Introduction

During August 2003, France sustained an unusual heat wave that lasted 9 days and caused epidemic classic heatstroke. Excess mortality was estimated to be as high as 14,800 deaths, predominantly in the region of Paris (+134% mortality) [1]. Daily average peak temperature was 38–39°C; medium minimal temperature was at 23.4°C [1]. France was not the only European country facing such a heat wave, for Spain, Italy, Portugal and the United Kingdom were also concerned. French hospitals experienced massive admittance for heat-related disorders, mainly classic heat stroke, a medical entity unknown to most medical staff. Although exertional heat stroke is relatively well documented in the medical literature, clinical and biological features of classic heat stroke are more limited [2–5] especially with regard to the course of organ dysfunction. In this report, we describe the clinical and biological course of early organ dysfunction in a cohort of patients suffering classic heat stroke admitted in our intensive care units.

Patients and methods

The Institutional review board approved the study. Eligible patients were found in the emergency ward when demonstrating all three of Bouchama’s criteria: alteration of mental status (coma, delirium, disorientation, seizures), body core temperature of more than 40.6°C or documented evidence of cooling before the first record temperature and a reliable history of compatible environmental exposure: hot, dry or flushed skin [6]. Transfer to the intensive care unit occurred prior to any resuscitation. Patient selection and inclusion were prospective as ICU admissions went along. On arrival in the intensive care unit, all patients were cooled using external cooling with groin and axillary ice packs, continuous manual thin water spray and ambient air fan. Temperature of intra-venous fluids was ambient at the start of the heat wave. A few patients thereafter received cold intra-venous fluids (4°C). No patient received dantrolene sodium infusion because of its proven ineffectiveness [7]. All patients were managed to maintain mean arterial pressure (MAP) above 65 mmHg, heart rate below 100 bpm and urine output above 1 mL/kg/min. All patients received basic crystalloid infusion. Anaemia (defined as haemoglobin level below 70 g/l) and coagulation disorders were treated by administration of erythrocytes, platelets, and fresh frozen plasma according to clinical and laboratory results.

Organ dysfunction was assessed at admission and thereafter daily, from Day 0 to Day 7, using the Sequential Organ Failure Assessment (SOFA) score items including classic laboratory data, necessity of mechanical ventilation or vasoactive support (epinephrine or norepinephrine >0.1 μg/kg/min) [8]. Demographic characteristics, past medical history, Glasgow coma score, initial and maximum body core temperatures measured by rectal temperature, cooling time required to achieve a body core temperature of 38.5°C from ICU admission and other necessary data were collected using a pre-designed data collection form. Simplified Acute Physiologic Score (SAPS II) [9], length of intensive care unit and hospital stay were also calculated. A 1-year follow-up to assess neurological disability amongst survivors was organised.

Continuous variables are given as median and interquartile (25–75%) range. Data at admission and at 24 h were compared using non-parametric Wilcoxon’s matched-pairs signed-ranks test. Data comparing ICU survivors and non-survivors were compared using non-parametric two-tailed p value Mann–Whitney’s U test. Categorical data are given as number of cases and percentages and were compared using non-parametric two-sided p value Fisher’s exact test. Progress in achieving cooling was analysed in ICU survivors and non-survivors using the Kaplan–Meier product limit method. Curve comparison used the Mantel–Haenszel logrank test. Statistical analysis was performed using GraphPad Prism Software, San Diego, CA, USA. Difference was considered to be statistically significant at two-sided p value <0.05.

Results

Patients’ demographics

A total of 22 patients were admitted within 6 days (August 6th to August 13th). Sex ratio was 1 to 1. Median age was 68.5 years (61.3–76.8). 91% of patients suffered at least one underlying illness or was having chronic medication. 36% (8/22) were found to have psychiatric medication (namely phenothiazine, lithium and selective serotonin reuptake inhibitors), 45% (10/22) were found to have chronic arterial hypertension and concomitant medication (diuretics, beta-blockers), 23% (5/22) suffered chronic respiratory disease (asthma, COPD) and 1 patient was discovered at necropsy to have been suffering pheochromocytoma.

Mortality and length of stay

Median ICU length of stay was 8 days (2.0–23.5). ICU mortality was 63.6% (14/22). 7 patients died within the first 7 days following multiple organ failure. 7 patients died after demonstrating persistent neurological disability (coma = 6, tetraplegia = 1). Median total hospital length of stay was 17 days (2–25.5). None of the patients discharged from ICU died within a 1-year follow-up.

