Supplementary appendix IV: Appraisal tables of included studies
Clomiphene citrate versus placebo
Brown 2009
Study:Brown, J., C. Farquhar, et al. (2009). "Clomiphene and anti-oestrogens for ovulation induction in PCOS." Cochrane Database of Systematic Reviews(4).Description of study: systematic review of RCTs. All were of cross-over design, however only first phase data were used.
Patient/population / Women of reproductive age with WHO group two anovulation.
N / 15 studies (Please note all these studies did not assess our comparison or outcomes of interest)
Clomiphene citrate versus placebo: 3 studies, n=133
Setting / One out of the three studies reported the setting in a Department of obstetrics and gynaecology (Garcia 1985). The three studies were conducted in US and Canada.
Intervention/indicator / Reference / Intervention
Cudmore 1966 / 50mg clomiphene citrate (days 1-14) for 3 cycles then 3 cycles (n=13).
No infertility treatment in the threemonths prior to the study, but sensitivity to clomiphene citrate is not reported.
Garcia 1985 / 50 mg clomiphene citrate, increased by 50mg if ovulation failed to occur, up to250 mg for 5 cycles then 5 cycles.
Sensitivity to clomiphene citrate not reported.
Johnson 1966 / 100mg clomiphene citrate days 6-10 for 1 cycle then 1 cycle.
Sensitivity to clomiphene citrate not reported.
Comparison/control / Reference / Comparison
Cudmore 1966 / Placebo (days 1-14) for 3 cycles then 3 cycles (n=9).
Garcia 1985 / Placebo, 1 tablet, increased up to 5 tablets similar to treatment for 5 cycles then 5 cycles.
Johnson 1966 / Placebo days 6-10 for 1 cycle then 1 cycle.
Outcomes / Ovulation and pregnancy rates per patient.
Inclusion Criteria / “Randomised controlled trials (RCT) that compared oral agentsfor ovulation induction (alone or in conjunction with medicaladjuncts) in anovulatory subfertility were considered for inclusionin the review.”
“Women of reproductive age with WHO group two anovulation.”
Exclusion Criteria / “Quasi-randomised trials were excluded. Cross-over trials were notincluded unless phase one data were available.”
“Women with hyperprolactinaemia or Cushing’s syndrome, orboth, were excluded and trials which reported that women withthese two conditions had been included were excluded from thereview. Trials containing women with WHO group one anovulationwere excluded.”
Study Validity
Is it clear that there were no conflicts of interest in the writing or funding of this review? / Yes
Does the review have a clearly- focused question? / Yes
Is a systematic review the appropriate method to answer the question? / Yes
Does the review have specified inclusion/exclusion criteria? / Yes
If there were specified inclusion/ exclusion criteria, were these appropriate? / Yes
Does the review document a comprehensive search strategy? / Yes
Were reviewers blind to authors, institutions and affiliations? / Not reported
Were 2 or more independent reviewers used for:
- application of inclusion criteria to assess eligibility of studies?
- extraction of data from study reports?
- appraisal of study quality?
Were the strengths and limitations of included studies and potential impact on the results discussed? / Yes
Was the validity of included trials appraised using appropriate criteria? / Yes
Is there a summary of the results of individual studies? / Partial
If meta-analyses were conducted, was it reasonable to do so? / Yes
If meta-analyses were conducted, was it done appropriately? / Yes
What is the overall risk of bias? / Low / Most of the criteria have been fulfilled and where criteria have not been fulfilled it is very unlikely the conclusions of the study would be affected.
Results
Odds ratio (fixed effect) [95% CI]
Clomiphene citrate versus placebo (in women with PCOS where sensitivity to clomiphene citrate is not reported)
Ovulation rate: 7.47 [3.24 to 17.23] I2=0% P < 0.00001 (3 trials, 133 participants) – favouring clomiphene citrate
Pregnancy rate: 5.77 [1.55 to 21.48] I2=0% P < 0.009 (3 trials, 133 participants) – favouring clomiphene citrate
Author’s Conclusions
“Strong evidence in favour of one anti-oestrogen or adjunctive agenthas not been found. This review shows evidence supporting theeffectiveness of the current first-line treatment, clomiphene citrate,in terms of pregnancies. There is no evidence of a differencein effect between clomiphene and tamoxifen or clomiphene alonebut the number of women studied is too small to be conclusive. No trials comparing tamoxifen and placebo could be found. Thereare insufficient data to determine the place of ketoconazole, tamoxifen,bromocriptine, hCGor hormone supplementation as anadjunct with clomiphene versus clomiphene alone in anovulatory,normoprolactinaemic women. Further trials are clearly needed inthese areas.Dexamethasone andCOCs appear to be promising adjunctsto clomiphene treatment and also require further research.”
