TABLE 5

Adjusted AH nonlinear negative binomial regression models exploring the relationship of environmental factors with influenza activity and the nonlinearity of AH and temperature with influenza A and B virusesa

Demographic or climatic factorAll influenza virusesInfluenza A virusInfluenza B virus
IRR (95% CI)P valueNonlinearity P valueIRR (95% CI)Association P valueNonlinearity P valueIRR (95% CI)P valueNonlinearity P value
Age group (yr)
    65+1.00NANA1.00NANA1.00NANA
    <10.69 (0.55–0.87)<0.0001*NA0.70 (0.52–0.95)0.0016*NA0.70 (0.47–1.04)<0.0001NA
    1–41.35 (1.10–1.67)<0.0001*NA1.21 (0.92–1.61)<0.0001NA1.67 (1.18–2.37)<0.0001*NA
    5–192.43 (1.97–3.01)<0.0001*NA1.78 (1.32–2.39)0.0007*NA4.07 (2.89–5.72)0.004*NA
    20–641.19 (1.00–1.41)0.002NA1.20 (0.96–1.50)0.0243NA1.26 (0.95–1.69)0.084NA
Outbreak status
    Yes1.00NANA1.00NANA1.00NANA
    No0.26 (0.23–0.31)<0.0001*NA0.24 (0.19–0.30)<0.0001*NA0.31 (0.24–0.41)<0.0001*NA
AHbNA<0.0001*0.0202*NA<0.0001*0.2348NA<0.0001*<0.0001
WS1.00 (0.98–1.02)0.82NA0.97 (0.94–1.00)0.0824NA1.04 (1.00–1.08)0.0152NA
Temp fluctuation1.03 (1.01–1.05)<0.0001*NA1.00 (0.98–1.02)0.4897NA1.07 (1.04–1.10)<0.0001*NA
  • a AH, absolute humidity; WS, wind speed; IRR, incidence relative risk, CI, confidence interval; NA, the measurement is not applicable for that variable. The AH nonlinear regression model explored the association of environmental factors with influenza activity as well as the nonlinearity of the association of AH with all influenza viruses and influenza A and B viruses. The left column lists independent/predictable variables for which this model was adjusted. The total weekly numbers of positive influenza A and B virus counts were used as dependent variables. A significant result (*) for association is considered when the 95% confidence interval does not cross 1 and the P value is <0.05. A significant result for nonlinearity is considered when the P value is <0.05. An incidence relative risk of 1.00 indicates the category used for reference/comparison. AH, temperature, and WS were measured by the use of weekly median measurements. Influenza A and influenza B viruses represent the total weekly numbers of positive specimens. All influenza viruses represent the sum of influenza A and B virus-positive specimens.

  • b AH was also examined for a nonlinear association with influenza A and B viruses.