Table 2.
Multivariable Analysis to Test the Effect of Different Covariates on the Risk of Developing Febrile Malaria
| Covariate | Hazard Ratio | Robust SE | z | P>|z| | Confidence Interval | |
|---|---|---|---|---|---|---|
| Lower | Upper | |||||
| Main | ||||||
| Transmission (high vs low) | 0.98 | 0.35 | −0.05 | .957 | 0.49 | 1.96 |
| Transmission (high vs mod-high) | 16.55 | 5.43 | 8.55 | <.0001 | 8.70 | 31.49 |
| Transformed age | 5.29 | 1.50 | 5.86 | <.0001 | 3.03 | 9.22 |
| Transmission (high vs low) × transformed age | 0.60 | 0.21 | −1.43 | .154 | 0.30 | 1.21 |
| Transmission (high vs mod-high) × transformed age | 0.24 | 0.07 | −4.59 | <.0001 | 0.13 | 0.44 |
| Infection status (uninfected vs asymptomatic) | 0.21 | 0.07 | −4.97 | <.0001 | 0.12 | 0.39 |
| Transmission (high vs low) × infection status (uninfected vs asymptomatic) | 2.43 | 0.45 | 4.81 | <.0001 | 1.69 | 3.49 |
| Transmission (high vs mod-high) × infection status (uninfected vs asymptomatic) | 0.25 | 0.07 | −5.29 | <.0001 | 0.15 | 0.42 |
| Infection status (uninfected vs asymptomatic) × transformed age | 3.32 | 0.93 | 4.27 | <.0001 | 1.91 | 5.76 |
| Year of survey | 0.95 | 0.01 | −5.72 | <.0001 | 0.93 | 0.97 |
| Infection status (uninfected vs asymptomatic) × year of survey | 1.12 | 0.02 | 5.35 | <.0001 | 1.07 | 1.17 |
| Time varying covariates | ||||||
| Infection status (uninfected vs asymptomatic) | 0.69 | 0.08 | −3.39 | 0.001 | 0.56 | 0.85 |
| Year of survey | 0.94 | 0.01 | −7.34 | <0.0001 | 0.92 | 0.95 |
Abbreviations: mod-high, moderate-high; SE, standard error.
This final model shows the effect of the different covariates on developing febrile episodes with alternative cutoff times of <90 and >90 days. The P values in bold represent those that were statistically significant (P < .05). The symbol × indicates an interaction between the respective covariates.