Our Comments/Summary
- High quality systematic review with low risk of bias.
- Clomiphene citrate maybe better than placebo for ovulation rate and pregnancy rate.
Appraisal tables of included studies: Metformin
Tang 2010
Study:Tang, T., J. M. Lord, et al. (2010). "Insulin-sensitising drugs (metformin, rosiglitazone, pioglitazone, D-chiro-inositol) for women with polycystic ovary syndrome, oligo amenorrhoea and subfertility." Cochrane Database of Systematic Reviews 2010(1): 1-131.Description of study: systematic review of RCTs.
Patient/population / Women with oligo/anovulatory PCOS based on the recent diagnosticcriteria set by the Rotterdam consensus.
N / 31 studies, n=2537 (Please note all these studies may not have assessed our comparison or outcomes of interest)
Metformin v placebo or no treatment: 15 studies, n=up to 875 (some outcomes may have not been measured in all studies)
Metformin v clomiphene citrate: 3 studies, n=up to 2470 (some outcomes may have not been measured in all studies)
Metformin + clomiphene citrate v clomiphene citrate: 11 studies, n= up to 2668 (some outcomes may have not been measured in all studies)
Setting / “The included studies originated from a number of countries includingVenezuela, Brazil, Germany, Egypt, UK, Italy, USA, Taiwan,Iran, Turkey, Malaysia, Holland, Hong Kong, Finland andDenmark.”
Intervention/indicator / See appendix 3 for population and intervention details for each study.
Comparison/control / See appendix 3 for population and intervention details for each study.
Outcomes / Primary outcomes: live birth rate, clinical pregnancy rate, ovulation rate; Secondary outcomes: miscarriage, multiple pregnancy and adverse events (gastrointestinal side effects).
Outcomes not relevant to this evidence review include menstrual frequency, blood pressure, body mass index, waist circumference and waist-hip ratio and hormone and metabolic profile.
Inclusion Criteria / Randomised controlled trials which were truly randomised. Crossover studies were to be includedbut only first phase data was included.
Women with oligo/anovulatory PCOS based on the recent diagnosticcriteria set by the Rotterdam consensus.
Metformin versus placebo or notreatment
Metformin versus clomiphene citrate
Metformin + clomiphene citrate versus clomiphene citrate
Exclusion Criteria / QuasiRCTs were not included.
Since troglitazone has been withdrawn from the market, studiesinvolving troglitazone were excluded from the current review.
Study Validity
Is it clear that there were no conflicts of interest in the writing or funding of this review? / Yes
Does the review have a clearly- focused question? / Yes
Is a systematic review the appropriate method to answer the question? / Yes
Does the review have specified inclusion/exclusion criteria? / Yes
If there were specified inclusion/ exclusion criteria, were these appropriate?? / Yes
Does the review document a comprehensive search strategy? / Yes
Were reviewers blind to authors, institutions and affiliations? / Not reported
Were 2 or more independent reviewers used for:
- application of inclusion criteria to assess eligibility of studies?
- extraction of data from study reports?
- appraisal of study quality?
Were the strengths and limitations of included studies and potential impact on the results discussed? / Partial
Was the validity of included trials appraised using appropriate criteria? / Yes
Is there a summary of the results of individual studies? / Yes
If meta-analyses were conducted, was it reasonable to do so? / Yes
If meta-analyses were conducted, was it done appropriately? / Yes / It would have been helpful to have subgroup effect estimates for women who are clomiphene citrate resistant for all comparisons and outcomes.
What is the overall risk of bias? / Low / Most of the criteria have been fulfilled and where criteria have not been fulfilled it is unlikely the conclusions of the study would be affected.
