Table 1.
Comparing observed vs. predicted maximum peaks in daily ILI-related ED visit volumes, in terms of magnitude and timing, Edmonton, Alberta, 2008–2014
| Time period | Median no. of visits/day | Observed peak ILI-related visit volume and date | Predicteda peak ILI-related visit volume and date | Difference between observed and predicteda maximum peak ILI-related volumes (relative %) and date | No. (%) of the 7b days ahead of the seasonal peak when predicted volume was within 30% of observed volume | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Non-parametric method | Model 2: GLMM | Model 3: GLMM and locally-predicted residual | Main Model | ||||||||
| Pre-H1N1 (1 Aug 08–31 Mar 09) | 19 | 52 47 |
Thu, Jan 1/09 Mon, Mar 2/09 |
42.5 47.6 |
Sun, Dec 28/08 Sun, Feb 22/09 |
−9.5 (−18.2%) 0.6 (1.2%) |
−4 days −8 days |
9 (64.3) | 9 (64.3) | 11 (78.6) | 11 (78.6) |
| H1N1 Wave 1 (1 Apr 09–31 Jul 09) | 25 | 57 | Sun, May 3/09 | 58.9 | Mon, May 4/09 | 1.9 (3.3%) | +1 day | 5 (71.4) | 1 (14.3) | 2 (28.6) | 3 (42.9) |
| H1N1 Wave 2 (1 Oct 09 –5 Dec 09) | 50 | 275 | Wed, Oct 28/09 | 297.3 | Sun, Nov 1/09 | 22.3 (8.1%) | +4 days | 2 (28.6) | 0 (0.0) | 0 (0.0) | 3 (42.9) |
| Post-H1N1 (6 Dec 09–31 Jul 10) | 20.5 | 51 | Tue, Mar 2/10 | 50.2 | Sat, Mar 6/10 | −0.8 (−1.6%) | +4 days | 3 (42.9) | 7 (100) | 5 (71.4) | 5 (71.4) |
| 2010–11 (1 Aug 10–31 Jul 11) | 21 | 62 | Sun, Dec 26/10 | 48.7 | Sun, Dec 26/10 | −13.3 (−21.4%) | 0 days | 2 (28.6) | 5 (71.4) | 5 (71.4) | 4 (57.1) |
| 2011–12 (1 Aug 11–31 Jul 12) | 22.5 | 59 54 |
Sun, Jan 1/12 Tue, Feb 28/12 |
50.5 53.9 |
Mon, Dec 26/11 Sat, Mar 3/12 |
−8.5 (−14.4%) −0.1 (−0.3%) |
−6 days +4 days |
7 (50.0) | 14 (100) | 14 (100) | 12 (85.7) |
| 2012–13 (1 Aug 12–31 Jul 13) | 26 | 125 | Wed, Dec 26/12 |
145.6 | Sun, Dec 30/12 | 20.6 (16.5%) | +4 days | 6 (85.7) | 0 (0.0) | 1 (14.3) | 7 (100) |
| 2013–14 (1 Aug 13–15 Feb 14) | 32 | 93 | Sun, Dec 29/13 | 110.3 | Sun, Dec 29/13 | 17.3 (18.6%) | 0 days | 4 (57.1) | 0 (0.0) | 1 (14.3) | 6 (85.7) |
a
Predicted volumes are based on our Main Model.
b
The pre-H1N1 and 2011–2012 season each had two similarly sized maximum peak volumes that occurred at notably distinct times within each of these periods; therefore, we examined the combined no. (%) over the 7 days ahead of each of these peaks (14 days total).