tornado deaths

Summary: There is no statistically significant change to tornado frequencies and there has been substantial progress in reducing the death rate due to tornadoes.  For my healthcare readers, there may be insight into how the public perceives and responds to risk in the analyses and procedures used for tornado warnings.  It's a very different situation, so it's not easy to say how this affects healthcare.

This year had another burst of tornadoes and tornado deaths.  The occurence of tornados is tremendously variable.  There are bursts like this in among all the noise.  There is no significant trend.  The following figures are from the work of Harold Brooks, of the National Severe Storms Laboratory (NSSL).  He did a lot of analysis in response to the 1998 outbreak, and has kept up the study of tornadoes and harm mitigation.

First, the occurrence of severe tornadoes.  This shows a count of tornadoes that caused a death.  Using this definition gets usable data back to about 1875.  The tornado intensity scales were not established until recently.  With a highly noisy signal like this, you need more than a few decades to see a trend.  The records for cause of death go back reasonably well into the late 1800's. Before then you can see that either there were almost no tornadoes (unlikely) or record keeping was inadequate (the real problem).  There is no visible trend line, and statistics confirm that there is no trend in tornado frequency over that period.

The actual death rate per capita shows really substantial improvement starting in the 1920's.  The reasons include:

• Better reporting (telephone, radio, cell phone, police networks)
• Better warning technologies (telephone, sirens, radio, television)
• Better warning psychology (actual human reactions are studied)
• Better detection before they hit the ground (radar)
• Better forecasting tools (computers, radar)
• Better buildings (safe rooms, overall better construction)
• Better medicine (antibiotics, trauma treatment improvements)

The data is insufficient to allocate the improvement against causes.  But it's very dramatic:

The unadjusted death rates also show an improvement.  Despite the suburban sprawl converting farmland into houses, the death rates also fall.  (The property damage figures show no improvement.  The greatly increased area with houses and the much higher value for houses has meant a staggering increase in property damage costs.)

 

So, no climate effect for major tornadoes.

There is a lot of good understanding of human psychology, reactions to risk, reactions to disaster, etc.  The biggest simple learning is that people do not react as shown on TV or as expected by pundits.  It's like some other areas where common sense is wrong.  Human reactions do make sense once you accept the observed behavior model. 

For example, there is a lot of obsessing about specificity with warnings.  The reality is that there will always be false alarms.  Even a real tornado can be perceived as a false alarm if the warning is not phrased right.  And there will be tornadoes that are not detected in time for a warning.  Fortunately, as "acts of god" the lawyers haven't yet demanded perfection in forecasts and warnings.  Which means that the meteorologists can have open and frank discussion of the proper level of false alarms for most effective harm reduction.  Too much demand for perfection means failure to raise the alarm.  Too many false alarms means failure to take protective measures.

Television affects the public future response.  The decision about whether an alarm was "true" or "false" seems to be affected by whether there are dramatic pictures of a nearby tornado.  Without those pictures, more alarms are viewed as false alarms.

Building construction, especially safe room details, makes a big difference also.  Immediate warnings are much less effective at night because people are asleep.  Safe rooms that are usable for sleep when there are potential tornados make a difference.

It's quite interesting for the insights into how people behave.  I've not yet found any simple lessons for healthcare beyond the first simple rule of human factors:

Common sense is wrong.  Relax your preconceptions and accept that you need to study what people really do and how they really react in the real world.