Abstract
Fever is a frequent complication after traumatic spinal cord injury (TSCI) and may exacerbate secondary neurological damage through increased metabolic demand, inflammation, and impaired perfusion. Despite its potential impact, the role of fever in TSCI remains understudied, with only two prior studies addressing this issue, both using heterogeneous methodologies and yielding inconclusive results. Most existing work dichotomizes fever as present or absent, an approach that fails to capture the cumulative thermal stress experienced by the injured spinal cord. The concept of fever burden, integrating both intensity and duration of hyperthermia, may therefore provide a more accurate and clinically relevant measure. We conducted a retrospective observational study of adults with acute TSCI admitted to the Neurocritical Care Unit of the University Hospital of Geneva (2022–2024). Inclusion required surgical intervention within 48 h and continuous temperature monitoring; patients with concomitant severe traumatic brain injury or inability to provide consent were excluded. Core temperature was recorded minute-by-minute via bladder probe and aggregated to hourly values. Fever burden was defined as degree-hours above 37.9°C over the first 72 h. Neurological outcome was assessed with the American Spinal Injury Association (ASIA) Impairment Scale at admission and hospital discharge. Associations were evaluated using multivariable ordinal logistic regression adjusting for age, sex, and Injury Severity Score (ISS). Twenty-four patients were included (mean age 47.3 ± 19.2 years; 66.7% male; mean ISS 25.6 ± 12.7). Fever episodes occurred in 79.2% of patients, representing 15.8% of monitored time. Greater fever burden was independently associated with poorer neurological recovery (OR 1.29, 95% CI 1.14–1.53, p < 0.001). Older age was associated with reduced odds of bad outcome (OR 0.97 per year, 95% CI 0.97–0.98, p < 0.001). Early fever burden after TSCI is common and independently associated with worse ASIA outcomes at discharge. These findings support precise temperature surveillance and suggest targeted fever management as a modifiable therapeutic focus. Larger, multicenter prospective studies are warranted to validate these results.
Keywords
Introduction
Fever occurs in up to two-thirds of patients with acute traumatic spinal cord injury (TSCI) within the first two weeks postinjury, a period marked by altered spinal cord metabolic stress.1–4 Despite its high prevalence, the impact of fever on neurological recovery in TSCI remains poorly understood and insufficiently investigated.
While data from subarachnoid hemorrhage (SAH), ischemic stroke, and traumatic brain injury (TBI) suggest that fever is associated with worsened neurological outcomes,5–7 this correlation has not been validated in TSCI. Given the shared secondary injury mechanisms between TSCI and TBI, including ischemia, excitotoxicity, and neuroinflammation, the potential detrimental effects of fever in TSCI warrant further investigation.3,8–10 Notably, temperature management strategies differ between TBI and TSCI, despite their pathophysiological similarities.
While fever in TSCI has been well described, only two studies have examined its relationship with neurological outcome,11,12 and only one has distinguished between neurogenic, inflammatory, and infectious fever. Preliminary evidence suggests that infectious fever develops later during hospitalization, whereas neurogenic and inflammatory fever predominate in the early phase.4,11–15 Given this temporal pattern, fever within the first 72 h postinjury is more likely to reflect an inflammatory or neurogenic response, potentially influencing secondary injury processes and neurological recovery.
This study aimed to investigate the association between early fever burden and neurological outcome during the first 72 h of intensive care, a critical window where inflammatory and neurogenic fever are most prevalent. By focusing on this early phase, we sought to determine whether fever burden in the acute postinjury period is associated with worse neurological recovery and to explore the potential role of targeted fever management strategies in optimizing outcomes for TSCI patients.
Methods
Study design and patient population
This retrospective observational study analyzed patients admitted with TSCI to the Neurocritical Care Unit (NCCU) between 2022 and the end of 2024. Ethic approval was given by the regional Ethical Office Committee (ID: 2025-01006). A total of 25 patients were identified; however, one was excluded due to missing data, resulting in a final cohort of 24 patients. Eligible patients were ≥18 years old with a TSCI who underwent surgical intervention within 48 h of injury. Patients with severe TBI or those unable to provide consent were excluded from the study.
