Skip to main content
. 2017 May 11;23(4):481–500. doi: 10.1093/humupd/dmx011

Table V.

Quality assessment of observational and clinical trials assessing risk of endometriosis recurrence after HRT.

Quality assessment No of patients Effect Evidence quality
Studies Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations Intervention Control Relative (95% CI)
Comparison of HRT with no HRT HRT No treatment
Matorras et al. (2002) Randomized trial Seriousa Not serious Not serious Very seriousb None 4/115 (3.5%) 0/57 (0.0%) RR 4.50 (0.25 to 82.17)c

  • ⨁◯◯◯

  • VERY LOW
Rattanachaiyanont et al. (2003) Observational study Very seriousd Not serious Not serious Very seriousb None 4/90 (4.4%) 0/17 (0.0%) RR 1.78 (0.10 to 31.64)c

  • ⨁◯◯◯

  • VERY LOW
Acien et al. (2013) Observational study Seriouse Not serious Not serious Not applicablef None 0/11 (0%) 0/8 (0%) Not calculableg

  • ⨁◯◯◯

  • VERY LOW
Comparison of delayed HRT with immediate HRT Delayed HRT Immediate HRT
Hickman et al. (1998) Observational study Not serious Not serious Not serious Not serious None

  • (>6 weeks from surgery)

  • 7/35 (20.0%)

  • (≤6 weeks from surgery)

  • 4/60 (6.7%)

 HR 5.74 (1.31 to 25.23)h

  • ⨁⨁◯◯

  • LOW
Arumugam and Damodaran (1998) Observational study Very seriousi Not serious Not serious Not applicable None

  • (5 months from surgery)

  • 0/8 (0%)

  • (3 months from surgery)

  • 0/5 (0%)

 Not calculableg

  • ⨁◯◯◯

  • VERY LOW
Comparison of HRT with tibolone HRT Tibolone
Fedele et al. (1999) Randomized trial Seriousj Not serious Not serious Very seriousb None 4/10 (40.0%) 1/11 (9.1%) RR 4.40 (0.59 to 33.07)c

  • ⨁◯◯◯

  • VERY LOW
Comparison of oestrogen-only HRT with combined HRT Oestrogen-only HRT Combined HRT (continuous and cyclical)
Rattanachaiyanont et al. (2003) Observational study Very seriousd Not serious Not serious Very seriousb None 4/50 (8.0%) 0/40 (0.0%) RR 7.24 (0.40 to 130.54)c

  • ⨁◯◯◯

  • VERY LOW

CI, confidence interval; HR, hazard ratio; HRT, hormone replacement therapy; RR, risk ratio.

aHigh risk of performance bias—single blinded study, with physician unaware of treatment allocation, but with access to hormone results (which would have indicated treatment with HRT or not). High risk of detection bias, as assessment for recurrence was only carried out if the clinician felt this was warranted, which may have been influenced by the participant (who was not blind to treatment allocation).

bVery wide CI for RR.

cRR calculated by the authors using Review Manager (RevMan) [Computer program]. Version 5.3. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014.

dHigh risk of selection bias as unclear why women were allocated to different HRT regimens (or no HRT). High risk of detection bias, as researchers would have been aware of the woman's HRT status when assessing presence of recurrence (by reviewing medical records).

eRisk of detection bias, as criteria for designating recurrence are not clearly stated.

fNot applicable as odds ratio and CI cannot be calculated.

gNo events in either group, therefore odds ratio not calculable.

hHR adjusted for stage of endometriosis, age at time of hysterectomy and postoperative adjunct medroxyprogesterone therapy.

iHigh risk of selection bias (unclear why some women started HRT after 3 months and some after 5 months), and high risk of detection bias (recurrence was only based on CA 125 levels).

jNo description of blinding for the trial, and no scoring system is reported for pain, therefore risk of detection bias.