Volume 26, Number 11—November 2020
Research
Nowcasting (Short-Term Forecasting) of Influenza Epidemics in Local Settings, Sweden, 2008–2019
, Olle Eriksson, Örjan Dahlström, Benjamin John Cowling, Matthew Biggerstaff, Gunnar Ljunggren, Anna Jöud, Emanuel Istefan, and Toomas Timpka
Table 2
Performance of peak-timing and peak-intensity predictions from evaluation of nowcasting for detection and prediction of local influenza epidemics, Sweden, 2008–2019
| Influenza virus activity | Time-to-peak* |
Peak-intensity category, cases/day/100,000 population†§ |
||||||
|---|---|---|---|---|---|---|---|---|
| Prediction date | Predicted | Error | Interpretation | Predicted | Factual | Interpretation | ||
| 2009 A(H1N1) | ||||||||
| Stockholm | 2009 Sep 13 | 8 | 56 | Poor | Medium (5.0) | Very high (12.4) | Poor | |
| West Gothia | 2009 Sep 13 | 8 | 56 | Poor | Low (2.2) | Very high (13.7) | Poor | |
| Scania |
2009 Sep 25 |
10 |
42 |
Poor |
Low (1.4) |
High (6.4) |
Poor |
|
| 2010–11 A(H1N1) and B | ||||||||
| Stockholm | 2011 Jan 14 | 10 | 28 | Poor | Medium (3.4) | Medium (3.5) | Excellent | |
| West Gothia | 2011 Jan 14 | 10 | 14 | Poor | Medium (4.3) | High (6.1) | Tolerable | |
| Scania |
2011 Jan 10 |
11 |
22 |
Poor |
Medium (2.9) |
High (5.5) |
Poor |
|
| 2011–12 A(H3N2) | ||||||||
| Stockholm | 2012 Feb 27 | 8 | −8 | Tolerable | High (7.4) | Very high (9.4) | Good | |
| West Gothia | 2012 Feb 27 | 8 | −8 | Tolerable | High (7.8) | Very high (9.6) | Good | |
| Scania |
2012 Feb 27 |
8 |
−8 |
Tolerable |
Medium (4.0) |
High (6.8) |
Poor |
|
| 2012–13 A(H3N2), A(H1N1), and B | ||||||||
| Stockholm | 2013 Feb 10 | 8 | −7 | Good | Very high (10.3) | Very high (12.2) | Excellent | |
| West Gothia | 2013 Feb 10 | 8 | −7 | Good | Very high (10.3) | Very high (11.9) | Excellent | |
| Scania |
2019 Feb 8 |
10 |
−7 |
Good |
High (7.3) |
Very high (10.7) |
Good |
|
| 2013–14 A(H3N2), A(H1N1), and B | ||||||||
| Stockholm | 2014 Feb 16 | 8 | −7 | Good | Medium (2.7) | Medium (3.0) | Excellent | |
| West Gothia | 2014 Feb 16 | 8 | −7 | Good | Medium (3.5) | Medium (2.9) | Excellent | |
| Scania |
2014 Feb 17 |
8 |
−1 |
Excellent |
Medium (3.2) |
Medium (4.2) |
Excellent |
|
| 2014–15 A(H3N2) and B | ||||||||
| Stockholm | 2015 Feb 22 | 8 | 6 | Good | Medium (4.5) | High (6.5) | Tolerable | |
| West Gothia | 2015 Feb 22 | 8 | 6 | Good | Very high (7.9) | Very high (8.3) | Excellent | |
| Scania |
2015 Feb 14 |
9 |
0 |
Excellent |
Medium (3.9) |
Very high (8.1) |
Poor |
|
| 2015–16 A(H1N1) and B | ||||||||
| Stockholm | 2016 Feb 7 | 8 | 0 | Excellent | High (6.7) | Very high (8.2) | Tolerable | |
| West Gothia | 2016 Feb 7 | 8 | 7 | Good | High (7.6) | Very high (11.6) | Good | |
| Scania |
2016 Feb 6 |
9 |
7 |
Good |
Medium (4.3) |
Very high (10.4) |
Poor |
|
| 2016–17 A(H3N2) | ||||||||
| Stockholm | 2017 Jan 1 | 8 | −7 | Good | Very high (8.2) | High (6.8) | Good | |
| West Gothia | 2017 Feb 12 | 8 | 7 | Good | Medium (3.3) | Medium (3.7) | Excellent | |
| Scania |
2017 Feb 5 |
8 |
14 |
Poor |
Medium (4.2) |
Medium (5.1) |
Excellent |
|
| 2017–18 A(H3N2) and B | ||||||||
| Stockholm | 2018 Feb 18 | 8 | −7 | Good | Very high (14.4) | Very high (11.6) | Excellent | |
| West Gothia | 2018 Feb 18 | 8 | 0 | Excellent | Medium (5.2) | High (5.9) | Good | |
| Scania |
2018 Feb 4 |
8 |
14 |
Poor |
Medium (4.2) |
Very high (14.0) |
Poor |
|
| 2018–19 A(H1N1) | ||||||||
| Stockholm | 2019 Feb 3 | 8 | 0 | Excellent | Very high (14.4) | High (6.2) | Poor | |
| West Gothia | 2019 Feb 3 | 8 | 7 | Good | Medium (4.0) | Medium (3.4) | Excellent | |
| Scania | 2019 Feb 3 | 8 | −7 | Good | Medium (2.8) | Medium (5.2) | Excellent | |
*Time-to-peak (days) determined using syndromic telenursing data. Positive value means that the peak was predicted to be reached before the actual peak occurs, whereas negative value means that the peak is predicted after the actual peak occurs. †Peak-intensity category determined using clinical influenza-diagnosis data. §Using clinical influenza data (Table 1), the start of the epidemic was detected on December 27. On February 1, using syndromic data, the peak in clinical influenza data was forecasted to occur 8 days later (February 9), but the peak actually occurred on February 2 (7 days earlier than forecasted). Also, on February 1, the clinical influenza data intensity was forecasted to be high.