Table 1.
Characteristics of the Studies Selected to Estimate Influenza-Associated Excess Mortality
| Characteristic | No. of Studies (n = 103)a | Reference(s) |
|---|---|---|
| Statistical modeling technique | ||
| Multiplier method | 4 | 8, 23–25 |
| Regression model with a proxy for influenza activity | 54 | 6, 13–16, 21, 26–45, 62, 64, 68, 69, 71–94 |
| Serfling-type model | 28 | 35, 38, 40, 48, 54, 61, 62, 88, 95–114 |
| Other | 28 | 20, 22, 31, 46–63, 88, 115–120 |
| Consideration of effects of time in regression models with an influenza activity proxy | ||
| Yes | 51 | 6, 13–16, 21, 26–45, 62, 64, 68, 69, 71–81, 83–88, 90–93 |
| No | 3 | 82, 89, 94 |
| Time lag between influenza activity and outcome in regression models | ||
| 0 week/month/year | 35 | 6, 13, 14, 21, 30, 32–35, 37–44, 62, 64, 71–73, 75, 77–80, 82–84, 86–90 |
| 1 week | 6 | 16, 27–29, 31, 81 |
| Otherb | 14 | 15, 26, 32, 36, 45, 68, 69, 74, 76, 85, 91–94 |
| Whether RSV activity was considered in the regression models with an influenza activity proxy | ||
| Yes | 18 | 6, 13, 16, 27, 28, 31, 32, 35–45 |
| No | 38 | 6, 14, 15, 21, 26, 29, 30, 32–34, 62, 64, 68, 69, 71–94 |
| Income level of country/region | ||
| Low | 1 | 60 |
| Middle | 16 | 15, 21, 26, 35, 43, 44, 60, 90, 94, 96, 99, 103, 105, 116–118 |
| High | 89 | 6, 13, 14, 16, 20, 21, 23–25, 27–42, 45–64, 68, 69, 71–89, 91–93, 97–102, 104–115, 119, 120 |
| Otherc | 4 | 8, 21, 22, 95 |
| Climatic zone | ||
| Temperate | 81 | 6, 13, 14, 20, 21, 23–25, 29, 35–42, 45, 48–64, 68, 69, 71–81, 83–94, 96–100, 104–119 |
| Tropical/subtropical | 25 | 15, 16, 21, 26–28, 30–35, 43, 44, 46, 47, 55, 60, 82, 99, 101–103, 105, 120 |
| Otherc | 4 | 8, 21, 22, 95 |
Abbreviation: RSV, respiratory syncytial virus.
a Only statistical methods used in the main analysis of each study were included in our analysis. Multiple statistical methods or models could have been used in a single study to estimate influenza-associated mortality; therefore, different methods/models were counted separately in our data analysis. Similarly, some studies could also have reported estimates for more than 1 country or region, so those studies would have been counted more than once in the table. Therefore, the subtotal in each category may differ from the total number of studies (n = 103).
b Refers to studies applying a lag of 2 weeks; studies applying multiple lags; studies applying a moving average proxy for influenza activity in week 0 through week −2 with equal weights or in week −1 through week −2 with unequal weights; and studies without an indication of a lag.
c Refers to studies reporting a combined estimate for multiple countries/regions.