Temperature monitoring and fever burden calculation
Core body temperature was continuously recorded using a bladder thermic probe, with data acquisition at 1-min granularity and storage in an electronic patient data management system. Hourly temperature values were derived as the mean of all minute-by-minute recordings within the preceding hour.
Fever was defined as T° >37.9°C. Data underwent preprocessing to remove artifacts, including implausible values (<33°C or >42°C). Missing temperature values were linearly interpolated when up to three consecutive hourly values were unavailable, accounting for 10.2% of interpolated values. Patient days were excluded from analysis if >20% of temperature values per day were missing following interpolation, requiring at least 17 available temperature values per 24-h period for inclusion.
The daily fever burden (°C × hours) was quantified by summing the hourly temperature values exceeding 37.9°C. If the temperature remained below this threshold, a value of 0 was assigned. Fever burden was subsequently categorized into four groups based on incremental body temperature thresholds of 0.25°C starting from 37.9°C, defined as No, Low, Moderate, and High.
ICU fever management
In accordance with our institutional practice, all patients underwent blood cultures at the onset of fever, while sputum cultures were obtained when clinically indicated. Antipyretic therapy with paracetamol (1 g, four times daily) was routinely administered for analgesia and temperature control according to unit protocol. Standardized intensive care management also included blood pressure optimization with mean arterial pressure targets above 80 mmHg for the first 5 days of hospitalization.
Neurological outcome assessment
Neurological outcome was assessed using the American Spinal Injury Association (ASIA) Impairment Scale at admission and hospital discharge. Patients were classified into two outcome groups based on ASIA score progression: those with neurological improvement between admission and discharge were categorized as having a good outcome, while those with no improvement or worsening of ASIA score were classified as having a poor outcome.
Statistics
All statistical analyses were performed using R (Version 2024.04.2 + 764).
Descriptive statistics were reported as mean (SD) for continuous variables and frequencies (%) for categorical variables. Comparisons between patients with good and poor neurological outcomes were conducted using the Student’s t-test or Wilcoxon rank-sum test for continuous variables, depending on normality, and the chi-square test or Fisher’s exact test for categorical variables (Table 1).
Baseline Demographic and Clinical Characteristics according to Neurological Outcome.
Median (IQR); n(%).
Wilcoxon rank sum test; Fresher’s exact test; Wilcoxon rank sum exact test.
Continuous variables are reported as median (interquartile range) and categorical variables as n (%). Temperature corresponds to the patient-level mean of hourly interpolated body temperature measurements collected during the first 72 h of intensive care unit admission, as defined in the Methods. Comparisons between outcome groups were performed using the Wilcoxon rank-sum test for continuous variables and Fisher’s exact test for categorical variables.
To explore the relationship between fever burden and neurological outcome, univariate ordinal logistic regression was performed to identify potential predictors for inclusion in the final multivariable ordinal logistic regression. Variables with p < 0.10 in univariate analysis were considered for inclusion in the multivariable model. The final ordinal mixed-effects model implemented using the ordinal package incorporated fever burden, age, sex, and Injury Severity Score (ISS) as fixed effects, while patient ID was included as a random intercept to account for repeated measures within individuals.
Statistical significance was defined as p < 0.05, and all tests were two-tailed.
Results
A total of 24 patients were included, with a mean age of 47.3 (SD 19.2) years, and 66.7% were male. The mean ISS was 25.6 (SD 12.7). Regarding injury distribution, 50% sustained cervical spinal cord injuries, 29.2% thoracic, and 20.8% lumbar injuries. The mean body temperature during the study was 37.28°C (SD 0.66). All patients underwent surgical fixation within 24 h. The ASIA impairment scores at admission were 53.9% ASIA A, 8.1% ASIA B, 9% ASIA C, and 1.9% ASIA D.
