Figure 1.
CT – computed tomography; DVT – deep vein thrombosis; SOB – shortness-of-breath; VTE – venous thromboembolism.
Table 1.
Case characteristics.
Figure 2.
Odds ratio for each month versus winter.
Data are shown with 95% confidence intervals and red denotes a statistically significant difference (P<0.05) versus winter (defined as the average number of cases in December, January, and February). (A) Detroit VTE. (B) Detroit no-VTE. (C) Suburban VTE. (D) Suburban no-VTE.
Figure 3.
Distribution of monthly venous thromboembolism (VTE) and no-VTE cases.
X-axis indicates the month of the year (January = 1). Dashed line indicates uniform distribution. (A) Detroit VTE (n = 1,490). This was the only distribution that differed from uniform (P = 0.0046; Rayleigh test). Distribution mean (red arrow) June 24th; the 95% confidence intervals (May 19th to July 31st) are indicated by the shaded rectangle. (B) Distribution of Detroit cases that did not have a VTE (n = 1,123); did not differ from uniform (P = 0.50), but differed from the Detroit VTE cases (P<0.02; Kuiper’s test). (C) Suburban VTE (n = 417); did not differ from uniform (P = 0.13). (D) Suburban cases that did not have a VTE (n = 280); did not differ from uniform (P = 0.21) or the suburban VTE distribution.
Table 2.
Seasonal distribution of venous thromboembolism in Detroit as a function of potential risk factors.
Figure 4.
Periodic regression of venous thromboembolism (VTE) cases (Detroit).
Black circles indicate the monthly averages. The periodic regression is shown in red (P = 0.028; R2 = 0.55); June 30th peak. Periodic mean = 124.2 cases/month (dashed red line).
Figure 5.
Delay from self-reported symptom onset to emergency department arrival.
(A) All cases with recorded times (n = 1,504). The mode (chequered bar; 31%) occurred at 3–6 days; however, the delay exceeded one week for more than 25% of cases (black bars). (B) Pulmonary embolus (n = 557). The mode (29%) occurred in the <1 day period; proportion with delays exceeding one week was 17%. (C) Deep vein thrombosis (n = 947). The mode (35%) occurred at 3–6 days; proportion with delays exceeding one week was 32%. The distribution for deep vein thrombosis differed from that for pulmonary embolus (P<0.001; Kuiper’s two sample test).
Figure 6.
Weekly distribution of particulate matter (PM) parameters and venous thromboembolism (VTE) cases (Detroit hospitals).
Periodic regressions are shown in red. (A) 2.5 µm PM (P = 0.049; R2 = 0.12); January 26th peak. (B) 10 µm PM (P = 0.0003; R2 = 0.28); July 2nd peak. (C) Coarse particles (P<0.0001; R2 = 0.40); July 11th peak. (D) VTE cases per week per year (P = 0.018; R2 = 0.15); July 3rd peak.
Figure 7.
Distribution of non-idiopathic venous thromboembolism (VTE).
(A) Monthly distribution (January = 1): dashed line indicates uniform distribution. Distribution differed from uniform (P<0.05 Kuiper’s test; P = 0.0004 Rayleigh test); mean (red arrow) = July 21st (shaded rectangle indicates the 95% confidence intervals; June 21st to August 20th). (B) Weekly distribution. Solid red line shows the periodic regression (P = 0.0004; R2 = 0.27); July 25th peak.