Early organ dysfunction course, temperature and cooling time

Median body temperature on admission was 41.1°C. Median cooling rate was 0.005°C/min (0.002–0.01). Median time to reach 38.5°C rectal temperature was 5 h 30 min (2–23 h 45 min). Within 24 h of admission, although cooling procedure had been started, body temperature initially increased in all patients before beginning to decrease (Table 1). In all patients, respiratory (p < 0.01), circulatory (p < 0.01), haematological (p = 0.01), hepatic (p < 0.01) and renal function (p < 0.01) deteriorated within the first 24 h of ICU management (Table 1). Neurological status also worsened but this did not reach statistical significance.

Table 1 Data comparison between admission and within 24 h after admission maximal values
Full size table

First 24 h data comparison in respect to ICU-mortality are shown in Table 2. One patient died on admission. Required cooling time was significantly shorter amongst ICU survivors as shown in Fig. 1 (logrank test p < 0.01, hazard ratio 0.25, 95% confidence interval 0.01–0.32). 6 patients did not reach thermal goal at 24 h (2 survivors and 4 non-survivors). Serum lactate (p = 0.05), cardiac troponin I (cTnI) (p = 0.01) and creatinine levels (p < 0.01) proved to be significantly higher in non-survivors whilst PaO2/FiO2 ratio (p = 0.04) was significantly lower. ECG recording was performed in all patients and did not reveal any remarkable pattern. Extra-renal therapy was required in 4 patients of which 3 deceased. SOFA score at 24 h was significantly higher than on admission (p < 0.01). SOFA score and SAPS II score at 24 h were correlated with mortality.

Table 2 Data comparison at 24 h between ICU-stay survivors and non-survivors
Full size table
Fig. 1
figure 1

Comparison in percentage of ICU survivors and non-survivors achieving the cooling target of 38.5°C body core temperature against time required to achieve the target within the first 24 h of admission. Data are expressed as percentage and time (logrank test p < 0.01)

Full size image

Discussion

We report here the organ dysfunction characteristics and outcome amongst 22 patients suffering non-exertional heatstroke admitted in ICU during the 2003 heat wave. We report a 63.6% total death rate during this epidemic of non-exertional heat stroke. In previous studies, mortality rate ranged from less than 10 to 39.7% [5, 10]. Heatstroke related death is mostly seen in the elderly, with comorbid conditions such as psychiatric, respiratory or cardiovascular illness and thermoregulatory response impairing medication. Our patients were no different in that regard from those of previous reports [5, 11]. We do speculate that our much higher death rate is explained by a dramatic delay in body cooling. Non-cooling in heat stroke is known to be fatal and delayed cooling may not be able to halt the process, once multi-organ dysfunction has started [12]. Witnesses discovered most patients. We were therefore unable to determine the very start of symptoms and the real duration of heat exposure. Required cooling time was statistically longer in non-survivors (p = 0.04) and cooling procedures were not able to reverse organ dysfunction (p = 0.01). We therefore suggest that duration of heat exposure prior to medical access was longer in non-survivors compared to survivors.

Respiratory, circulatory, haematological, hepatic and renal function all deteriorated within the first 24 h of ICU management (Table 1). We speculate that the intensity of inflammatory response triggered by heatstroke overweighed at start the capacity of external body cooling. Such a fact has up to now not been described elsewhere and demands experimental confirmation.

Cooling rate in former studies ranged from 0.14 ± 0.11 to 0.27 ± 0.03°C/min [13, 14]. We did not make use of any proper “body-cooling unit” as described by Weiner and Khogali [15] in 1980. The cooling device we used during this event was designed at a time of sudden mass casualties and more closely resembled a field device than a proper evaporative bed [14, 16]. Such a method was probably sub-optimal and may also explain why our cooling rates were so very much lower, therefore impacting on our ICU-mortality.

Serum cardiac troponin I levels raised in all patients within the first 24 h and were found to be significantly elevated amongst non-survivors. The release of cardiac troponin may be due to reversible or irreversible myocyte damage and myocardial strain [17]. It is suggested that raised troponin may predict hospital and ICU mortality in the case of sepsis and septic shock [18]. Cardiac and circulatory abnormalities during heat stroke are known [19] but data concerning cTnI are, to our knowledge, non-existent. Heat is known to be cytotoxic per se for many tissues including myocardial tissue [6]. Giving an answer to the clinical significance of raised serum troponin levels in our patients is difficult considering that no systematic coronarography or histological analysis of cardiac tissue biopsy was performed.

Conclusion

Medical staff should be aware that heat stroke related organ dysfunctions can demonstrate a rapidly worsening organ dysfunction course leading to death even though cooling procedures and intensive care management are promptly started. Interestingly, serum cardiac troponin I is raised amongst non-survivors. This is new, and for the moment, remains unexplained. Further research will be necessary to understand and confirm this finding.