Results
Odds ratio (fixed effect) [95% CI]
Metformin versus placebo (in women with PCOS not reported to be CC-resistant as well as women with CC-resistant PCOS)
Live birth rate: 1.0 [0.16, 6.39] I2=0% p=1.0 (2 studies, 50 participants) – no favour
Pregnancy rate: 3.86 [2.18, 6.84] I2=0% p<0.00001 (6 studies, 479 participants) – favouring metformin
Ovulation rate: 2.12 [1.50,3.00] I2=69% p=0.000019 (13 studies, 875 participants) – favouring metformin
Miscarriage rate: 0.30 [0.06, 1.59] I2=NA p=0.16 (1 study, 51 participants) – favouring metformin
GI related adverse events: 9.23 [4.18, 20.37] I2=25% p<0.00001 (5 studies, 253 participants) – favouring placebo
Metformin versus clomiphene citrate(in women with PCOS not reported to be CC-resistant)
Live birth rate: 0.67 [0.44, 1.02] I2=93% p=0.061 (3 studies, 600 participants) – favouring clomiphene citrate
Pregnancy rate: 0.63 [0.43, 0.92] I2=91% p=0.018 (3 studies, 600 participants) – favouring clomiphene citrate
Ovulation rate: 0.48 [0.41, 0.57] I2=78% p<0.00001 (3 studies, 2470 participants) – favouring clomiphene citrate
Miscarriage rate: 0.94 [0.42, 2.07] I2=84% p=0.87 (3 studies, 145 participants) – no favour
Multiple pregnancy rate: 0.33 [0.02, 6.69] I2=0% p=0.47 (3 studies, 145 participants) – favouring metformin
Metformin + clomiphene citrate versus clomiphene citrate(in women with PCOS not reported to be CC-resistant as well as women with CC-resistant PCOS)
Live birth rate: 1.05 [0.75, 1.47] I2=48% p=0.78 (4 studies, 752 participants) – no favour
Pregnancy rate: 1.48 [1.12, 1.95] I2=58% p=0.0058 (7 studies, 976 participants) – favouring metformin + clomiphene citrate
Miscarriage rate: 1.48 [0.90, 2.43] I2=0% p=0.12 (5 studies, 745 participants) – favouring clomiphene citrate alone
Multiple pregnancy rate: 0.50 [0.12, 2.02] I2=0% p=0.33 (4 studies, 665 participants) – favouring metformin + clomiphene citrate
GI related adverse events: 3.40 [2.08, 5.54] I2=NA p<0.00001 (1 study, 418 participants) – favouring clomiphene citrate alone
Ovulation rate (CC-sensitive): 3.55 [0.65, 19.37] I2=NA p=0.14 (1 study, 56 participants) – favouring metformin + clomiphene citrate
Ovulation rate (CC-resistant): 4.86 [2.43, 9.74] I2=0% p<0.00001 (5 studies, 179 participants) – favouring metformin + clomiphene citrate
Ovulation rate (CC-sensitivity not defined): 1.65 [1.40, 1.94] I2=85% p<0.00001 (5 studies, 2433 participants) – favouring metformin + clomiphene citrate
Ovulation rate (total): 1.76 [1.51, 2.06] I2=74% p<0.00001 (11 studies, 2668 participants) – favouring metformin + clomiphene citrate
Author’s Conclusions
“In agreement with the previous review, metformin is still of benefit in improving clinical pregnancy and ovulation rates. However, thereis no evidence that metformin improves live birth rates whether it is used alone or in combination with clomiphene citrate, or when comparedwith clomiphene citrate. Therefore, the use of metformin in improving reproductive outcomes in women with PCOS appears to be limited.”
Our Comments/Summary
- High quality systematic review with low risk of bias.
- There is high heterogeneity in most meta-analyses. Subgroup by cc-resistance may have provided clarity about the effectiveness of metformin in this group of women where clomiphene citrate might not be the preferred option for treatment.
- In women with PCOS, including a combination of women with cc-resistance and women with undefined cc-sensitivity, metformin may be effective (better than placebo) for pregnancy rate and may be associated with less GI related adverse events.
- In women with cc-resistant PCOS, addition of metformin to clomiphene citrate may be better than clomiphene citrate alone for ovulation rate.
Palomba 2009
Study:Palomba, S., R. Pasquali, et al. (2009). "Clomiphene citrate, metformin or both as first-step approach in treating anovulatory infertility in patients with polycystic ovary syndrome (PCOS): a systematic review of head-to-head randomized controlled studies and meta-analysis." Clinical Endocrinology 70(2): 311-21.Description of study: Systematic review of RCTs.
Patient/population / Women with well defined diagnosisof PCOS.
N / 4 studies, n=1066
Metformin + clomiphene citrate v metformin: 2 studies, n= 741
Quality appraisal and results for the following two comparisons are not discussed here as the same studies have been meta-analysed in Tang 2010:
Metformin + clomiphene citrate v clomiphene citrate: 3 studies, n= 951
Metformin v clomiphene citrate: 3 studies, n=841
Setting / United States and Asia.
Intervention/indicator / See appendix 3 for population and intervention details for each study.
Comparison/control / See appendix 3 for population and intervention details for each study.
Outcomes / “Our primary end-point was the live birth rate.Secondary end-pointswere the rates of ovulation, pregnancy, abortion and discontinuationfor adverse events.”
