR 1.13, 95% CI 1.02–1.25) and the pandemic phase (L2, 2020–2022: aCR 1.17, 95% CI 1.07–1.28) relative to 2013–2017, indicating that the rise began before the pandemic and continued through it. The findings were materially unchanged in the log‐transformed linear models and after excluding stays longer than 180 days (Table S1).
3.5. Exploratory Temporal Comparison Across Pandemic Periods
To describe how the case‐mix and costs evolved across the pandemic period, the cohort was divided into three consecutive time periods for an exploratory temporal comparison, which is descriptive and not intended to estimate the causal effect of COVID‐19. The proportion of patients aged over 61 years was higher in the L2 group (χ 2 = 38.41, p < 0.001). The retired individuals were more commonly seen in the L1 and L2 groups (χ 2 = 57.89, p < 0.001). The L2 group had the lowest proportion of severe injuries (χ 2 = 14.34, p = 0.026; AIS = A), the shortest median LOS (χ 2 = 6.14, p = 0.046) and the highest daily hospitalization costs (χ 2 = 92.26, p < 0.001, Table 3).
4. Discussion
This study reports the decade‐long shift in the epidemiology and economic burden of cervical spinal cord injury. Time and economic development have not only altered people's lifestyles and work patterns but have also transformed CSCI incidence patterns. The economic burden of C4/C5 injuries is intrinsically linked to their specific neurological outcomes. Unlike C6 injuries, where wrist extension is preserved, facilitating some independence in transfers and feeding, patients with C4/C5 injuries are often completely dependent. Furthermore, C4/C5 involvement often compromises the diaphragmatic function, necessitating intensive respiratory management and higher nursing‐care intensity. This high dependency likely explains our study's sustained high daily costs. Considering the incidence of CSCI, research on this unique level remains relatively scarce. Studying this specific subgroup offers valuable clinical insights into rehabilitation needs and economic impact.
Studies in the United States have shown that between 2015 and 2019, the average age of TSCI patients rose from 45 to 51.6 years, compared to data from 2008 [10]. An Australian study reported a 5% annual increase in the incidence of TSCI in elderly individuals [11]. Similarly, in Canada, the proportion of TSCI patients aged 65 or older increased from 22% in 2005 to 38% in 2019. By 2019, 66% of cases had occurred in individuals over 45 years of age, a reversal of the trend from 2005, when most cases occurred in younger adults [12].
Studies in China have also reported an increasing incidence of TCSCI among elderly patients, prompting national policies and prevention research [13]. Our findings corroborate these trends, demonstrating an increasing age at the onset of C4/C5 TCSCI and an increasing proportion of elderly individuals in the injured cohort.
In terms of etiology, previous studies have shown a decline in the incidence of TSCI caused by MVC, while the proportion of injuries due to falls has been rising [13, 14], especially among elderly individuals [15].
The reason for this observation can be that, with the gradual improvement and strengthened enforcement of national traffic policies, injuries caused by motor vehicle collisions have decreased. It can also be that the proportion of elderly patients with TCSCI has been increasing. This aligns with the rising incidence of falls among the elderly, who may experience falls due to physical factors such as sarcopenia and diminished balance function. Consequently, future preventive strategies should emphasize stability training for older adults to prevent falls.
Caution is warranted when comparing cost figures across studies, as most previous publications have reported costs for all SCI levels combined during acute care, whereas our study exclusively captured neurological‐level C4/C5 patients during stable inpatient rehabilitation. With this caveat, a 2024 study from China found that among 10,945 patients with acute TSCI (all levels), the average total cost was 71,300 RMB (USD 11,500) and the average daily cost was 4400 RMB (USD 700). In that cohort, total acute TSCI cost decreased over the study period (−4.7%), while the daily cost did not significantly change (1.0% increase) [16].
Combined with the acute care spending reported in previous studies on a large Chinese cohort [17], the cost would be nearly USD 90,000 per year for patients in China, which is similar to the spending reported in Canada and the Republic of Korea [18, 19]. However, there have been no clear reports on the economic burden of patients with cervical spinal cord injury, which accounted for the largest proportion of costs in prior research. Our study reports economic‐burden trends over a decade for patients with C4/C5 TCSCI, which revealed that the median adjusted daily cost was 1330 RMB (USD 215) and the adjusted total cost was 80,010 RMB (USD 12,919). Our study found that the adjusted daily costs were lower than those reported in previous studies, whereas the total hospitalization costs were higher [16]. The lower daily hospitalization costs in our study may be attributed to our data set comprising patients in stable rehabilitation phases, which inherently incur lower daily costs than those with acute injuries. The increase in total hospitalization costs could stem from prior studies examining total costs across all phases, coupled with higher overall expenses associated with patients with cervical spinal cord injury. Importantly, the costs reported here, inpatient rehabilitation expenditures during the post‐acute stable phase, represent only one component of the lifetime economic burden of high‐cervical SCI. International estimates of lifetime costs exceed USD 1–3 million per patient when considering acute surgical care, long‐term rehabilitation, assistive technology, caregiver burden, and indirect productivity losses. Our~USD 13,000 per rehabilitation admission should not be interpreted as the total economic burden; it provides a rehabilitation phase benchmark for this specific NLI group that was previously absent from the literature. Our findings revealed that over the 10‐year period, the inflammation‐adjusted daily hospitalization cost increased significantly in the second half (2018–2022).
