Table 2.
Final negative binomial regression (‘estimation’) model for monthly Ross River virus notifications in the Mildura Local Government Area, Victoria, Australia, July 2000–June 2011
| Variable (units) | Lag (months) | Levels | IRR | s.e. | 95% CI | P value |
|---|---|---|---|---|---|---|
| log2 (maximum vapour pressure, hPa) | 1 | 3·47 | 1·40 | 1·57–7·66 | 0·002 | |
| Culex annulirostris (mean trap count) | 1 | ⩾1000 | 7·07 | 3·37 | 2·78–18·0 | <0·001 |
| 10–999 | 1·83 | 0·46 | 1·12–2·98 | |||
| <10 | 1·00 | – | (ref.) | |||
| Aedes camptorhynchus (mean trap count) | 2 | ⩾10 | 1·55 | 0·23 | 1·16–2·09 | 0·003 |
| <10 | 1·00 | – | (ref.) | |||
| log2 (precipitation, mm) | 4 | 1·25 | 0·08 | 1·09–1·42 | 0·001 | |
| Number of days with precipitation >1 mm | 6 | 1·08 | 0·04 | 1·00–1·17 | 0·044 | |
| Interaction terms | ||||||
| Maximum Murray river flow × Southern Oscillation Index | ||||||
| RIVER | 3 | ⩾16 268 ML | 0·15 | 0·13 | 0·03–0·81 | 0·027 |
| SOI | 6 | ⩾1·7 | 1·56 | 0·50 | 0·83–2·94 | 0·167 |
| −5·4–1·7 | 1·99 | 0·66 | 1·04–3·80 | 0·037 | ||
| <−5·4 | 1·00 | – | (ref.) | |||
| RIVER × SOI | ⩾1·7 | 5·65 | 4·69 | 1·11–28·7 | 0·037 | |
| −5·4–1·7 | 0·33 | 0·42 | 0·03–4·10 | 0·385 | ||
| Sea surface temperature × minimum sea level | ||||||
| SST | 2 | ⩾26·8 °C | 0·32 | 0·09 | 0·19–0·55 | <0·001 |
| SEALVLmin | 7 | ⩾13·2 cm | 0·87 | 0·25 | 0·49–1·55 | 0·644 |
| SST × SEALVLmin | 3·99 | 1·72 | 1·71–9·29 | 0·001 | ||
| Est. | s.e. | 95% CI | P value* | |||
| Dispersion parameter (α) | 0·04 | 0·03 | 0·01–0·21 | 0·009 |
IRR, Incidence rate ratio; s.e., standard error of IRR or dispersion parameter; CI, confidence interval.; Est., point estimate of dispersion parameter.
AIC = 318·564, n = 132, degrees of freedom = 16, log likelihood (model = −143·282; null model = –226·598), maximum likelihood R2 = 0·717, deviance-based goodness-of-fit (P = 0·50).
* P value for dispersion parameter estimated using a likelihood ratio test that α is non-zero.