Table 2.
Factora | bb (s.e.) | 95% CI | p valuec | f 2d |
---|---|---|---|---|
IgG1 | 0.38 | |||
Linear term | 1.10 (0.16) | 0.78, 1.41 | < 0.001 | |
Interaction term | − 0.28 (0.05) | − 0.38, − 0.18 | < 0.001 | |
IgG2 | 0.02 | |||
Linear term | 0.11 (0.13) | − 0.14, 0.37 | 0.388 | |
Interaction term | − 0.01 (0.05) | − 0.10, 0.08 | 0.805 | |
IgG3 | 0.18 | |||
Linear term | 0.54 (0.15) | 0.24, 0.83 | < 0.001 | |
Interaction term | − 0.07 (0.05) | − 0.18, 0.03 | 0.177 | |
IgG4 | 0.02 | |||
Linear term | 0.21 (0.86) | − 1.47, 1.90 | 0.802 | |
Interaction term | 0.07 (0.31) | − 0.53, 0.68 | 0.808 | |
IgM | 0.03 | |||
Linear term | − 0.21 (0.18) | − 0.57, 0.38 | 0.233 | |
Interaction term | − 0.02 (0.07) | − 0.11, 0.15 | 0.792 |
aLinear mixed modeling (LMM) analyses where the natural log of C1q-fixation was regressed on the natural log of antibody responses (linear term) conditioning on other antibody response factors to provide an independent association. This LMM model included a term for the natural log of time (not shown) and relaxed the constraint of a consistent association between C1q-fixation and antibody responses across time (interaction term). The LMMs applied a random intercept for study participant and random slope for time
bRegression coefficient (b) and standard error (s.e.) represent the percent change in participant C1q-fixation level for a percent increase in antibody responses
cProbability values based on Wald statistics
dCohen’s f2 represents the ratio of the unique variance explained by a specific antibody response factor to the variance explained by an intercept-only model. Higher values indicate a stronger effect