Association between serum creatinine levels at emergency admission and outcomes in patients with exertional heat stroke: a multicenter cohort study

Exertional heat stroke (EHS) is a life-threatening condition characterized by profound central nervous system (CNS) dysfunction and core temperature typically >40 °C.^[1] This condition involves complex pathophysiological processes in which heat triggers a cascade of dysregulated inflammatory responses, endothelial dysfunction, coagulation abnormalities, and muscle damage. These processes can lead to multiorgan failure, significantly increasing the risk of mortality.^[2] Given the severity of EHS, early identification and timely intervention are crucial. However, there are no specific diagnostic markers for EHS,^[1] highlighting the need to identify reliable clinical parameters that can assist early decision-making.

Serum creatinine (Scr) is one of the most widely used biomarkers in clinical practice for assessing renal function. It is pivotal not only for diagnosing chronic kidney disease but also for defining acute kidney injury (AKI) as a core criterion.^[3] Moreover, Scr has demonstrated prognostic value in various diseases, such as being a predictor of mortality in patients suffering from paraquat poisoning^[4] and being associated with prolonged mechanical ventilation, extended intensive care unit (ICU) stays and longer hospitalizations in patients undergoing cardiac surgery.^[5] Furthermore, Scr measured at admission has been linked to adverse outcomes in critically ill patients, including mortality and the need for renal replacement therapy.^[6]

These findings suggest that Scr may be relevant for prognostic assessment in EHS patients. However, there is a lack of specific studies exploring the relationship between Scr levels and EHS prognosis. Additionally, existing research has focused primarily on Scr changes in ICU patients.

To address this gap, we conducted a multicenter cohort study to investigate the relationship between Scr at ED admission and outcomes in EHS patients. Based on the literature, we hypothesized that elevated Scr levels would be associated with poor outcomes. Given that nervous system susceptibility in EHS and mortality are endpoints, we focused on the relationship of Scr with these two key outcomes. Additionally, we explored the dynamic changes in Scr levels within the first 24 h of admission, comparing patients with different clinical outcomes, to further clarify the prognostic value of Scr in patients with EHS.

This multicenter study enrolled EHS patients from EDs at six tertiary hospitals, comprising a retrospective cohort (January-September 2022) and a prospective cohort (June 2023-September 2024). Since the retrospective cohort data were anonymized, written informed consent was waived. Informed consent, either written or verbal, was obtained from the patients or their families in the prospective cohort, as required by hospital protocols.

Eligible patients met the diagnostic criteria for EHS, as defined by the 2021 edition of the Expert Consensus on the Emergency Diagnosis and Treatment of Heat Stroke.^[7] Furthermore, considering that prehospital cooling interventions may result in potential temperature reduction, we set an initial temperature above 39.5 °C on arrival to the EDs as an inclusion criterion. To account for measurement differences between cohorts (tympanic thermometry in retrospective data vs. rectal thermometry in prospective data), we performed a normalization adjustment of +0.5 °C for tympanic membrane readings, which was consistent with established clinical guidelines.^[8]

The exclusion criteria were as follows: need for immediate cardiopulmonary resuscitation; the presence of other severe conditions upon admission, such as cerebral hemorrhage, traumatic brain injury, or liver abscess; preexisting renal failure; and discharge from the ED within 24 h, where patient outcomes could not be determined. No patients were excluded because of missing Scr data, as Scr testing was routinely performed in the EDs of the participating hospitals.

All patients received rapid cooling and fluid resuscitation in the ED, followed by organ function support according to treatment guidelines. Based on the severity of their condition, patients were either admitted to the ICU or general ward, kept under observation, or discharged.

We collected data on patient demographics, medical history, admission vital signs, laboratory test results, and hospitalization outcomes. Additionally, all Scr values within the first 24 h of admission were recorded. Data were extracted from electronic medical records. In the EDs of the participating hospitals, critically ill patients routinely undergo complete blood count tests, biochemical assessments (including heart, kidney, and liver function markers), coagulation function tests, and blood gas analyses, with repeat tests based on clinical status until the patient was discharged.

The primary outcomes were neurological sequelae at discharge and in-hospital mortality. The secondary outcomes included acute respiratory failure (ARF), acute heart failure (AHF), the use of vasoactive drugs, the ICU length of stay, and the total hospital stay. Neurological sequelae at discharge were defined as in-hospital death or deterioration in neurological function compared with the baseline status, such as ataxia, speech impairment, and cognitive dysfunction.^[9] The criteria for ARF and AHF were based on a previous study.^[10] Since Scr is the main diagnostic marker for AKI, including patients requiring continuous renal replacement therapy (CRRT), AKI and CRRT were not considered outcomes in this study. Additionally, rhabdomyolysis, a common complication of EHS and a key contributor to elevated Scr levels, was also excluded as an outcome.

