Figure 1

Identifying confounding factors in temporal diagnosis. A. Aggregated number of diagnoses from 1997 to 2009 (blue) show a strong increasing trend over time, modeled by the red line. When this trend is subtracted out, the remaining signal (magenta) shows no overall trend but a seasonal trend. B. De-trended total diagnoses display 6 month periodicity, as shown by the periodogram. When the years are plotted on top of each other (each colored line represents a year, with the bold black line the average), the spring and fall show consistent peaks in diagnosis each year. C. For each diagnosis, the number of cases is compared with the seasonal pattern in incidence. The most frequently occurring diagnoses show the most correlation with the overall spring-fall peak incidence; the overall trend causes false detection of periodic signal. Correlation between number of cases (x-axis) and correspondence with the overall spring-fall trend (y-axis) is 0.41.