Fever episodes (T° >37.9°C) occurred in 79.2% of patients (n = 19). The proportion of total monitored time spent above 37.9°C was 15.8%, and 34.4% of the time was above 37.5°C. Hypothermic episodes (T° < 36°C) were rare (1% of recorded time). The mean daily fever burden (T° >37.9°C) was 15.9%, with an average fever duration of 9.6 min (0.16 h) per day. (Fig. 1)
Mean hourly fever burden (°C above 37.9) following TSCI, illustrating the significant peaks in fever burden over time (72 h). TSCI, traumatic spinal cord injury.
To examine the association between fever burden and neurological outcome, patients were classified as having either Good or Poor recovery, with a good outcome defined as an improvement in ASIA score between admission and discharge. (Fig. 2) Hospital length of stay did not differ significantly between patients with good and poor neurological recovery (median [IQR]: 7 [6] vs. 8 [7] days, p = 0.70). Ordinal logistic regression was first performed to determine potential predictors for inclusion in the ordinal mixed-effects model. In the multivariable ordinal logistic regression model, both fever burden and age were independently associated with ASIA outcome. Higher fever burden was significantly related to poorer neurological recovery (OR 1.29, 95% CI 1.14–1.53, p < 0.001), while increasing age was associated with reduced odds of worst neurological outcome (OR 0.97 per year, 95% CI 0.97–0.98, p < 0.001). Neither sex nor ISS showed a significant association with neurological outcome (p = 0.43 and p = 0.67, respectively) (Table 2).
Alluvial plot illustrating the progression of ASIA scores by Fever Burden. This plot shows the relationship between the initial ASIA score (ASIA Entry) and the final ASIA score (ASIA Final), with the patients categorized into two outcome groups: Good and Bad. Good outcome was defined as an improvement in ASIA score between admission and discharge. ASIA, American Spinal Injury Association.
Multivariable Ordinal Logistic Regression Results
Multivariable ordinal logistic regression model assessing predictors of ASIA final outcome. Higher fever burden was independently associated with poorer neurological recovery. Odds ratios (OR) <1 indicate reduced odds of achieving a better ASIA grade, whereas OR >1 indicate increased odds of worse neurological status (ASIA A = complete injury; ASIA E = normal).
Discussion
These findings suggest that fever burden is a modifiable risk factor in spinal cord injury recovery and emphasize the potential role of targeted fever management strategies in improving neurological outcomes.
In this retrospective study, we found that a higher fever burden during the acute phase of TSCI was significantly associated with poorer neurological outcomes at hospital discharge. Importantly, this association persisted independently of baseline injury severity, suggesting that fever may contribute directly to secondary injury mechanisms rather than simply reflecting disease severity. These observations are consistent with evidence from other acute neurological disorders, such as SAH, ischemic stroke, and TBI, where fever has consistently been linked to unfavorable outcomes.5,7,16 In the SCI population, Ülger et al. recently demonstrated that noninfectious fever was associated with increased mortality, further supporting the detrimental role of dysregulated thermoregulation after spinal trauma. 12 By quantifying fever burden as a cumulative, continuous measure, our study extends this knowledge by showing that total exposure to hyperthermia, rather than the mere occurrence of fever, may be critical in shaping recovery trajectories.
The biological plausibility of our findings is supported by both experimental and clinical evidence. Preclinical models have shown that even moderate increases in temperature can exacerbate lesion site metabolic distress, mitochondrial dysfunction, and excitotoxicity, thereby aggravating neuronal and glial cell loss.3,10,11,17–20 Hyperthermia also amplifies neuroinflammation, which is increasingly recognized as a major contributor to secondary spinal cord damage. Clinical studies similarly highlight metabolic crises and inflammatory cascades as central mechanisms of ongoing injury.21,22 Taken together, these data suggest that fever may act both as a marker of severity and as a mediator of secondary injury, amplifying pathophysiological processes that hinder neurological recovery.
