Traditional case-mix adjustments via multivariate regression based methods cannot completely account for the selection bias.
The foregoing analysis shows that traditional multivariate regression based methods tend to overestimate the impact of admission to stroke unit on outcomes in the presence of overt selection.
At each stated follow up point, outcomes will be compared using multivariate regression based methods (e.g. logistic or least squares) adjusting follow up scores for baseline scores (where available) and key baseline characteristics (e.g. age/sex, race/ethnicity).
The multivariate regression based minimum P value approach yielded maximal z-statistics at SBP thresholds of 100 mmHg in the ED, 104 mmHg in the ambulance, and 110 mmHg in the MECU while the univariate analyses yielded maximal z-statistics at SBP thresholds of 95 mmHg in the ED, 103 mmHg in the ambulance, and 101 mmHg in the MECU (see Additional file 1).
The probability of patients receiving a specific antifracture drug prescription at hospital discharge as well as their 1-year adherence to prescribed drugs were compared post-FPS versus pre-FPS using multivariate regression based on a generalized linear model procedure (Tables 1 and 2, respectively).
Covariates previously hypothesized to have an effect on asthma status (native status, breast-feeding, maternal smoking, income quintile, education quartile, maternal age, birth weight, and gestational length) were included in multivariate regressions based on statistical significance in bivariate regressions.
Propensity scores predicting use of saxagliptin or sitagliptin were generated using multivariate logistic regression based on demographic characteristics, comorbidities in the baseline period, comorbidity indices, baseline resources use, and baseline pharmacotherapy before the index date.
To verify that the ER status-dependent prognostic power of proliferation is not affected by bias in tumour stage, we combined the ER calls in the four reference cohorts and performed a multivariate Cox regression based on lymph-node status, ER status, MKI67 expression and their interaction.
Multivariate logistic regression based on the significant variables from univariate logistic regression showed that for both cut-points neck circumference and witnessed apnoeas were independent predictors of SDB.
The SADPM (14) is a multivariate logistic regression based on age, sex, BMI, ethnicity, FPG, HDL cholesterol, and blood pressure to compute a risk score for future T2DM.
To compensate for the missing values in the logistic regression models, multiple imputation was employed using multivariate normal regression, based on age, sex, and smoking data; multiple imputation results in less biased findings when dealing with missing covariate data [ 13].
