Epidemiological Trends in Search and Rescue Incidents Documented by the Alpine Club of Canada From 1970 to 2005

The medical aspect of search and rescue (SAR) in Canada is a poorly understood element of prehospital care. There are two Canadian epidemiological studies that examined the morbidity and fatality associated with SAR incidents. One focused on mountain SAR operations in Banff, Yoho, and Kootenay National Parks and found that musculoskeletal injuries account for the majority of these incidents. The second study reviewed 272 medical action reports from the Canadian Forces Search and Rescue Technician program during a 4-year period and indicated that trauma-related cases were predominant. However, both studies underscore the typical array of medical conditions seen in any hospital setting are also being encountered during SAR missions. Their restricted geographic range and their small sample size limit these studies, respectively.

To date there are no provincial or national level investigations that analyze the epidemiology of SAR incidents in Canada. In an effort to widen our understanding of these patterns, this study has examined the SAR reports collected by the Alpine Club of Canada (ACC). Further insight into the epidemiology of SAR incidents is beneficial to all stakeholders involved in the medical care of these individuals. This includes SAR teams, emergency medical service personnel, and hospital healthcare professionals. Beyond medical relevance, expanding our knowledge of these trends is of value to the outdoor recreational tourism industry. Ultimately, an objective understanding of these patterns will contribute to evidence-based risk analysis, prevention strategies, and best-practice recommendations.

This study was a retrospective, cross-sectional review of the SAR incident database maintained by the ACC from January 1, 1970, until June 12, 2005. The ACC does not conduct SAR operations but does collect information on incidents from multiple sources as shown in Table 1. The database was initially populated through archival searches of ACC literature and self-reporting by ACC members. This process of data collection expanded to include information from personnel affiliated with the organizations outlined in Table 1. The ACC also created an online format for data acquisition through their website. For each SAR incident, the ACC collected the following: accident ID, date, route ID, description of event, analysis of event, cause, rescue mode, number of injured, injury description, number of fatalities, location, information source, and type of activity. For unknown reasons the database has not been formally maintained since June 12, 2005.

The characteristics of the ACC database are shown in Table 1, along with the exclusion criteria. The following definitions were used: Incident was an event that led to the creation of one or more casualties; casualty was a person who sustained any type of morbidity or fatality; injury was any condition due to a traumatic mechanism of injury; and illness was all other conditions not related to traumatic mechanism of injury. To organize the information contained in the ACC database, it was sometimes necessary to group similar terms for a given variable. This process was applied to the following variables: data source, mode of extraction by rescue personnel, activity undertaken during the incident, presumptive cause of the incident, anatomic location of the traumatic injury, and types of nontraumatic illness. Specific definitions have been provided for presumptive cause of the incident, anatomic location of the traumatic injury, and types of nontraumatic illness within their respective figures or tables. The proximity to the nearest urban center for a given incident was based on the shortest linear distance between 2 points. An incident may have included multiple victims; the term casualty captured this and represented the sum of victim(s) who sustained any type of morbidity or fatality. The aforementioned variables were described by presenting frequencies and percentages with the corresponding 95% CI. Microsoft Excel version 12.3.6 for Mac (Microsoft Corp, Redmond, WA) and Stata v13.1 (StataCorp LP, College Station, TX) was used for this analysis, along with ArcGIS (Esri, Redlands, CA). The University of Calgary Research Ethics Board (REB13-0815) gave approval for this study.

The ACC database contained 1088 incidents with 1377 casualties (Table 1). Of these casualties, 944 (68.6%; 95% CI, 64.2 to 73.1) suffered an injury or illness and 433 (31.4%; 95% CI, 28.6 to 34.6) were fatalities. Age data were available for 523 (38.0%; 95% CI, 34.8 to 41.4) of the casualties. Sex data were available for 1032 (74.9%; 95% CI, 70.4 to 79.7) of all casualties. When sex was reported, males accounted for 76.1% of all casualties and 82.3% of the fatalities (male fatalities were 4.8 [95% CI, 4.3 to 5.5] times higher than female fatalities). Figure 1 displays the number of casualties as a function of year and shows a stark increase in the number of casualties in 1999 and subsequent years. On a monthly basis there was a bimodal distribution of casualties with peaks in February (the winter peak was smaller) and August (Figure 2). Ground personnel (49.7%; 95% CI, 46.0 to 53.4) and helicopter (37.4%; 95% CI, 34.2 to 40.8) rescue were the predominant modes of extracting casualties. For helicopter extractions, ground personnel were also involved and ground SAR groups participated in 87.1% of all incidents (Table 2). Mountaineering (29.3%; 95% CI, 26.5 to 32.3) and hiking (27.0%; 95% CI, 24.3 to 29.9) resulted in more than half of all casualties that yielded any type of morbidity, whereas mountaineering (34.9%; 95% CI, 29.5 to 40.9) and skiing, ski mountaineering, or snowboarding (27.9%; 95% CI, 23.2 to 33.4) accounted for nearly two thirds of all fatalities (Table 3). Human error and slips and falls were the major contributors to the presumptive cause of incidents (Figure 3). The lower limb was the most common anatomic location of traumatic injury, accounting for 41.6% (95% CI, 37.6 to 45.9) of these injuries (Table 4). Hypothermia (49.4%; 95% CI, 41.8 to 57.8), exhaustion (18.7%; 95% CI, 14.2 to 24.2), frostbite (11.0%; 95% CI, 7.6 to 15.3), and dehydration (7.4%; 95% CI, 4.7 to 11.1) accounted for the majority of all nontraumatic illness. Data were obtained from across Canada with the exception of Prince Edward Island, Saskatchewan, and the Northwest Territories (Figure 3). British Columbia and Alberta accounted for 91.6% (95% CI, 86.0 to 97.5) of the incidents in the database. The closest urban centers with the highest incident rates were Vancouver and Golden in British Columbia and Banff, Canmore-Kananaskis, and Jasper in Alberta (Figure 4).

