Statistical modeling of the relation of IgG1, IgG3, and IgA FVIII-binding antibodies in healthy individuals with covariates age, sex, and location of the plasma center using a zero-inflated Poisson regression model.

The comparison of the full specified model with the null model using the likelihood ratio test resulted in a P value of .8938 indicating that age, sex, and location of the plasma center taken together as predictor variables (not just individually) did not result in a statistically significant improvement in the model fit.

The comparison of the full specified model with the null model using the likelihood ratio test resulted in a P value of .0803 indicating that age, sex, and location of the plasma center taken together as predictor variables (not just individually) did not result in a statistically significant improvement in the model fit.

The comparison of the full specified model with the null model using the likelihood ratio test resulted in a P value of .0055 indicating that age, sex, and location of the plasma center taken together as predictor variables (not just individually) resulted in a statistically significant improvement in the model fit compared to the trivial model. The full specified model was further refined based on clinical input and the akaike information criterion (AIC) criteria resulting in a final zero-inflated Poisson regression model consisting of the intercept only in the titer ≥ 1:20 component and with covariates age and plasma center in the titer < 1:20 component which resulted also in a statistically significant improvement in the model fit compared to the null model (P = .0051).

Null model: zero-inflated Poisson regression model without covariates (ie, trivial model with intercepts only).

Full specified model: zero-inflated Poisson regression model with covariates age, sex, and location of the plasma center in the titer ≥ 1:20 component as well as in the titer < 1:20 component.