Unraveling seasonality in population averages: an examination of seasonal variation in glucose levels in diabetes patients using a large population-based data set.

It has been shown that the population average blood glucose level of diabetes patients shows seasonal variation, with higher levels in the winter than summer. However, seasonality in the population averages could be due to a tendency in the individual to seasonal variation, or alternatively due to occasional high winter readings (spiking), with different individuals showing this increase in different winters. A method was developed to rule out spiking as the dominant pattern underlying the seasonal variation in the population averages. Three years of data from three community-serving laboratories in Israel were retrieved. Diabetes patients (N = 3243) with a blood glucose result every winter and summer over the study period were selected. For each individual, the following were calculated: seasonal average glucose for all winters and summers over the period of study (2006-2009), winter-summer difference for each adjacent winter-summer pair, and average of these five differences, an index of the degree of spikiness in the pattern of the six seasonal levels, and number of times out of five that each winter-summer difference was positive. Seasonal population averages were examined. The distribution of the individual's differences between adjacent seasons (winter minus summer) was examined and compared between subgroups. Seasonal population averages were reexamined in groups divided according to the index of the degree of spikiness in the individual's glucose pattern over the series of seasons. Seasonal population averages showed higher winter than summer levels. The overall median winter-summer difference on the individual level was 8 mg/dL (0.4 mmol/L). In 16.9% (95% confidence interval [CI]: 15.6-18.2%) of the population, all five winter-summer differences were positive versus 3.6% (95% CI: 3.0-4.2%) where all five winter-summer differences were negative. Seasonal variation in the population averages was not attenuated in the group having the lowest spikiness index; comparison of the distributions of the winter-summer differences in the high-, medium-, and low-spikiness groups showed no significant difference (p = .213). Therefore, seasonality in the population average blood glucose in diabetes patients is not just the result of occasional high measurements in different individuals in different winters, but presumably reflects a general periodic tendency in individuals for winter glucose levels to be higher than summer levels.