Interestingly, we observed a contradictory trend: while LOS significantly decreased in the post‐pandemic period (Phase L2), the inflation‐adjusted daily hospitalization costs increased. This trend can be attributed to two primary factors. First, demographic shifts played a role, patients admitted between 2018 and 2022 were significantly older (49.5 ± 15.8 years). Elderly patients typically present with more comorbidities and require more intensive daily medical care [17]; Second, the drivers of the residual cost increase cannot be identified from these data. Inpatient charges accumulated over a decade are shaped not only by clinical treatment intensity, but also by revisions to hospital charging structures and service pricing, changes in reimbursement and insurance policies, and shifts in the cost of medicines and consumables. Because our data set records only aggregate daily and total charges rather than itemized fee components, we cannot determine which of these contributed or to what extent [20]. Future research may also be needed to establish a national cervical spinal cord injury burden database that identifies the total annual and average daily costs of CSCI in China.
The temporal rise in inflation‐adjusted per‐diem cost persisted and was accentuated after multivariable adjustment for age, sex, AIS grade, neurological level, etiology, and occupation (adjusted cost ratio 1.15 for daily cost; 1.15 for total cost after adjusting for length of stay), indicating that it was not merely an artifact of the older, more dependent case‐mix admitted in the later period. The dominant measurable cost driver was injury severity (AIS grade), with progressively lower costs across less complete lesions, whereas length of stay dominated the total cost. The divergence between a shortening length of stay and a rising per‐diem cost is compatible with greater daily treatment intensity rather than a changing underlying patient mix. However, the present data cannot demonstrate this: component‐level cost data were not available, and administrative changes in pricing or reimbursement could produce the same pattern, so any attribution to newer rehabilitation technologies, pharmacotherapy, or assistive‐device use would be speculative. Because the increase was already present in the pre‐pandemic phase (2018–2019) and continued thereafter, it is best interpreted as a secular trend rather than a purely pandemic‐related effect. The findings directly supported by this data set are the aging of the admitted cohort, the changing distribution of injury mechanisms, the shortening of inpatient stays, and the higher per‐diem cost that persists after case‐mix adjustment and full price deflation; the mechanisms discussed above are contextual interpretations that were not measured here.
The case‐mix shift we observed, older patients, and a growing share of low‐level‐fall injuries, has direct implications for the long‐term burden of chronic disabilities. Unlike the historically dominant young high‐energy trauma population, older adults with C4/C5 injuries more often sustain functionally devastating lesions superimposed on pre‐existing frailty, sarcopenia, and multimorbidity. These patients face protracted respiratory and ventilatory dependence, higher nursing‐care intensity, and a greater risk of secondary complications, converting an acute injury into a decade‐long care commitment. As population aging accelerates in China, the cumulative prevalence of survivors living with high‐cervical SCI and the associated demand for caregiving, assistive technology, and lifelong follow‐up is likely to grow faster than the incidence figures alone would suggest, a pattern consistent with the wider literature on aging‐related chronic disability.
These findings have implications for rehabilitation‐resource allocation. A patient population that is older, more dependent, and increasingly fall‐related calls for deliberate planning of geriatric‐competent SCI rehabilitation capacity, expanded pulmonary and respiratory rehabilitation services, and fall‐prevention public‐health programs targeting community‐dwelling older adults. In contrast, shorter length of stay alongside higher case‐mix‐adjusted per‐diem cost implies intensifying daily resource use within compressed admissions, with consequences for diagnosis‐related group and bundled‐payment reimbursement design, workforce and bed‐capacity planning, and equitable distribution of specialist rehabilitation beyond a small number of national centers. Embedding case‐mix‐adjusted cost benchmarks, such as those reported here into capacity and reimbursement planning would help health systems anticipate demographic transitions' resource consequences rather than react to them.
