Volume 17, Number 10—October 2011
Dispatch
Outbreak of West Nile Virus Infection in Greece, 2010
, Anna Papa, George Theocharopoulos, Georgios Dougas, Maria Athanasiou, Marios Detsis, Agoritsa Baka, Theodoros Lytras, Kassiani Mellou, Stefanos Bonovas, and Takis Panagiotopoulos
Table 2
Predictive factors of death for 197 patients with West Nile neuroinvasive disease analyzed by univariate and multivariate analysis, Greece, July–October 2010*
| Characteristic | No. deaths, n = 33† | Case-fatality rate, % | Crude risk ratio (95% CI) | Adjusted risk ratio‡ (95% CI) |
|---|---|---|---|---|
| Age group, y | ||||
| 40–59 | 1§ | 2.50 | Reference | Reference |
| 60–69 | 2 | 6.90 | 2.76 (0.26–28.99) | 2.72 (0.26–28.40) |
| 70–79 | 15 | 17.65 | 7.06 (0.97–51.59) | 6.13 (0.83–45.17) |
| >80 |
15 |
34.88 |
13.95 (1.93–100.9) |
11.41 (1.56–83.52) |
| Sex | ||||
| F | 10 | 11.36 | Reference | NA |
| M |
23 |
21.10 |
1.86 (0.93–3.69) |
NA |
| Underlying diseases | ||||
| None | 2 | 3.92 | Reference | NA |
| >1 |
31 |
21.23 |
5.41 (1.34–21.82) |
NA |
| Hypertension | ||||
| No | 19 | 15.70 | Reference | NA |
| Yes |
14 |
18.42 |
1.17 (0.63–2.20) |
NA |
| Heart disease | ||||
| No | 18 | 12.00 | Reference | Reference |
| Yes |
15 |
32.61 |
2.72 (1.49–4.95) |
2.03 (1.14–3.64) |
| Diabetes | ||||
| No | 23 | 15.44 | Reference | NA |
| Yes |
10 |
20.83 |
1.35 (0.69–2.63) |
NA |
| Immunosuppression | ||||
| No | 31 | 17.42 | Reference | NA |
| Yes |
2 |
10.53 |
0.60 (0.16–2.33) |
NA |
| Cancer | ||||
| No | 29 | 16.11 | Reference | NA |
| Yes |
4 |
23.53 |
1.46 (0.58–3.66) |
NA |
| Stroke | ||||
| No | 28 | 15.30 | Reference | NA |
| Yes |
5 |
35.71 |
2.33 (1.07–5.10) |
NA |
| Renal failure | ||||
| No | 31 | 16.40 | Reference | NA |
| Yes | 2 | 25.00 | 1.52 (0.44–5.28) | NA |
*CI, confidence interval; NA, not applicable.
†Two additional patients with nonneuroinvasive disease died, and those deaths were not included in this analysis.
‡In logistic regression analysis, initial models included all variables for which the p value was <0.05 or the odds ratio was >1.1 or <0.90. Therefore, all variables were included in the initial models. Variables were removed 1 at a time depending on results of statistical testing (p<0.05), by using the likelihood-ratio test. All variables that remained significant in the final logistic regression model were included in the binomial regression model for the estimation of adjusted risk ratios.
§ Belonged to the 40–49-year age group.