Global positive DHTb | Resolution of baseline vertigob | Likertc | |||||
---|---|---|---|---|---|---|---|
Odds ratio (95% CI) | P-value | Odds ratio (95% CI) | P-value | M.Ef (95% CI) | P-value | ||
Intercept term | (Reference) | 0.259 (0.074 to 0.907) | 0.035 | 0.833 (0.128 to 5.401) | 0.848 | 1.25 (0.20 to 2.16) | 0.002 |
Follow-up visit | Week | Reference | – | Reference | – | Reference | – |
Month | 0.760 (0.327 to 1.766) | 0.523 | 5.646 (1.615 to 19.736) | 0.007 | −0.52 (−1.00 to 0.03) | 0.103 | |
Year | 0.305 (0.105 to 0.887) | 0.029 | 21.856 (4.182 to 114.215) | <0.001 | −1.83 (−2.74 to −0.93) | <0.001 | |
Treatment group | Control | Reference | – | Reference | – | Reference | – |
Intervention | 1.748 (0.435 to 7.033) | 0.431 | 0.508 (0.062 to 4.189) | 0.53 | 0.56 (−0.30 to 1.52) | 0.202 | |
Presence of baseline nystagmus | 1.296 (0.283 to 5.937) | 0.738 | 2.734 (0.169 to 44.210) | 0.479 | 0.38 (−0.45 to 1.13) | 0.433 | |
Daily use of betahistine | 1.426 (0.973 to 2.091) | 0.069 | 0.374 (0.158 to 0.889) | 0.026 | 0.66 (0.45 to 0.82) | <0.001 | |
Two-way interactions Group (intervention): Nystagmus | Intervention with baseline nystagmus | 0.095 (0.010 to 0.924) | 0.043 | 15.418 (0.583 to 407.714) | 0.102 | −1.73 (−2.95 to −0.51) | 0.011 |
Nystagmus: betahistine units/d | – | 0.313 (0.090 to 1.093) | 0.069 |
↵a Models obtained from all two-way interactions and three-way interactions interacting with visit and intervention group, and through a variable selection elimination process according to the Akaike Information Criterion.
↵b Mixed logistic regression. Results are shown as exponential of the coefficients obtained from the model (expressed as odds ratios except for the intercept term), with a 95% CI estimate and the P-values reflecting the statistical significance of the corresponding coefficients.
↵c Mixed Tobit regression. Marginal effects of the fixed effects of the multivariate mixed Tobit model assuming left censoring at zero and right censoring at 10, with a 95% CI and P-value reflecting the statistical significance of the related coefficient. Marginal effect calculated as the median of the individual marginal effects (explained in Cunillera (2014)).21 M.Ef = marginal effects.