Inclusion Criteria / The authors state that the articles were reviewed for inclusion and exclusion criteria but don’t state what the criteria were. Characteristics of study populations, interventions and outcomes of the included studies are described.
Exclusion Criteria / “Specifically, retrospective, case–control,nonrandomized and quasi-randomized trials and case reports/serieswere excluded.”
Study Validity
Is it clear that there were no conflicts of interest in the writing or funding of this review? / No
Does the review have a clearly- focused question? / Yes
Is a systematic review the appropriate method to answer the question? / Yes
Does the review have specified inclusion/exclusion criteria? / Partial
If there were specified inclusion/ exclusion criteria, were these appropriate?? / Yes
Does the review document a comprehensive search strategy? / Yes
Were reviewers blind to authors, institutions and affiliations? / No / “Two independent reviewers, not blinded at any pointto the authors or sources of publication, identified and selectedthe RCTs that met the inclusion criteria.”
Were 2 or more independent reviewers used for:
- application of inclusion criteria to assess eligibility of studies?
- extraction of data from study reports?
- appraisal of study quality?
Were the strengths and limitations of included studies and potential impact on the results discussed? / Yes
Was the validity of included trials appraised using appropriate criteria? / Yes
Is there a summary of the results of individual studies? / Yes
If meta-analyses were conducted, was it reasonable to do so? / Yes
If meta-analyses were conducted, was it done appropriately? / Yes / However, the Cochrane Q statistic for measure of heterogeneity is no longer used. I2 may be a more rigorous measure of heterogeneity.
What is the overall risk of bias? / Low / Most of the criteria have been fulfilled and where criteria have not been fulfilled it is unlikely the conclusions of the study would be affected.
Results
Odds ratio (fixed effect) [95% CI]
Metformin + clomiphene citrate versus metformin (in women with PCOS whose sensitivity to CC has not been defined) - 2 studies, 741 participants
Live birth rate: 0.23 [0.13, 0.40] Q=0.533 (p=0.465) p<0.0001 – favouring metformin + clomiphene citrate
Pregnancy rate: 0.23 [0.14, 0.37] Q=0.244 (p=0.622) p<0.0001 – favouring metformin + clomiphene citrate
Ovulation rate: 0.23 [0.15, 0.34] Q=0.948 (p=0.330) p<0.0001 – favouring metformin + clomiphene citrate
Miscarriage rate: 0.47 [0.22, 1.0] Q=0.052 (p=0.820) p=0.069 – favouring metformin + clomiphene citrate
Adverse events: 0.86 [0.23, 3.04] Q=0.03 (p=1.0) p=0.993 – favouring metformin + clomiphene citrate
Author’s Conclusions
“In PCOS patients with anovulatory infertility andnot previously treated, the administration of metformin plus CC isnot better than monotherapy (metformin alone or CC alone),whereas to date no specific recommendation can be given regardingthe use of CC or metformin as first-step drug.”
Our Comments/Summary
- High quality systematic review with low risk of bias.
- Heterogeneity is reported as not significant for all outcomes.
- In women with PCOS, where cc-sensitivity is undefined, combined metformin and clomiphene citrate may be better than metformin alone for live birth rate, pregnancy rate and ovulation rate.
- The authors raise an important point: “A main drawback in the design of the RCTs that assessed the efficacy of combined CC plus metformin vs. CC (18,25,26) or metformin (18,26) alone was that metformin was started either concurrently with (18,26) or after only one month (25) of CC initiation rather than after three or more months of pretreatment with metformin. Metformin has a slower onset of action than CC; hence, the studies as designed were unintentionally biased against demonstrating a value of combining CC with metformin.”
Creanga 2010
Study:Creanga, A. A., H. M. Bradley, et al. (2008). "Use of metformin in polycystic ovary syndrome: a meta-analysis." Obstetrics & Gynecology 111(4): 959-68.Description of study: systematic review of RCTs.
Patient/population / Women with PCOS seeking pregnancy.
N / 17 studies, n=1639(Please note all these studies may not have assessed our comparison or outcomes of interest)
Metformin v placebo: 9 studies, n=397
Metformin + clomiphene citrate v clomiphene citrate: 11 studies, n= 1042
Setting / The included studies originated from a number of countries includingVenezuela, Egypt, UK, Italy, USA, Taiwan,Turkey, Netherlands, Hong Kong and Jordan.
Intervention/indicator / See appendix 3 for population and intervention details for each study.
Comparison/control / See appendix 3 for population and intervention details for each study.
Outcomes / Ovulation and early pregnancy rate.
