Plain Talk on Fatal Accident Rates

Mon, Apr 21, 2008 — David Evans


And what they do not reveal in terms of net hazard

Aviation safety has never been better, claimed Nicholas Sabatini at a 10 April 2008 hearing of the U.S. Senate Commerce Subcommittee on Aviation. The essence of Sabatini’s argument is presented below, followed by a few comments about the potentially misleading aspect of his remarks:

“Some people may dismiss claims like ‘this is the safest period ever’ because they have heard this claim in the past. For at least the past 70 years, aviation safety has improved by a third or more every decade. In fact, the pace of improvement has accelerated recently and we believe the pace of improvement will continue to accelerate for the next decade or more.

“This context is important. Over the past five years, on-board fatalities have occurred at a rate of about 1 fatal accident in every 15 million passenger flights. We see no reason why that figure cannot become one in 30 million or even one in 40 million flights within 10 or 15 years. The system’s performance is now so strong that we decided several years ago to develop a new measure to express the risk of fatality in commercial aviation. In addition to traditional data on fatal accidents per 100,000 flight hours or 100,000 departures, the FAA now uses fatalities per 100 million persons flown as a basic measure of the system’s performance. This includes all fatalities, whether they occur onboard a passenger or cargo flight, or whether they occur off the aircraft on the airport or elsewhere. [The logic of these last two sentences must be questioned. Sabatini first states the measure is by the number of persons flown, which means occupants of the aircraft, then he says the measure includes fatalities occurring off the aircraft, which could include service personnel killed on the ramp.}

“To offer a sense of scale, immediately after World War II, that measure yielded nearly 1,500 fatalities per 100 million persons flown. By the early 1960s, the measure had improved to about 500 fatalities per 100 million persons flown. By the mid-1990s, that measure had fallen to about 45 fatalities per 100 million persons flown.

“Now, in a typical year, e experience rates of 5 to 8 fatalities per 100 million persons flown within the next decade. By comparing that level of safety to where we were just 20 years ago, or even a decade ago, we begin to get some sense of how safe the system has become – and it will continue to get better over the long run.”

To Sabatini’s rosy view, a few comments are in order. First, relating safety to 100 million persons flown pumps up the denominator. Airliners can accommodate 100-500 passengers per flight. One could argue that by going to this measure of so many persons flown, the FAA has instantly “improved” air safety while not reducing accidents at all. The traveling public certainly does not measure safety in terms of fatalities per 100 million persons aboard. For the average passenger, the rate of accidents per week is probably the more relevant.

Second, Sabatini prefers to count only fatal accidents, thereby eliminating accidents involving personal injuries and damage to aircraft. The National Transportation Safety Board (NTSB) maintains a database of all accidents, fatal and otherwise. For the most recent year in which the NTSB has compiled statistics, 24 airline accidents are shown for 2007, non fatal. In 2006, only two of the airliner accidents were fatal but, as for 2007, the toll includes hull losses where nobody was killed or injured or the airplane was damaged beyond repair. And, by the way, there were fewer fatalities in 2003 than in 2006, but there were 51 total accidents. So whether one is using 2003 or 2006/2007 as a baseline, on average accidents were occurring about once a week or, more recently, about once every two weeks.

Third, Sabatini is counting only airliner fatalities. What was left out? General aviation (GA), for which the NTSB also maintains a database. In this category, there were 1,631 accidents in 2007 in which 491 were killed. On average, about 31 accidents were occurring each week, or about 4 per day – killing 1.30 persons per day, on average. It should also be mentioned that the 491 killed in GA accidents is the equivalent of roughly two widebody jets worth of passengers. It is fair to argue that if two widebody jets crashed, there would be much hand-wringing over the crisis in aviation safety. Yet a similar total loss of life, accumulated over far more accidents (because smaller GA airplanes carry fewer people), does not raise similar alarms.

Fourth, Sabatini is not including incidents in his account. An incident might be described as an accident that got lucky; but for a few seconds or few feet difference, the outcome could have been far worse.

It would seem that a couple things need to be done. The most important is to get the FAA on the same page as the NTSB and count accidents the same way. The NTSB accident database ought to be the repository of governing statistics. The second thing is to address safety from two perspectives: airliners and GA flights. Both statistics are needed to assess the safety of the aviation system.

Most importantly, a database of incidents needs to be created. Uniformly compiled incident statistics are not maintained by the FAA or the NTSB, but these events are the precursors to accidents and are of themselves a useful measure of safety.

Comments are closed.

Nolan Law Group