Table III.

(a) Pertinent summary statistics on the survival endpoints of interest. Disease-free survival (DFS) is defined as time until a clinically detectable recurrence at any site (no melanoma deaths occurred without recurrence). Distant-disease-free survival (DDFS) is specific to recurrence at a distant site and includes death from melanoma without a distant recurrence. The two subgroups are patients with sentinel-node metastases (node +) and those absent metastases (node −). (b) Results from the two-sample ITT analysis estimating the effectiveness of the new standard of care; the parameter estimates are from Weibull accelerated failure time models. (c) Results from a model to estimate efficacy of lymphadenectomy on node-positive patients without covariates. (d) Follow-up analysis of DDFS to investigate covariate effects.

	Treatment, node + (N = 122)	Treatment, node − (N = 642)	Control (N = 500)
(a) Summary statistics on survival endpoints (N = 1264)
Recurrence at any site—per cent	44.3	16.0	26.8
Distant recurrence—per cent	36.9	12.3	17.8
DFS (months)—median	37.9	59.1	54.5
DDFS (months)—median	43.8	59.8	59.9
(b) Two-sample intention-to-treat test of effectiveness (N = 1264)
	Effectiveness	95 per cent interval	p-Value
DFS	0.344	(0.0782, 0.609)	0.0112
DDFS	0.0708	(−0.201, 0.343)	0.610
(c) Two-sample test of efficacy in node-positive subgroup (N = 1264)
	Efficacy (ψ̂1)	95 per cent interval	p-Value
DFS	1.593	(1.167, 2.019)	2.4E–13
DDFS	0.937	(0.119, 1.755)	0.0247
(d) Regression analysis of DDFS with covariates (N = 1147)
Shared parameters	Estimate	95 per cent interval	p-Value
Breslow thickness	−0.528	(−0.739, −0.318)	8.6E–7
Trunk (site)	−0.486	(−0.766, −0.205)	6.8E–4
Node +
Treatment efficacy	1.135	(0.402, 1.869)	0.0024
Ulceration	0.179	(−0.377, 0.735)	0.527
Node −
Ulceration	−0.639	(−1.007, −0.270)	6.8E–4