TABLE 3

GRADE and Summary of Finding Table

Quality AssessmentNo. of PatientsEffectQuality
No. of StudiesStudy DesignRisk of BiasInconsistencyIndirectnessImprecisionOther ConsiderationsMetforminOCPRelative (95% CI)Absolute (95% CI)
Menstrual cycle regulation (follow-up: mean 6 mo; assessed with number of cycles per month)
2RCTSeriousaNot seriousNot seriousSeriousbNone2222MD 0.27 lower (0.33 lower to 0.21 lower)⨁⨁OO Low
Hirsutism (follow-up: range 6 to 24 mo; assessed with Ferriman Gallwey score)
3RCTVery seriouscNot seriousNot seriousSeriousc,dNone6265MD 0.05 higher (0.62 lower to 0.71 higher)⨁OOO Very low
BMI (follow-up: range 6 to 24 mo; assessed with kg/m2)
4RCTVery seriouscNot seriousNot seriousNot seriousNone7277MD 4.02 lower (5.23 lower to 2.81 lower)⨁⨁OO Low
Acne (follow-up: mean 6 mo; assessed with Cook’s numeric grading)
1RCTSeriouseNot seriousNot seriousSeriousfNone1615MD 0.3 higher (0.05 higher to 0.55 higher)⨁⨁OO Low
Dysglycemia (follow-up: range 6 to 24 mo; assessed with number of girls with diabetes and prediabetes)
2RCTSeriousgNot seriousNot seriousSeriousfNone7/38 (18.4%)19/43 (44.2%)RR 0.41 (0.19 to 0.86)0 fewer per 1000 (from 62 fewer to 358 fewer)⨁⨁OO Low
Total testosterone (follow-up: range 6 to 24 mo; assessed with nmol/L)
4RCTVery seriouscNot seriousSerioushNot seriousNone7277MD 1.2 higher (0.91 higher to 1.5 higher)⨁OOO Very low
Triglyceride (follow-up: mean 6 mo; assessed with mg/dL)
3RCTVery seriouscNot seriousSerioushSeriousb,iNone3237MD 9.69 lower (31.32 lower to 11.95 higher)⨁OOO Very low
Total Cholesterol (follow-up: mean 6 mo; assessed with mg/dL)
2RCTVery seriouscNot seriousSerioushSeriousbNone2225MD 43.23 lower (64.15 lower to 22.32 lower)⨁OOO Very low
LDL (follow-up: mean 6 mo; assessed with mg/dL)
2RCTVery seriouscNot seriousSerioushSeriousbNone2225MD 35.5 lower (57.45 lower to 13.55 lower)⨁OOO Very low
HDL (follow-up: mean 6 mo; assessed with mg/dL)
3RCTVery seriouscNot seriousSerioushSeriousbNone3237MD 2.24 higher (3.83 lower to 8.32 higher)⨁OOO Very low
  • Question: Among adolescents aged 11 to 19 y with PCOS, does the use of metformin compared with oral combined contraceptive pill improve menstrual cyclicity, reduce clinical hyperandrogenism, and improve metabolic profile? Setting: outpatients. O, downgrade of evidence level. MD, mean difference.

  • a One study performed semiopen the concealment of allocation for the metformin group, and had high loss of follow-up.

  • b Not meeting optimal information size criteria.

  • c Two out of 3 studies were high risk of bias (unblinded, no concealment, high loss of follow-up).

  • d CI contains MD = 0.

  • e Unblinded study.

  • f Not meeting optimal information size criteria.

  • g Unblinded study, high loss of follow-up.

  • h Surrogate outcome.

  • i Point estimates and CI were not precise.