Comparing data before and after the pandemic revealed that the decreasing trend in the admission share of the 46–60‐year age group was interrupted after 2020. Furthermore, the proportion of injuries in the 60–75 age group increased more noticeably post‐COVID. One possible explanation is that pandemic‐related restrictions limit the elderly's outdoor mobility, leading to deconditioning, reduced muscle strength, and compromised balance factors that increase home fall risk, although alternative mechanisms cannot be excluded. This finding is also supported by the etiological analysis. From an occupational perspective, the number of retirees among those affected by the pandemic steadily increased. These occupational shifts are descriptive and cannot, by themselves, establish that home‐quarantine policies caused the change; altered referral patterns, acute‐care survival, and patients' willingness or ability to attend rehabilitation may equally account for them. Importantly, these pandemic‐period associations are speculative, as our data set contains no direct information on activity levels, quarantine compliance, or pre‐injury mobility status. The observed demographic shifts could alternatively reflect pandemic‐related changes in referral patterns at the CRRC rather than a true population‐level change in injury incidence, a possibility that cannot be excluded without community‐level surveillance or multicenter registry data. Consequently, future preventive strategies must emphasize home safety and balance training for older individuals, and future research should seek to disentangle pandemic‐specific effects from the underlying secular demographic trends.
Beyond their epidemiological interpretations, changes spanning the pandemic period raise a translational question regarding how rehabilitation should be delivered. The pandemic accelerated interest in telerehabilitation, hybrid in‐person/remote models and community‐ or home‐based service delivery, and our observation of shorter admissions with intensified daily care is consistent with a broader inpatient–outpatient boundary reconfiguration. For a high‐dependency group, such as those with C4/C5 SCI, the translational challenge is to determine which components of rehabilitation can be safely decentralized, how respiratory and caregiver support can be sustained outside the inpatient setting, and how digital pathways can preserve the continuity of care without widening access disparities. Whether these service model shifts represent durable, evidence‐based improvements or transient disruptions is an empirical question that prospective, multicenter studies ideally linked to functional and cost outcomes need to resolve.
4.1. Limitations
This study has some important limitations. First, as this was a single‐center study, our findings may not be representative of all patients with C4/C5 TCSCI in China. The CRRC preferentially attracts patients with severe injuries and adequate resources, potentially excluding those who died during acute care, could not afford rehabilitation, or were treated locally. Regional disparities in healthcare access further limit the results' generalizability. Because the cohort was assembled by complete enumeration of consecutive eligible admissions rather than by random sampling, the results describe the population admitted to this center and should not be extrapolated to the wider Chinese TCSCI population without external validation. Second, retrospective data collection introduces information bias, particularly in injury‐etiology classification (fall height measurements relying on patient recall) and pre‐injury characteristics. Third, the COVID‐19 pandemic restrictions (2020‐2022) may have systematically biased our cohort through selective patient referrals and altered hospital protocols, confounding the true epidemiological trends with healthcare system disruptions. We cannot definitively separate the pandemic‐related changes from the artifacts of reduced data quality. Fourth, unmeasured confounders including comorbidities, insurance type, geographic origin, and injury‐to‐admission intervals were not systematically captured. This bias's probable direction is toward overestimating per‐diem cost increases in the later period, given confirmed aging of the cohort; this effect is likely modest relative to the robust cost difference (Z = −10.07, p < 0.001). Patients excluded because of incomplete records (n = 299) may have differed systematically from those retained, which we were unable to quantify. These patients were excluded precisely because one or more of the core study variables were missing from their records, and those records were not carried forward into the analytic data set. Therefore, a baseline comparison of the variables of interest cannot be reconstructed retrospectively, and we report this as an unresolved limitation rather than one that we are able to bound. Selection bias from this source cannot be excluded, and its direction remains undetermined. We have correspondingly avoided any inference that depends on the excluded patients resembling those retained; the case‐mix and cost findings rest on multivariable adjustment within the analyzed cohort, and the cost results were robust to log‐transformed modeling and exclusion of extreme lengths of stay. In addition, our cost data comprise aggregate daily and total inpatient charges without itemized fee components; therefore, the relative contributions of clinical intensity, service pricing, reimbursement policy, and consumable costs to the observed cost increase cannot be disentangled. Fifth, our economic assessment captured only direct inpatient rehabilitation costs and substantially underestimated the total burden by excluding acute care expenses, post‐discharge costs, indirect costs (lost productivity and caregiver burden), and intangible costs. International studies suggest that the lifetime costs for high cervical SCI exceed USD 1–3 million, compared to our reported approximately USD 13,000 rehabilitation hospitalization costs. Finally, exclusion criteria (LOS 7 days–1 year) introduced survivor bias, and lack of functional outcome measures limits rehabilitation‐effectiveness assessment. Future prospective multicenter studies with comprehensive cost capture are required.