Inclusion Criteria / “written in English, reported originaldata from randomized controlled trials conducted onhumans, included one or both of the comparisongroups of interest (metformin compared with placebo,and metformin plus clomiphene citrate compared with placeboplus clomiphene citrate), measured one or both of theprimary meta-analytic outcomes (ovulation and earlypregnancy), provided sufficient data to calculate astatistical measure of association between metforminuse and at least one of the primary outcomes, includedstudy subjects meeting the Rotterdam revisedNational Institutes of Health criteria for PCOS (atleast two of the following three: hyperandrogenism,anovulation or oligo-ovulation, polycystic ovaries onultrasonography), and included patients seeking pregnancy.”
Exclusion Criteria / See above.
Study Validity
Is it clear that there were no conflicts of interest in the writing or funding of this review? / Yes
Does the review have a clearly- focused question? / Yes
Is a systematic review the appropriate method to answer the question? / Yes
Does the review have specified inclusion/exclusion criteria? / Yes
If there were specified inclusion/ exclusion criteria, were these appropriate?? / Yes
Does the review document a comprehensive search strategy? / Yes
Were reviewers blind to authors, institutions and affiliations? / Not reported
Were 2 or more independent reviewers used for:
- application of inclusion criteria to assess eligibility of studies?
- extraction of data from study reports?
- appraisal of study quality?
Were the strengths and limitations of included studies and potential impact on the results discussed? / Yes
Was the validity of included trials appraised using appropriate criteria? / Yes
Is there a summary of the results of individual studies? / No
If meta-analyses were conducted, was it reasonable to do so? / Yes
If meta-analyses were conducted, was it done appropriately? / Yes
What is the overall risk of bias? / Low / Most of the criteria have been fulfilled and where criteria have not been fulfilled it is very unlikely the conclusions of the study would be affected.
Results
Odds ratios (random effects) [95% CI]
Metformin versus placebo
Ovulation rate (all women): 2.94 [1.43–6.02] (9 studies, 397 participants) – favouring metformin (no statistically significant heterogeneity X2=14.43, P=0.071)
Ovulation rate (CC-naive): 3.55 [1.46–8.65] (6 studies, 401 participants) – favouring metformin (no statistically significant heterogeneity X2 and P not reported)
Ovulation rate (CC-resistant): 1.32 [0.32–5.53] (3 studies, 103 participants) – favouring metformin (no statistically significant heterogeneity X2 and P not reported)
Early pregnancy rate (all women): 1.56 [0.74–3.33] (8 studies, 607 participants) – favouring metformin (no statistically significant heterogeneity X2=3.05, P=0.692)
Early pregnancy rate (CC-naive): 1.70 [0.75–3.83] (5 studies, 504 participants) – favouring metformin (heterogeneity X2 and P not reported)
Early pregnancy rate (CC-resistant): 0.93 [0.13–7.13] (3 studies, 103 participants) – favouring placebo (heterogeneity X2 and P not reported)
Metformin + clomiphene citrate versus clomiphene citrate
Ovulation rate (all women): 4.39 [1.94–9.96] (11 studies, 1042 participants) – favouring metformin + clomiphene citrate (statistically significant heterogeneity X2=47.72, P<0.001)
Ovulation rate (CC-naive): 3.84 [1.38–10.68] (5 studies, 832 participants) – favouring metformin + clomiphene citrate (statistically significant heterogeneity X2 and P not reported)
Ovulation rate (CC-resistant): 5.09 [1.44–17.98] (6 studies, 446 participants) – favouring metformin + clomiphene citrate (statistically significant heterogeneity X2 and P not reported)
Early pregnancy rate (all women): 2.67 [1.45–4.94] (11 studies, 1238 participants) – favouring metformin (no statistically significant heterogeneity X2=18.16, P=0.033)
Early pregnancy rate (CC-naive): 1.70 [0.99–2.94] (6 studies, 1017 participants) – favouring metformin (heterogeneity X2 and P not reported)
Early pregnancy rate (CC-resistant): 9.62 [2.95–31.45] (5 studies, 221 participants) – favouring metformin (heterogeneity X2 and P not reported)
Author’s Conclusions
“Using all available evidence, this metaanalysissuggests that metformin increases the likelihoodof ovulation and, in combination with clomiphene citrate, increasesthe odds of both ovulation and pregnancy inwomen with polycystic ovary syndrome.”
Our Comments/Summary
- High quality systematic review with low risk of bias.
- In women with PCOS (regardless of CC-sensitivity and in those who are CC-naive), metformin may be better than placebo.
Moll 2007