Time series studies relate the day-to-day variation in air pollution to the day-to-day variation in numbers of
deaths, hospital admissions, patients having symptoms etc. It has been argued that if air pollution advances death
only by a few days or weeks in extremely frail individuals, the public health implications of the relationships seen in
time series studies may be limited (14). Several investigators have now explored whether indeed this so-called
‘harvesting’ applies. If so, using longer averaging times in the time series would lead to reduction or even
disappearance of the relationship between air pollution and mortality: subjects who were dying a few days or weeks
early would not be around to die at their expected point in time. Thus, some days or weeks after high pollution days,
fewer people would die than expected, canceling out the excess observed after the high pollution days. In contrast to
this, using longer averaging times has produced higher rather than lower effect estimates (15). Still, the effect
estimates from time series studies are less than those from long-term studies, suggesting that some of the long-term
effects are not captured by variations of exposure in time over periods of weeks or months