Impact of aerial humidity on seasonal malaria: an ecological study in Zambia

Malaria Journal

Table 3 Final negative binomial models derived from backward selection of full model variables

	Model	AIC	P value					IRR
	Model	AIC	Humidity	Temperature	Year	Elevation	Rain	Humidity	Temperature	Rain
Final Whole Year Tree Models	No lag	1409.8	1.22 E−17	2.03 E−03	1.62 E−04	NA	0.891	1.111	1.074	1.000
	1 week lag	1457.9	3.23 E−16	5.20 E−09	4.35 E−05	NA	0.664	1.097	1.143	1.000
	2 week lag	1498.5	1.70 E−14	3.22 E−12	2.68 E−04	0.2750	0.308	1.083	1.180	0.995
Final Dry Season Tree Models	No lag	324.59	4.35 E−05	NA	NA	NA	0.349	1.102	NA	0.499
	1 week lag	290.8	7.77 E−04	NA	0.690	NA	0.127	1.093	NA	1.278
	2 week lag	281.45	0.522	NA	NA	NA	0.875	1.016	NA	1.031

This table includes the final models for whole year or dry season, based on tree or granary data, and zero-, 1- and 2-week lags. Lower akaike information criterion (AIC) indicates better fitting model. P-values for each variable included in the model are shown, and in bold are significant values. If a variable was not included in the final model it is listed as NA. The Incidence Rate Ratio (IRR), indicates the ratio by which malaria incidence increases for every unit increase of humidity or temperature, as other variables are held constant

ISSN: 1475-2875