An epidemiologist and daylight savings time

If you check the blog for tomorrow's Sunday Sermonette, it will be an hour earlier, astronomically speaking. That's because in the US the clocks are shifted forward by an hour for "Day Light Savings Time," starting at 2 a.m. tomorrow morning (before the Sermonette goes up). The time shift will last until November 7, next fall. The sudden discontinuity in time keeping has uses other than energy conservation, however, and clever epidemiologists have used such circumstances for their own purposes for years. Here's one example.

One of the nice things about the switch back in the fall is that you get an extra hour of sleep (in theory at least). Conversely, in the spring you lose an hour of sleep time. So Lambe and Cummings reasoned that on the Monday after the switch in the spring, people are more likely to have less sleep (on average) than the Monday the week before the switch. They looked at car crashes in Sweden on those two Mondays in the years between 1984 and 1995. The 4% increase in crash rate after the change was within what might be expected from usual variation, so either the effect is real but small or there is no effect at all (see Lambe and Cummings, "The shift to and from daylight savings time and motor vehicle crashes," Accid Anal Prev. 2000 Jul;32(4):609-11).

This is not the only epidemiological study that used the time shift to study things like sleep deprivation or ambient light at drive time, nor is it necessarily the best. We didn't do a thorough search because all we wanted to do is make a simple point. One of things epidemiologists like myself do is look around in the world for natural circumstances that are like experiments, observe them in just the right way, and arrange our observations to give us the most reliable information (often this means using statistical methods). While we can do experiments on sleep deprivation or ambient lighting in a driving simulator and get valuable information, it is still important to see what happens in the natural setting where experiments are impossible. Is the information consistent with what we saw with the simulations? If not, why not? What does it suggest we look at next? No single piece of information, no matter how good the study, is likely to settle the point (unless the effect is so powerful there is no question). Instead we have to put together information from different sources, each with different strengths and weaknesses, triangulating to arrive at a "final" answer (which in science really means a "provisional" answer). It's like putting together pieces of a jigsaw puzzle where you aren't sure if any particular piece is even part of the picture or if it is, what part it is. You need other pieces that complement it and you hope that eventually you will be able to get a glimpse of what the puzzle is portraying.

So for most people, tomorrow's time shift is something they'll have to take into account to negotiate their lives for a day or so. Remember, though, that some epidemiologist it might be a scientific opportunity.

And drive carefully.