Skip to main content

Table 1 Characteristics of included surrogate and final trials.

From: Treatment effect bias in randomised controlled trials using surrogate outcomes: a preliminary cohort study analysis

  Surrogate Trials (N=138) N(%) Final Trials (N=132) N(%)
Intervention clinical area   
Cardiovascular 31(22) 31(23)
Endocrinology 11(8) 4(3)
Gastrology/hepatology 11(8) 11(8)
Infectious disease 28(20) 28(21)
Nephrology/urology 4(3) 4(3)
Neurology 2(1) 3(2)
Obstetrics 5(4) 5(4)
Oncology 4(3) 4(3)
Other 36(26) 36(27)
Pulmunology 6(4) 6(5)
Population clinical area   
Cardiovascular 34(25) 31(23)
Endocrinology 14(10) 9(7)
Gastrology/hepatology 9(7) 10(8)
Infectious disease 21(15) 21(16)
Nephrology/urology 2(1) 5(4)
Neurology 2(1) 2(2)
Obstetrics 7(5) 8(6)
Oncology 4(3) 4(3)
Other 36(26) 36(27)
Pulmunology 9(7) 6(5)
Journal   
Annals 14(10) 10(8)
BMJ 11(8) 14(11)
JAMA 31(22) 31(23)
Lancet 28(20) 28(21)
NEJM 51(37) 49(37)
PLoS 3(2) -
Year   
2005 64(46) 63(48)
2006 74(54) 69(52)
  1. Surrogate trials were less likely to have adequate evidence of randomisation sequence generation and adopt the ITT principle. We also found clear evidence that final trials were more likely to observe a non-statistically significant (‘neutral’) treatment effect than surrogate trials (49% vs 23%) (Table 2).