Death of backcountry winter-sports practitioners in avalanches – A systematic review and meta-analysis of proportion of causes of avalanche death

This study estimates the proportions of the three major causes of avalanche death globally, and reviews potential factors influencing the proportions of causes of avalanche-related deaths (PCAD). By searching databases and consulting experts, we retrieved studies and registries in multiple languages, which examined PCAD. As a result, we retrieved 1,415 reports and included 37 for the study (22 for meta-analysis). We performed a meta-analysis to estimate pooled proportions. Between-study heterogeneity was assessed jointly by I2 and 95% prediction interval of pooled estimates. PCAD by trauma and asphyxia are 29% (95%CI 21–39%) and 82% (95%CI 72–88%), after the year of 2000. PCAD by hypothermia is 2% (95%CI 1–4%), estimated with studies having sufficient sample size. Time periods (before or after 2000), data representativeness (national subgroup), forensic procedures, and sample size explained between-study variation for proportions to a considerable extent. Factors influencing PCAD, that were either available or not available for quantitative synthesis, were summarized in a narrative systematic review (37 studies). In conclusion, we re-affirm asphyxia as the predominant cause of avalanche death, followed by trauma, and then hypothermia. Patterns of PCAD by trauma and asphyxia varied more after the year of 2000. A sample size > 75 is needed to estimate the proportion of hypothermia. PCAD discrepancies are lower in the data representing fatalities from a country than from regions. Without proper forensic diagnosis procedure, PCAD by trauma can be over-estimated. Under-reporting of forensic diagnostic criteria is an important bottleneck to the reliability of evidence in the field. Evidence on the role of other influencing factors to PCAD such as fatalities’ expertise and usage of mitigation gear is anecdotal and warrants further research. The results of meta-analysis build upon synthesizing and summarizing studies with moderate to high risk of bias and should be interpreted with caution.