We searched the following databases without limits on language: MEDLINE, SportDiscus, CINAHL, the Cochrane Reviews database, and the Physiotherapy Evidence Database. All dates (which varied according to database) were included, and the search was performed in April 2007. Previously known cases, review articles, and reference lists of available studies were cross-referenced for possible articles meeting inclusion criteria. Key words used for searches were cooling, cryotherapy, water immersion, cold-water immersion, ice-water immersion, icing, fanning, bath, baths, cooling modality, heat illness, heat illnesses, exertional heatstroke, exertional heat stroke, heat exhaustion, hyperthermia, hyperthermic, hyperpyrexia, exercise, exertion, running, football, military, runners, marathoner, physical activity, marathoning, soccer, and tennis. This original search revealed a total of 89 possible studies.
Specific inclusion criteria identified before data analysis included (1) exercise-associated hyperthermia, (2) pretreatment hyperthermia greater than 38.5°C (101.3°F), (3) a valid core body temperature measurement to characterize hyperthermia, (4) detailed explanation, sufficient for repeatability of the cooling modality methods, and (5) original research studies with human participants. Passive hyperthermia was excluded so that results from exercising conditions would reflect traditional athletic training and military settings. Individuals with classic heat stroke may respond differently to whole-body cooling than do EHS victims. Core body temperatures exceeding 38.5°C (101.3°F) normally occur in some athletic settings, and this cutoff point was selected to appropriately generalize research findings to common athletic training settings, even though rapid cooling is not necessary.3
Valid core body temperature measurement techniques included rectal thermistors, ingestible telemetric sensors, bladder catheters, and esophageal thermistors. Investigations using other temperature measures were excluded because these devices are known to report invalid results for exercising individuals.14–,16
Animal studies were not included. Case reports were not included in the overall data analysis but appear in a separate section.
The time between maximum core body temperature and treatment initiation is one determinant of the overall effectiveness of treatment. The main outcome for an EHS patient is based on the amount of time a victim remains above a critical temperature threshold. Also, if the time to treatment is 15 minutes versus 1 minute, this affects the cooling rate, because the body can gain or lose heat via radiation to the environment. The time to treatment was not considered an exclusionary criterion in our review, because authors offered enough information to provide a reasonable estimate of this measure. All groups explained their methods similarly, which led us to assume that less than 10 minutes separated the hyperthermia and initiation of treatment. We tried to contact authors of previous research to assess this variable if it was not reported originally.
Abstracts of the 89 identified papers were included in this review only if they met all inclusion criteria previously identified (). A total of 17 abstracts met these criteria. In their abstracts, some authors did not report complete methods or describe measures taken. For this reason, 1 reviewer completed a full-text evaluation of methods to assure that all inclusion criteria were met. This review identified 7 articles that met the inclusion criteria; a quality assessment review of these studies followed.
Selection process for articles included in the systematic review. PEDro indicates Physiotherapy Evidence Database.
Two reviewers independently assessed the methodologic quality of the studies based on the Physiotherapy Evidence Database (PEDro) scale.17
An a priori inclusion score of 4/10 was selected for an article to be included in data analysis. If discrepancies existed between reviewers, an open discussion took place to ensure that one had not missed or misinterpreted an aspect of the study. After discussion, if an agreement was not reached, a third independent reviewer assessed the article based on the scale. This reviewer was independent and was blinded to previous reviews and discussions. Initial κ statistics revealed strong agreement of 0.984. After a consensus meeting, the κ statistic for agreement was 1.000. A third independent reviewer was not needed for unresolved disagreement on PEDro scores. Final PEDro scale scores for the 7 articles that met the inclusion criteria are shown in .
Physiotherapy Evidence Database (PEDro) Scale Scores of Critically Reviewed Articles
A score of 4/10 on the PEDro scale was chosen because complete blinding of participants and therapists is impossible when assessing whole-body cooling modalities. Thus, the maximum score obtainable for this research is 7/10. Case reports and epidemiologic studies were not included because they achieved a zero score on the PEDro scale.
None of the studies included blinding of participants, assessors, or researchers. Three groups18–,20
did not randomly allocate volunteers to groups. One set of investigators18
was unable to conceal allocation to groups or control the fact that important prognostic indicators were not similar at the beginning of whole-body cooling. All studies that met the inclusion criteria, however, met the criterion score of 4/10.18–,24
To determine if recommendations for certain modalities were warranted, we identified 3 categories of cooling based on efficiency. Category A required approximately 20 minutes of cooling for an EHS patient with a maximum core body temperature of 42.2°C (108°F) to be cooled to 38.89°C (102°F). Category B required up to 40 minutes for the same body temperature reduction, and category C required more than 40 minutes of cooling. The cooling rates identified were category A, equal to or greater than 0.155°C · min−1; category B, greater than 0.078°C · min−1 but less than 0.155°C · min−1; and category C, less than or equal to 0.078°C · min−1.
Only cooling rate data were synthesized for this review. Some authors included other outcome variables, but because our original research question focused specifically on cooling efficiency after exercise-induced hyperthermia, we included only data related to these criteria in the current review.