The bimodal distribution of incidents in February and August was likely related to the usage of the mountain environments for winter and summer sports and activities. A similar trend was observed in the study by Wild.

Although challenging to interpret given the missing data (as seen in Figure 2), males accounted for 76.1% of all casualties and 82.3% of all fatalities when sex data were available. Despite this limitation, Wild reported a similar rate of fatalities among men, but a lower rate of male casualties (63.1%). Two American studies that examined wilderness SAR in New Hampshire during a 3-year period and a 10-year review of SAR incidents in Yosemite National Park reported male casualties accounted for 64.5% and 62.0%, respectively. The Yosemite National Park findings and a study conducted in Pima, Arizona, that reviewed SAR incidents during a 13-year period reported that male fatalities accounted for 85.9% and 87.0%, respectively. However, it should be underscored that the settings for these American studies would be considered different than the Canadian Rocky Mountains or coastal British Columbia and may not be representative.

Ground SAR personnel were used in 87.1% of all incidents, and a helicopter (in conjunction with ground SAR personnel) was necessary in just over one third of all incidents. Within North America, the use of helicopters during SAR activities ranges from 2.5% to 64.0%,^{,  ,}  although a study from the Mont-Blanc region of the French Alps documents helicopter use in 95% of incidents. This observation underscores the differences in geography and the regional response culture that can be seen in other areas of North America and Europe.

Activities that are typically considered to be high risk, for example, mountaineering, skiing, ski mountaineering, and snowboarding, accounted for 62.8% of the fatalities. The latter of these 2 categories occurred outside of commercial ski resort boundaries and is not afforded the safety net that can be found in ski resort areas. Hiking contributed to only 32.0% of nonfatal incidents, whereas in the study by Wild it accounted for 43.5%, and American studies reported 52.0% to 55.0%.^, 

Not surprisingly, the lower limb was the most common anatomic location for traumatic injury, accounting for 41.6% of injuries, in keeping with literature from the United States,^,  France, and Scotland. As is expected when encountering casualties in the outdoor environment, hypothermia, exhaustion, frostbite, and dehydration represented a significant proportion of all nontraumatic illnesses. However, the other nontraumatic presentations in Table 4 accounted for approximately 13% of incidents and highlights the importance of ground SAR personnel having to contend with a diverse set of medical encounters. The Canadian ground SAR community is trying to standardize the technical training received by individuals working in these organizations. The information presented in this study and other Canadian-based studies may help to guide the development of this curriculum from a medical standpoint.

The data contained in the ACC incident reports are limited in a variety of ways. As the ACC focuses on the alpine environment, there is a predilection toward incidents that occur in mountainous regions and, in particular, the Rocky Mountains range. At the reporting level, incidents have been gathered by multiple organizations over the years, with no systems in place for corroborating events or checking the completeness of reporting. Within individual reports, there are frequent gaps in data, especially related to the age and sex of casualties. Medical information is recorded from multiple sources with mixed levels of medical training—bystanders, first-aiders, first-responders, or medical professionals—and a level of inconsistency was considered to be inherent as there was no reference to a final diagnosis. Finally, the reported incidents are not correlated with the total population that pursues outdoor recreation within a given region or during a specific time frame. As a result of these limitations, the findings should not be considered representative but instead may only portray a subset of the outdoor recreationist population.

Because of the study design, a cross-sectional methodology, no associations could be derived between exposures and outcomes. Given the inability to establish a well-defined population (inconsistent demographic data) and the large number of possible risk factors (leading to the need for a stratified analysis), it was believed that a retrospective analytic methodology could not be used. However, accessing better quality data sets in the future may allow investigations to examine potential relationships between risk factors (demographic, outdoor activity, region of the country, etc) and the likelihood of injury, illness, or death.

Figure 1 highlights the fluctuations that are present in the ACC data and underscores the difficulty when trying to interpret trends as a function of time, in particular the increase in casualties that occurred in 1999 and subsequent years. This was related to a new information source being added to the ACC database, ie, the British Columbia Search and Rescue reports became available online. Beyond this there are relatively large swings in the number of casualties between 1975 and 1999 on an almost cyclical basis for which the authors do not have a clear explanation. Given the manner in which data were collected by the ACC, it is likely that a selection bias was inherent.

The study serves to illustrate trends that may be encountered by SAR personnel within British Columbia and Alberta. This study also highlights the need to identify or create a data set that meets more robust standards and from which good quality epidemiology can be derived. Barring the creation of Canadian-focused SAR epidemiology, it is conceivable that using American-based studies would be appropriate and would assist in guiding Canadian SAR organizations. If we take, for example, the SAR operations occurring in British Columbia and Alberta, this likely shares more in common with trends that would be witnessed in Washington, Idaho, or Montana than it would with the Canadian Prairies, Arctic, or Atlantic Coastal regions. However, a number of federal, provincial, and regional level agencies are involved in SAR activities and may operate databases that could be used to further explore the geographic and seasonal variations of Canadian SAR operations.

We would like to thank the Alpine Club of Canada and, in particular, Edwina Podemski and Ernst Bergman for their assistance with accessing the database and understanding its content. Tapan Talavlikar was an asset with the construction of the GIS map.