TABLE 3.

Estimated parameters from the network meta‐regression model using the two different scores developed from the LASSO model and prespecified model

Estimated parameters from IPD	LASSO model	Prespecified model
NMR model	Mean (95% Cr. interval)	Mean (95% Cr. interval)
γ0	2.30 (1.78, 2.8)	1.26 (0.95, 1.58)
δ DF	−0.92 (−1.20, −0.64)	−0.89 (−1.18, −0.60)
δ GA	−0.72 (−1.15, −0.28)	−0.71 (−1.15, −0.26)
δ N	−1.24 (−1.55, −0.93)	−1.22 (−1.53, −0.93)
γ DF	0.90 (−0.20, 1.98)	0.25 (−0.35, 0.87)
γ GA	0.64 (−1.02, 2.39)	0.23 (−0.71, 1.3)
γ N	−0.02 (−1.16, 1.07)	−0.26 (−1.01, 0.43)

Note: e ^γ0, OR of relapse in 2 years for one unit increase in logit‐risk in untreated patients (placebo); $e^{{\delta }_{\mathrm{DF}}}$, OR of relapse under Dimethyl Fumarate vs placebo at the study mean risk; $e^{{\delta }_{\mathrm{GA}}}$, OR of relapse under Glatiramer Acetate vs placebo at the study mean risk; $e^{{\delta }_N}$, OR of relapse under Natalizumab vs placebo at the study mean risk; $e^{{\gamma }_{\mathrm{DF}}}:$ OR of relapse under Dimethyl Fumarate vs placebo for one unit of increase in the logit risk; $e^{{\gamma }_{\mathrm{GA}}}$, OR of relapse under Glatiramer Acetate vs placebo for one unit of increase in the logit risk; $e^{{\gamma }_N},$ OR of relapse under Natalizumab vs placebo for one unit of increase in the logit risk.

Abbreviations: DF, Dimethyl Fumarate; GA, Glatiramer Acetate; IPD, individual patient data; N, Natalizumab; LASSO, least absolute shrinkage and selection operator; NMR, network meta‐regression.