Our results also raise important clinical considerations.
First, fever after TSCI may not exclusively reflect infectious complications but can also be neurogenic in origin, as previously described. This distinction has practical implications, as neurogenic fever may be underrecognized and undertreated, leading to unnecessary delays in initiating targeted temperature management. Second, the timing of fever appears important: early episodes, more likely to be noninfectious, may represent a particularly vulnerable window where intervention could be most effective. Finally, given the feasibility of continuous temperature monitoring in the ICU, fever burden represents a pragmatic, readily accessible, and potentially modifiable risk factor that could be systematically integrated into neurocritical care protocols.
Limitations
This study has several limitations. First, its retrospective single-center design and relatively small cohort limit the generalizability of our findings and increase the risk of selection bias. Second, we assessed neurological outcome only at hospital discharge, without long-term follow-up, which may underestimate the true impact of fever burden on recovery. Third, differences in surgical timing within the 48-h window could not be modeled and may have influenced neurological outcome. Finally, unmeasured confounders may have influenced both fever patterns and neurological outcomes.
Conclusion
In this study, a higher early fever burden was associated with poorer neurological outcomes in patients with TSCI. These findings suggest that fever burden may represent a clinically relevant marker associated with neurological recovery in the acute phase. Given the observational design and limited sample size, causality cannot be inferred, and larger prospective studies are required to confirm these associations.
Transparency, Rigor, and Reproducibility Summary
This study was not preregistered as it was a retrospective observational analysis of patients admitted with acute TSCI to the NCCU of the University Hospital of Geneva between 2022 and 2024. The analytic plan was prespecified by the study team before data extraction, although not formally registered in a public repository.
A total of 25 patients were screened, of whom one was excluded for missing data, resulting in a final cohort of 24 patients. Eligibility criteria were predefined: adult patients (≥18 years) with TSCI requiring surgical intervention within 24 h. Patients with concomitant severe TBI or inability to provide consent were excluded. Continuous temperature data were obtained via bladder probe monitoring integrated into the electronic patient data management system. Neurological outcome was assessed in all included patients using the American Spinal Injury Association (ASIA) Impairment Scale at admission and discharge.
Outcome assessments were performed by trained clinical staff as part of routine care. Data collection and extraction were performed by investigators aware of patient clinical characteristics, but analyses followed a prespecified plan. Preprocessing steps included the removal of physiologically implausible values and linear interpolation of missing intervals shorter than 3 h. Missing hourly temperature values accounted for 10.2% of the dataset. Patients with >20% missing data per day were excluded from analysis.
Temperature was acquired continuously at 1-min resolution and aggregated into hourly averages. Fever burden was defined as degree-hours above 37.9°C, a reproducible and clinically interpretable metric used in prior neurocritical care studies. All analyses were conducted in R (version 2024.04.2 + 764).
Given the exploratory nature and small sample size, no a priori power calculation was performed. Statistical significance was defined as p < 0.05 (two-tailed), and effect sizes with 95% confidence intervals are reported. Multiple comparisons were limited, as fever burden was the primary exposure and neurological outcome the primary end-point.
The study was reviewed and approved by the Cantonal Ethics Committee of Geneva.
This is the first report from our group; replication in larger, multicenter cohorts is warranted. Deidentified data will be available on reasonable request to the corresponding author, subject to institutional and ethical approval.
Footnotes
Authors’ Contributions
T.R.: Conceptualization and methodology. T.R. and D.L.: Validation. T.R. and D.L.: Formal analysis. T.R.: Writing—original draft preparation. T.R. and Q.H.: Writing—review and editing. Q.H., D.P.S., and Q.H.: Supervision. All authors read and approved the final version of the article.
Data Availability Statement
All data generated or analyzed during this study are included in this published article and its supplementary information files.