The association between Scr levels at ED admission and study outcomes was evaluated via logistic or linear regression models. Scr was analyzed as both a continuous variable (per 10 µmol/L increment) and a categorical variable (>140 µmol/L vs. ≤140 µmol/L). Multivariate models were established and adjusted for comorbidities and vital signs. The outcome-specific covariates included troponin I (AHF) and blood gas (ARF).

A generalized additive mixed model (GAMM) was employed to visualize Scr trends within the first 24 h of admission, particularly comparing patients with and without neurological sequelae. This analytical approach overcomes longitudinal data limitations through three key strengths: (1) managing irregular measurement intervals; (2) preserving statistical power despite missing data points; and (3) capturing nonlinear temporal associations through flexible smoothing splines.^[11]

Continuous variables are expressed as the means ± standard deviations or medians (interquartile ranges) and were compared via t tests or Mann-Whitney U tests. Categorical variables are expressed as percentages and were analyzed via Chi-square tests or Fisher’s exact tests. Statistical analysis was conducted via EmpowerStats (www.empowerstats.com) and R (www.r-project.org). All tests were two-sided, with P-values <0.05 considered statistically significant.

Ultimately, 177 patients were included in the final analysis (supplementary Figure 1), with a mean age of 59.7±12.8 years and 71.2% males. Upon admission, the mean core body temperature was (41.3±0.9) °C and the median Glasgow Coma Scale (GCS) score was 7 (5-11). The incidences of AHF, ARF, and AKI were 10.7%, 40.7%, and 37.3%, respectively. Neurological sequelae were observed in 26 patients (14.7%), and 16 patients (9.0%) died (supplementary Table 1).

At admission, the median Scr level was 114.20 (88.50-140.20) µmol/L, with 45 patients presenting Scr levels above 140 µmol/L. When 140 µmol/L was used as a threshold for subgroup analysis, patients in the high-Scr group were more likely to have cardiovascular diseases and neurological abnormalities, as well as lower systolic blood pressure (SBP), oxygen saturation (SpO₂), and GCS scores. Additionally, potassium, sodium, and chloride levels were relatively high. Other indicators of organ function, including aspartate aminotransferase (AST), troponin I, BNP, and lactate, were significantly elevated in the high-Scr group, whereas the standard base excess (SBE) level was lower. Patients in the high-Scr group had a higher incidence of adverse outcomes and a longer hospital stay than did those in the low-Scr group.

Each 10 µmol/L increase in Scr independently predicted neurological sequelae (adjusted odds ratio [aOR]: 1.14, 95% confidence interval [95% CI] 1.02-1.27, P=0.021) and in-hospital mortality (aOR: 1.14, 95% CI 1.02-1.27, P=0.016) after adjusting for cardiovascular disease, mental disorders, core temperature, SBP, SpO₂, and GCS score (Table 1).

Table 1.

Logistic regression of serum creatinine for outcomes in patients with exertional heat stroke

Serum creatinine	Neurological sequelae at discharge, OR (95% CI)					Hospital mortality, OR (95% CI)
	Univariate	P-value	Adjust model	P-value		Univariate	P-value	Adjust model	P-value
Per 10 µmol/L increase	1.15 (1.07-1.24)	<0.001	1.14 (1.02-1.27)	0.021		1.13 (1.05-1.23)	0.002	1.14 (1.02-1.27)	0.016
≤ 140 µmol/L	Reference		Reference			Reference		Reference
> 140 µmol/L	10.33 (4.07-26.22)	<0.001	7.70 (2.36-25.80)	<0.001		6.00 (2.04-17.65)	0.001	4.05 (1.13-14.55)	0.031

Model adjusted for mental disorders, cardiovascular disease, core temperature, systolic pressure, pulse oxygen saturation, and Glasgow Coma Scale score; OR: odds ratio; 95% CI: 95% confidence interval.

After adjusting for potential confounders, the high-Scr group had a higher risk of neurological sequelae (aOR 7.70, 95% CI 2.36-25.80, P<0.001) and higher in-hospital mortality (aOR 4.05, 95% CI 1.13-14.55, P=0.031) compared to the low-Scr group (Table 1).

Scr levels were also significantly associated with other adverse outcomes, including vasopressor use, AHF, ARF, and prolonged ICU stays. After adjusting for covariates, most associations remained, except for ARF (supplementary Figure 2).

We employed a GAMM to examine the 24-hour Scr trends between patients with and without neurological sequelae. As presented in Figure 1 and supplementary Table 2, patients with neurological sequelae experienced a significant increase in Scr levels over 24 h (β 51.60, 95% CI 37.03-66.17, P<0.001), whereas those without sequelae exhibited a marked decline (β -57.47 [-86.98 to -27.95], P<0.001). On average, the Scr level in the sequelae group rose by 3.64 µmol/L per hour (95% CI 2.73-4.55, P<0.001) compared with that in the non-sequelae group.

