To the Editor:
We thank Dr Fortini et al and Dr Van Tilburg for their thoughtful feedback. We agree that although our model provides the estimate of 51 hours for when to consider the termination of a search and rescue mission, it cannot be used alone to determine when to stop a search. There are many factors that search managers must take into consideration when confronted with “real-life” situations that we simply could not account for in our models given the source of the data. We believe that a refinement of this initial estimate is warranted, perhaps by collecting data prospectively and including information on both lost person-related and environment-related factors. Although we believe our methodology for building and validating the model used for this analysis is valid, we agree that numbers can be presented many different ways with the potential for differing interpretations. Regardless of the statistical analysis undertaken, it is important to bear in mind that this time value should be interpreted as a guideline to better inform the search effort, rather than an absolute stopping point.
Some of what troubles Dr Fortini et al and Dr Van Tilburg is, we believe, the ongoing tension between a population-based interpretation of results vs interpretations relevant to an individual lost person. From our data, by the end of the 50th hour of searching, 99.3% of all survivors and 98.8% of all lost persons had already been located. Those are good results both mathematically and practically. However, the 41 real people who were still missing at the end of the 50th hour (1.2% of all missing) and their friends and families and the rescuers will not be comforted by these numbers. All that matters to them, very understandably, is the 1 person who hasn’t yet been found. And of those remaining 41 people, 24 (59%) were later found alive. As no decision rule will consistently identify 100% of the group of interest (in this case, survivors), there will periodically be cases that defy the odds (thankfully). We agree that when there is some possibility of survival, this possibility (however remote) may be enough to continue searching. Generally, these decisions are made on a case-by-case basis, with many variables factoring into the decision-making process. As noted by Dr Van Tilburg, there have been several recent examples in Oregon. However, there is frequently a balance to be sought between available resources, the safety of search personnel, financing, and likelihood of survival. Assuming a fixed amount of resources, personnel, and funding to serve an entire region, it may not always be possible to continue a search without a reasonable chance of survival.
Dr Fortini et al correctly point out that at 51 hours, the survival probability remains relatively high, which he argues does not make continued searching “futile.” We agree. However, we would expect survival probability at each time point to remain high and do observe so (Table). Survival is a more common outcome than death in this cohort (94% of all lost persons survived); therefore, survival at any time point remains the more likely outcome than not.
Survival probabilities for lost persons, by 10-hour search intervals
We acknowledge that a statistical model does not and cannot substitute for real-life experience and human judgment and cannot account for the emotional value we all place on saving a single life whenever possible. The pressures to continue searches indefinitely can be tremendous, as Dr Van Tilburg correctly points out. We recognize that rare events do indeed happen—people can survive extreme circumstances beyond reasonable expectation.
We are glad that our study has added to this debate and sparked continued interest in the topic, and we greatly appreciate the interest and comments provided. We hope that the search and rescue community will continue to explore these important questions and that, with improved measurement of critical factors related to real-life search management and decision making, will point to a clearer and better answer to such critical questions.