Skip to main content

Table 4 Methodological characteristics of test-treatment trials

From: Test-treatment RCTs are susceptible to bias: a review of the methodological quality of randomized trials that evaluate diagnostic tests

Trial Quality Item Trials, n (%)
Randomized sequence allocation
 Adequate 59 (57)
 Inadequate 2 (2)
 Unclear 42 (41)
Allocation concealment
 Adequate 38 (37)
 Inadequate 3 (3)
 Unclear 62 (60)
Blinding
 Patients 5 (5)
 Care-providers 4 (4)
 Outcome assessors 22 (21)
aSingle-blind 20 (19)
bDouble-blind 5 (5)
 No blinding 78 (76)
Reporting of primary outcome assessment
 Complete 53 (51)
 Partial 33 (32)
 Absent 17 (17)
Reporting of participant flow
 Complete 44 (43)
 Partial 58 (56)
 Absent 1 (1)
Missing data
 Complete 41 (40)
 Attrition ≤10% 30 (29)
 Attrition >10% 25 (24)
 Incomplete, cannot calculate 4 (4)
 Unclear if complete 3 (3)
Differential attrition
 ≥ 5% between arms 16 (16)
 ≥ 20% between arms 1 (1)
Intention-to-treat (ITT)
 Patients analyzed as randomized 72 (70)
 Complete or imputed data and analyzed as randomized 30 (29)
 Not conducted 31 (30)
Inconsistent outcome assessment 21 (20)
Inappropriate subgroup analysis 9 (9)
Sample size
 Power calculation reported 81 (79)
cMedian trial sample size [IQR] 309 [153–731]
cMedian study arm sample size [IQR] 166 [72–297]
  1. ablinding either patients or care-providers or outcome assessors; bblinding at least two of: patients, care-providers or outcome assessors; cbased on numbers randomized in all 103 trials