Figure 1. Association between the serum creatinine level and neurological sequelae during the first 24 h using GAMM. The neurological sequelae was notably influenced by serum creatinine levels during the first 24 h. Patients with neurological sequelae demonstrated a significant increase in creatinine levels over this period, whereas those without sequelae showed a marked decrease. GAMM: generalized additive mixed model.

Our study demonstrated that Scr at ED admission was significantly associated with neurological sequelae and in-hospital mortality. Scr was also correlated with other adverse outcomes, including vasopressor use, AHF, ARF, and prolonged ICU stays. Furthermore, the 24-hour post-admission trend in Scr levels differed significantly between patients with and without neurological sequelae.

Elevated Scr in EHS patients may result from multiple mechanisms: heat and the ensuing systemic inflammatory response, along with coagulation abnormalities, directly damage renal function by impairing creatinine filtration and excretion. Additionally, in EHS, the redistribution of blood flow to dissipate heat reduces renal perfusion, causing renal hypoperfusion. Furthermore, dehydration, which is commonly observed in EHS due to hypovolemia from excessive sweating, further exacerbates Scr elevation.^[1] Finally, rhabdomyolysis—characterized by the release of muscle contents such as myoglobin into the circulation—increases renal strain and may lead to AKI.^[1]

Thus, elevated Scr in EHS patients reflects not only kidney injury but also systemic multiorgan dysfunction. Our findings indicate that Scr changes correlate with other organ function markers (e.g., AST and troponin), further supporting its value in assessing multiorgan dysfunction. Specifically, patients with elevated Scr at admission are more likely to require vasopressors, develop AHF, and have prolonged ICU and hospital stays. Previous studies indicated that elevated Scr levels are associated with worse outcomes in ICU patients with HS.^[12,13] Our research extends these findings to the early emergency care phase, highlighting the prognostic value of Scr as an early marker.

Although Scr is a widely used biomarker, its delayed increase after AKI may underestimate early renal dysfunction in patients with EHS. Therefore, we further analyzed the dynamic changes in Scr levels within 24 h post-admission. We observed that patients without neurologic sequelae had a significant decrease in Scr after active treatment, whereas patients with neurologic sequelae had persistently elevated Scr levels despite adequate hydration during the acute phase (which would normally reduce creatinine levels).^[5] This dynamic change suggests that Scr has prognostic value at admission, and its trend may help clinicians monitor disease progression and adjust treatment strategies.

While few studies have directly examined the relationship between Scr and neurological injury in EHS patients, existing studies have shown a positive correlation between elevated Scr levels and the risk of neurological complications, such as heat stroke-induced brain hemorrhage.^[14] We speculate that, in addition to the direct neurotoxic effects of heat exposure, impaired renal function may hinder the clearance of medium- and large-molecule toxins, which have potential neurotoxic effects.^[15] These mechanisms may partially explain the observed association between elevated Scr levels and neurological sequelae in our study.

Several limitations should be acknowledged. First, the inclusion of both retrospective and prospective cohorts may introduce variability in treatment strategies, potentially confounding the results. Second, Scr levels are influenced by factors such as muscle mass, diet, and medication use, which were not fully controlled in this study. Third, the analysis is limited to short-term outcomes during hospitalization, and future research should examine the relationship between admission Scr levels and the long-term neurological prognosis of EHS patients to address the limitation of focusing solely on immediate hospital outcomes. Finally, in the areas served by the participating hospitals, EHS cases are sporadic and are primarily observed among farmers working outdoors, workers in high-temperature environments, and individuals with impaired self-care abilities who continue outdoor activities in extreme heat. Thus, compared with cohorts involving athletes or military personnel, our patient population was older and more likely to have underlying medical conditions, making them more susceptible to adverse outcomes.

In summary, this study demonstrated that the Scr level at emergency admission may be a predictor of prognosis in patients with EHS and may be associated with neurological sequelae, in-hospital mortality, and other adverse outcomes. Dynamic changes in Scr further underscore its utility in the early management of EHS patients. As a routine and readily available biomarker, Scr could be included in early risk assessment and clinical decision-making processes for EHS patients.

Funding: This study was funded by Research Fund of Zhejiang Provincial Health Commission (2025KY8662025).

Ethical approval: This study was approved by the Ethics Committee of the Second Affiliated Hospital of Zhejiang University School of Medicine (approval 2022-0913 and 2023-0517).

Conflicts of interest: The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Contributors: DPJ (Dingping Jin) and LC (Lan Chen) are co-first authors. DPJ: conceptualization, formal analysis, writing - original draft, writing - review &editing; LC: conceptualization, formal analysis, writing - original draft, writing - review &editing; SYW, JNR, HZ, XLW, LYL, HMM: data collection and analysis; XQF: conceptualization, methodology, validation, writing - review & editing, supervision, project administration. All authors read and approved the final manuscript.

All the supplementary files in this paper are available at http://wjem.com.cn.