Pascal Haegeli
Research report prepared for WorkSafeBC, Avisualanche Consulting, Vancouver BC, p. 106.
Publication year: 2012

Executive summary

The goal of the present study was to collect detailed information on Canadian avalanche incidents involving avalanche balloon packs and to conduct a preliminary statistical analysis on the effect of avalanche balloon pack on the survival rate of individuals seriously involved in avalanches in Canada. More specifically, the project aimed to address the following three research questions:

  1. Is there a significant difference in the survival rates between avalanche victims with and without avalanche balloon packs?
  2. Are avalanche balloon packs able to significantly reduce the severity of burials?
  3. Does the use of avalanche balloon packs lead to higher rates of trauma among avalanche victims?

Information on avalanche accidents involving avalanche balloon packs was collected using a variety of sources including existing databases of avalanche incidents involving avalanche balloon packs, reports from the British Columbia Coroner Service, InfoEx reports, accident report submissions to the Canadian Avalanche Centre and incident related questions in the online survey about the challenges associated with the operational implementation of avalanche balloon packs. Any indications about a relevant incident were follow up with personal interviews with the accident party, rescue personnel, or accident investigators to confirm the involvement of an avalanche balloon pack and gather detailed accident information. In total, the data collection effort resulted in 43 well-documents incidents involving 92 individuals.

The results of the present evaluation critically depend on what types of avalanche involvements are included in the dataset. For the present analysis, only serious involvements were considered, which were defined as individuals who were considerably involved in the flow of the avalanche (i.e., were unable to remaining standing during the involvement) and/or were buried by the avalanche. This criterion ensured that the evaluation only included cases, where a deployed avalanche balloon pack actually had a realistic chance to affect the outcome of the involvement. This strict inclusion condition reduces the dataset to 34 avalanche incidents involving a total of 71 individuals. Within this dataset, 64% (46 of 71) of the involved individuals were wearing an avalanche balloon pack at the time of the accident. The overall survival rate in the dataset was 73% (52 of 71). Overall survival rates were significantly higher in incidents reported by professional avalanche safety operations (89%; 33 of 37) than amateur recreationists (56%; 19 of 34), clearly highlighting the superior reporting practices by avalanche professionals (Pearson’s chi-squared test; p-value = 0.004).

Thirty-seven percent (17 of 46) of the individuals wearing an avalanche balloon pack at the time of the accident were unable to successfully deploy the safety device. Following the approach of previous studies on the effectiveness of avalanche balloon packs, these cases were combined with the successful deployments and then compared against avalanche victims who were not equipped the safety device. This perspective provides a comprehensive evaluation of the safety device as it views improperly deploy avalanche balloon packs as failures of the safety device and includes the consequences of these failures in the overall evaluation of the device.

A univariate comparison of the survival rates between users and non-users of avalanche balloon packs in the complete dataset reveals that the use of this safety device results in a statistically significant increase of the survival rate from 56% for non-users to 83% for users (Pearson’s chi -squared test; p-value = 0.03; 95% confidence interval: 4% to 49%). The 27 percentage point improvement in the survival rate can be interpreted that in the given dataset one additional life could have been saved with every additional 3.7 victims equipped with avalanche balloon packs. This perspective is equivalent to the ‘numbers needed to treat’, a measure frequently used in epidemiological studies to assess the effectiveness of health-care interventions.

Additional factors that significantly affected the survival rate in the complete dataset included avalanche size (Wilcox rank-sum test: p-value = 0.006), terrain character (alpine, treeline, and below treeline; Pearson’s chi-squared test: p-value = 0.036), relative location of victims when avalanche released (start zone, track, and runout; Wilcox rank-sum test: p-value = 0.020) and marginally whether the victim was an avalanche professional or not (Pearson’s chi-squared test: p-value = 0.094). Only a multivariate analysis would be able to control for the additional influences and properly isolate the effect of avalanche balloon packs, but the small size of the current dataset prevented such an analysis.

Detailed information on the degree of burial was only known for 93% (66 of 71) of the involved individuals. A Wilcox rank-sum test indicated significant differences in the degree of burial (p-value = 0.002) with users of avalanche balloon packs experiencing significantly shallower burials. No significant differences were observed in injury rates among avalanche survivors with and without avalanche balloon packs (Pearson’s chi-squared test: p-value = 0.863).

To reduce the effect of the inferior reporting practices of amateur recreationists, the analysis was repeated using only incidents that were reported from professional avalanche safety operations. This dataset consisted of only 37 victims involved in 18 avalanche incidents, which represented 52% (37 of 71) of the complete dataset. In this dataset, the use of avalanche balloon packs also resulted in a statistically significant improvement of the survival rate by 36 percentage points from 64% for non-users to 100% for users (Pearson’s chi-squared test: p-value = 0.007; 95% confidence interval: 3% to 41%).

Since the samples of balloon pack users in the previous two comparisons included single involvements with avalanche balloon pack while the control group was drawn only from accident s with multiple involvements, the control group is likely biased towards larger accidents. This has the potential to overestimate the benefits of avalanche balloon packs. In an attempt to control for this issue, the complete dataset was reduced a second time to only include incidents where both users and non-users of avalanche balloon packs were involved. This approach also partially reduces the variability among other contributing factors since the characteristics of the accidents included in the user and non-user groups are now more evenly matched. This reduced dataset consisted of only six incidents with a total of 35 involvements, which represented 49% (35 of 71) of the complete dataset. Within this sample, victims who were wearing an avalanche balloon pack had an 80% survival chance, which was 24 percentage points higher than the 56% survival rate of non-users. Even though this improvement is in the same range as the previous two estimates, it is statistically not significant due to the small sample size (Pearson’s chi squared test: p-value = 0.347; 95% confidence interval: -8% to 56%).

Even though the described results indicate that avalanche balloon packs are able to reduce the severity of burials and increase survival rates, critical burials (i.e., burials with impaired airways) and fatalities do occur even with completely inflated avalanche balloon packs. In the complete dataset, 24% (7 of 29) of the individuals with a fully deployed avalanche balloon pack were critically buried and 17% (5 of 29) did not survive their avalanche involvement. While this dataset is too small to offer conclusive statistical evidence, a qualitative examination of these incidents indicates that the effectiveness of avalanche balloon packs is reduced in larger avalanches, when victims are caught in the track or runout zone of the avalanche and when terrain traps are present.

Overall, the results of the present univariate analyses confirm the positive results of existing scientific evaluations of avalanche balloon packs (Tschirky et al., 2000; Brugger & Falk, 2002; Brugger et al., 2003; Brugger et al., 2007). However, only a multivariate analysis would be able to control for the additional influences on avalanche survival and properly isolate the effect of avalanche balloon packs. Furthermore, the qualitative analysis of avalanche incidents that resulted in fatalities despite fully deployed avalanche balloon packs provide important insights about the limitations in the effectiveness of this safety device. Continuous collection of high-quality avalanche accident information is crucial for further improving our understanding of avalanche survival and the effectiveness of avalanche safety equipment.

Contact Pascal Haegeli for a copy of this report.