5. Conclusion
Over the past decade, the mean age of patients with neurological‐level C4/C5 TCSCI rose significantly (43.9–49.5 years), the proportion injured by low‐level falls increased substantially (20.7%–33.8%), and inflation‐adjusted per‐diem inpatient rehabilitation costs rose despite shorter hospital stays. These findings are consistent with an aging‐related shift in injury mechanisms, and the concurrent changes observed during the COVID‐19 period are reported as an exploratory temporal association rather than evidence of a causal pandemic effect. We recommend that policymakers implement fall‐prevention programs targeting older adults, and adapt rehabilitation capacities accordingly.
Author Contributions
Conceptualization: Juan Li, Wei Gao, Lianjun Gao. Methodology and formal analysis: Feng Yang, Wei Gao, Lianjun Gao. Neurological (AIS grade) verification: Feng Yang, Wei Gao, Lianjun Gao. Data curation: Juan Li, Yiming Zhang, Xuan Gao, Qing Wang. Writing–original draft: Juan Li, Lianjun Gao. Writing–review and editing: Lianjun Gao, Wei Gao. Supervision and oversight of data integrity: Juan Li, Lianjun Gao. All authors read and approved the final version of the manuscript and agree to be accountable for all aspects of the work. This work was supported by the China Rehabilitation Research Center Fund (Grant No. 2023ZX‐Q12).
Ethics Statement
This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the China Rehabilitation Research Center (CRRC) (approval number CRRC‐IEC‐RF‐SC‐002‐01). Given the retrospective design and the exclusive use of de‐identified medical records, the requirement for written informed consent was waived by the Ethics Committee. All personal identifiers were removed before analysis and no individually identifiable information is reported.
Consent
The authors have nothing to report.
Conflicts of Interest
The authors declare no conflicts of interest.
Use of Large Language Models, AI and Machine Learning Tools
During the preparation of this work the authors used a large language model‐based assistant (Claude, Anthropic, USA) to support the statistical analysis code and to assist with drafting and language editing of parts of the manuscript. All statistical output was re‐derived from the source data set and verified by the authors, and all text was reviewed and edited by the authors, who take full responsibility for the integrity of the data, the accuracy of the analyses and the content of the publication.
Supporting information
Supporting File
Acknowledgments
We thank Dr. Peng Kun Li for assistance with the statistical analysis.
Data Availability Statement
De‐identified aggregate annual data, the data dictionary and the analysis code supporting the findings of this study have been deposited in a public repository (DOI to be added on acceptance). Individual‐level patient data are not publicly available because of institutional data‐governance and patient‐privacy restrictions, but are available from the corresponding author on reasonable request and subject to approval by the China Rehabilitation Research Center.
References
- 1. Lu Y., Shang Z., Zhang W., et al., “Global Incidence and Characteristics of Spinal Cord Injury Since 2000–2021: A Systematic Review and Meta‐Analysis,” BMC Medicine 22 (2024): 285, 10.1186/s12916-024-03514-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Sun J., Yuan W., Zheng R., et al., “Traumatic Spinal Injury‐Related Hospitalizations in the United States, 2016–2019: A Retrospective Study,” International Journal of Surgery 109, no. 12 (2023): 3827–3835, 10.1097/JS9.0000000000000696. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Barbiellini Amidei C., Salmaso L., Bellio S., and Saia M., “Epidemiology of Traumatic Spinal Cord Injury: A Large Population‐Based Study,” Spinal Cord 60, no. 9 (2022): 812–819, 10.1038/s41393-022-00795-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Hu Y., Li L., Hong B., et al., “Epidemiological Features of Traumatic Spinal Cord Injury in China: A Systematic Review and Meta‐Analysis,” Frontiers in Neurology 14 (2023): 1131791, 10.3389/fneur.2023.1131791. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Chen C., Qiao X., Liu W., Fekete C., and Reinhardt J. D., “Epidemiology of Spinal Cord Injury in China: A Systematic Review of the Chinese and English Literature,” Spinal Cord 60 (2022): 1050–1061, 10.1038/s41393-022-00826-6. [DOI] [PubMed] [Google Scholar]
- 6. Aarabi B., Albrecht J. S., Simard J. M., et al., “Trends in Demographics and Markers of Injury Severity in Traumatic Cervical Spinal Cord Injury,” Journal of Neurotrauma 38 (2021): 756–764, 10.1089/neu.2020.7415. [DOI] [PubMed] [Google Scholar]
