Multivariable linear regression model.

Combinations of parameters identified as significant in bivariable modeling were included in multivariable linear regression models, with stepwise elimination of variables that were not statistically significant in the model. n = 74, Akaike information criterion (AIC) = 585.15, R2adj = 0.40.aThe model with the best fit for CXCL10 was stratified by HIV status and as such the slopes for HIV positive and HIV negative groups are different and are shown separately.bThis term describes the difference in slope between the HIV positive and HIV negative models. The p value indicates that the slopes are significantly different.Note. p values Abbreviations: CI, confidence interval; LPS, lipopolysaccharide.