Table 2.

Primary and Subgroup Analyses of the Primary Outcomes.^*

Outcome	Prednisone Alone		Thymectomy plus Prednisone		Estimated Difference (95% CI)†	P Value‡
	value	no. of patients	value	no. of patients
Primary analyses
Time-weighted average QMG score over 3-yr period	8.99±4.93	56	6.15±4.09	62	2.85 (0.47 to 5.22)	<0.001
Time-weighted average alternate-day prednisone dose over 3-yr period (mg)	60±27	56	44±21	61	16 (7 to 25)	<0.001
Subgroup analyses
Time-weighted average QMG score
Prednisone use at enrollment						0.86
Yes	9.10±5.06	46	6.30±3.89	47	2.80 (0.11 to 5.49)	0.004
No	8.84±4.60	9	5.66±4.79	15	3.18 (−3.03 to 9.39)	0.12
Sex						0.57
Female	9.73±5.16	38	6.47±4.13	46	3.26 (0.34 to 6.18)	0.002
Male	7.45±4.11	18	5.23±3.95	16	2.22 (−1.96 to 6.40)	0.12
Age at disease onset						0.74
<40 yr	9.60±5.32	34	6.50±4.41	42	3.10 (−0.13 to 6.33)	0.007
≥40 yr	7.85±3.50	18	5.33±2.79	18	2.52 (−0.65 to 5.69)	0.02
Time-weighted average alternate-day prednisone dose (mg)
Prednisone use at enrollment						0.37
Yes	61±28	46	44±22	46	17 (7 to 28)	0.002
No	48±19	9	42±20	15	7 (−10 to 24)	0.40
Sex						0.32
Female	59±25	37	46±23	45	13 (3 to 24)	0.01
Male	61±32	19	38±15	16	23 (6 to 40)	0.009
Age at disease onset						0.94
<40 yr	61±27	33	45±22	41	16 (4 to 27)	0.007
≥40 yr	56±28	19	41±21	18	15 (−1 to 32)	0.07

^*CI denotes confidence interval.

^†We used 95% confidence intervals in all analyses except for analyses involving the QMG score, for which we used 99.5% confidence intervals, per protocol.

^‡P values for between-group comparisons are based on two independent sample Student’s t-tests. P values for interaction with treatment were based on fitting a general linear model separately for each variable.