Table 2

GRADE evidence profile: risks and benefits of temporarily discontinuing medications to prevent acute kidney injury

Quality assessmentNumber of patientsEffect
StudiesRisk of biasInconsistencyIndirectnessImprecisionOther considerationsContinuationDiscontinuationRelative
(95% CI)
(95% CI)
Incidence of acute kidney injury
3 RCTsNot seriousNot seriousNot seriousSerious*Publication bias strongly suspected†27/248 (10.9%)21/274 (7.7%)RR 1.48
(0.84 to 2.60)
52 more per 1000
(from 17 fewer to 174 more)
3 RCTs3 CohortsVery serious‡Not seriousNot seriousNot seriousPublication bias strongly suspected†134/520 (25.8%)323/1099 (29.4%)RR 1.14
(0.96 to 1.36)
36 more per 1000
(from 10 fewer to 93 more)
Very low
  • *Wide CI and few events.

  • †Non-randomised studies appear would have been unlikely to have been written up for publication if findings had been negative; therefore, similar studies with negative findings considered likely.

  • ‡RCTS, no serious concerns regarding risk of bias. Two cohort studies, one judged moderate risk of bias due to possibility of residual confound, one judged critical risk of bias as did not control for confounding.

  • RR, Risk ratio.