- 7. Wu F., Zheng Y., Ren B., Huang L., and Yang D., “Current Epidemiological Profile and Characteristics of Traumatic Cervical Spinal Cord Injury in Nanchang, China,” Journal of Spinal Cord Medicine 45 (2022): 556–563, 10.1080/10790268.2021.1949188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Wang T. Y., Park C., Zhang H., et al., “Management of Acute Traumatic Spinal Cord Injury: A Review of the Literature,” Frontiers in Surgery 8 (2021): 698736, 10.3389/fsurg.2021.698736. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Kirshblum S. C., Burns S. P., Biering‐Sorensen F., et al., “International Standards for Neurological Classification of Spinal Cord Injury (Revised 2011),” Journal of Spinal Cord Medicine 34, no. 6 (2011): 535–546, 10.1179/204577211X13207446293695. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Chen Y., Wen H., Baidwan N. K., and DeVivo M. J., “Demographic and Health Profiles of People Living With Traumatic Spinal Cord Injury in the United States During 2015–2019: Findings From the Spinal Cord Injury Model Systems Database,” Archives of Physical Medicine and Rehabilitation 103, no. 4 (2022): 622–633, 10.1016/j.apmr.2021.11.001. [DOI] [PubMed] [Google Scholar]
- 11. Beck B., Cameron P. A., Braaf S., et al., “Traumatic Spinal Cord Injury in Victoria, 2007–2016,” Medical Journal of Australia 210 (2019): 360–366, 10.5694/mja2.50143. [DOI] [PubMed] [Google Scholar]
- 12. Thorogood N. P., Noonan V. K., Chen X., et al., “Incidence and Prevalence of Traumatic Spinal Cord Injury in Canada Using Health Administrative Data,” Frontiers in Neurology 14 (2023): 1201025, 10.3389/fneur.2023.1201025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Jiang B., Sun D., Sun H., et al., “Prevalence, Incidence, and External Causes of Traumatic Spinal Cord Injury in China: A Nationally Representative Cross‐Sectional Survey,” Frontiers in Neurology 12, no. 2021 (2022): 784647, 10.3389/fneur.2021.784647. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Golestani A., Shobeiri P., Sadeghi‐Naini M., et al., “Epidemiology of Traumatic Spinal Cord Injury in Developing Countries From 2009 to 2020: A Systematic Review and Meta‐Analysis,” Neuroepidemiology 56 (2022): 219–239, 10.1159/000524867. [DOI] [PubMed] [Google Scholar]
- 15. Barriga‐Martín A., Pérez‐Ruiz P., Muñoz‐Rodríguez J. R., et al., “Epidemiology of Traumatic Spinal Cord Injury in Spain: A Ten‐Year Analysis of Trend of Clinical and Demographic Characteristics,” Journal of Spinal Cord Medicine 48 (2025): 884–890, 10.1080/10790268.2024.2375889. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Zhou H., Lou Y., Chen L., et al., “Epidemiological and Clinical Features, Treatment Status, and Economic Burden of Traumatic Spinal Cord Injury in China: A Hospital‐Based Retrospective Study,” Neural Regeneration Research 19 (2024): 1126–1132, 10.4103/1673-5374.382257. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Chan B. C. F., Craven B. C., and Furlan J. C., “A Scoping Review on Health Economics in Neurosurgery for Acute Spine Trauma,” Neurosurgical Focus 44 (2018): E15, 10.3171/2018.2.FOCUS17778. [DOI] [PubMed] [Google Scholar]
- 18. Malekzadeh H., Golpayegani M., Ghodsi Z., et al., “Direct Cost of Illness for Spinal Cord Injury: A Systematic Review,” Global Spine Journal 12 (2022): 1267–1281, 10.1177/21925682211031190. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Bae S. W., Shin H.‐I., Bang M. S., and Lee M.‐Y., “Epidemiology of Work‐Related Traumatic Spinal Cord Injury: An Analysis of Workers' Compensation Claims in Korea, 2011–2019,” Journal of Occupational & Environmental Medicine 65 (2023): e453–e457, 10.1097/JOM.0000000000002852. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Sánchez‐Raya J. and Sampol J., “Spinal Cord Injury and COVID‐19: Some Thoughts After the First Wave,” Spinal Cord 58 (2020): 841–843, 10.1038/s41393-020-0524-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting File
Data Availability Statement
De‐identified aggregate annual data, the data dictionary and the analysis code supporting the findings of this study have been deposited in a public repository (DOI to be added on acceptance). Individual‐level patient data are not publicly available because of institutional data‐governance and patient‐privacy restrictions, but are available from the corresponding author on reasonable request and subject to approval by the China Rehabilitation Research Center.