I was thinking about our incredibly warm winter during lunch today.  When I got back to my computer, I saw an interesting study.  The study by the Kessler foundation was released this week showing warmer temperatures are associated with worse cognitive statuses of people with MS. http://kesslerfoundation.org/media/displaynews.php?id=220

While this information comes as no surprise to me, somebody who barely functions when warm, it does pose some interesting methodological problems for further studies or clinical trials.  The variation of individual reactions to temperatures may make larger studies even more important despite the added costs.  If a study has only a minimum number of patients thought to allow the determination of a technically statistically significant finding from a math point of view, it may in fact be reporting the significance of a small number of people’s  reaction to a difference in temperature or season.

If one could have enough people in a study, then one could adjust the results by the average impact the temperatures would have caused.  In other surveys, this same approach is done and referred to as applying a “seasonal adjustment.”  If one thinks of hiring, the term takes on a more obvious role than in a medical study.  Stores hire sales people for the Christmas season.  So when they say, “the hiring for Nov is up 5% from October Seasonally adjusted” it doesn’t just mean the economy added 5% more jobs.  If the economy normally adds 10% on average from October to November, the 5% is cumulative with the 10% normal boost.  Depending on methodology the 5% increase may be before or after the adjustment.

This article makes a case for such an adjustment when looking at MS stats and clinical trials.  If one takes this information at face value, I would be trying to run my clinical trials for new drugs/procedures during the coldest times of the year in the coldest places as my end points so I could hide/offset some negatives in the positive impact of the change in temperature.  Of note, the studies used for this were small with 40 MS patients for one time period and 45 MS patients from a different pool at the second point in time.  There are some with MS who can’t take the cold but are fine in the heat, the opposite of my reaction.  Had there been a couple of them in the trial, the results could have come back opposite what is being published here.

I am not saying I disbelieve the results or think they don’t point to some thing of importance when trying to determine the validity of different tests and trials.  In fact, based on personal experience, I would believe the results most likely valid enough to warrant further testing to determine a seasonal adjustment.  In a large study, this type of seasonal adjustment may lead to more correct identification of significant trends rather than those trends caused solely by weather.

One last note on the structure of the study’s results: they are reporting a significant correlation.  There is a saying “correlation does not equal causation.”  For example, in a NY City study, ice cream sales increased as outdoor violent crime increased.  Does ice cream cause us to murder each other?  We may all scream for ice cream, but murder?  Isn’t it just as plausible to think people buy and eat ice cream during the spring, summer, and fall when going outside is comfortable.  Then they buy less ice cream in the winter when they stay inside and thus aren’t outside to either perpetrate violent crimes or be victims of them? (note the seasonal adjustment needed for even this study)