Table 1.
Comparison of early alert dates identified by statistical process control charts for full-season charting and modeled real-time charting of positive rapid influenza tests in the UUPCRN population for 4 seasons.
| Influenza season | Date epidemic started | Full-season charting | Modeled real-time charting | ||
|---|---|---|---|---|---|
| Early alert* date relative to epidemic onset (days) |
No. visits to trigger early alert* (% of visits) |
Early alert† date relative to epidemic onset (days) |
No. visits to trigger early alert†
(% of visits) |
||
| 2004-05 | Jan 13 | −13 | 2 (2.7%) | −17 | 2 (0.3%) |
| 2005-06 | Dec 13 | −2 | 3 (3.6%) | −6 | 2 (0.2%) |
| 2006-07 | Feb 2 | +2 | 2 (2.4%) | −15 | 4 (0.4%) |
| 2007-08 | Feb 3 | −15 | 3 (2.4%) | −24 | 2 (0.2%) |
|
| |||||
| Mean‡ (sd) | −7.0 (8.3) | −15.5 (7.4) | |||
| Median | −7.5 | −16.0 | |||
*Early alert occurred on the date that rapid test positivity exceeded 3 sigma.
†Early alert occurred on the date that rapid test positivity exceeded 3 sigma and did not return to the center line on the following day.
‡One-sided t-test of means was significant (P = 0.039).