Types of studies

We included randomized controlled trials (RCTs) (excluding cluster‐randomized or cross‐over trials) comparing HRT of any regimen and duration of administration versus placebo or no hormone therapy, or trials comparing different regimens or duration of administration of HRT.

Types of participants

We included women of any age who were diagnosed with any stage of EOC and had surgical treatment, regardless of chemotherapy treatment.

Types of interventions

We included studies of HRT used after treatment for EOC. We did not study the aetiological effect of HRT used before diagnosis. We included any HRT (oestrogen alone or combined with progestin, oestrogen agonist/antagonist, progestin, or testosterone, and tibolone), of any regimen or duration of administration, compared with placebo or no hormone therapy. Due to pharmacokinetic differences between different regimens and durations of administration, we investigated the following three comparisons.

1. HRT versus placebo or no HRT

2. Different regimens of HRT

3. Different durations of HRT administration

Types of outcome measures

Electronic searches

We searched the following electronic databases on 12 June 2019:

• the Cochrane Central Register of Controlled Trials (CENTRAL; 2019, Issue 6), in the Cochrane library (Appendix 1)

• MEDLINE via Ovid (1946 to 12 June 2019) (Appendix 2)

• Embase via Ovid (1980 to 2019 week 23) (Appendix 3)

Searching other resources

All relevant articles were identified on PubMed and using the 'related articles' feature; a further search was carried out for newly published articles. We searched the following registries for ongoing trials.

• Metaregister (www.controlled‐trials.com/rct)

• Physicians Data Query (www.cancer.gov/publications/pdq)

• World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) (www.who.int/ictrp/en/)

• ClinicalTrails.gov (www.clinicaltrials.gov)

• National Cancer Institute (www.cancer.gov/about‐cancer/treatment/clinical‐trials)

If we had identified any ongoing trials that had not been published, we would have approached the principal investigators and major co‐operative groups active in this area, to ask for relevant data.

We handsearched the citation lists of included studies, key textbooks, and previous systematic reviews, and contacted experts in the field to identify further reports of trials. We also handsearched conferences abstracts from the following sources.

• Gynecologic Oncology (Annual Meeting of the American Society of Gynecologic Oncology)

• International Journal of Gynecological Cancer (Annual Meeting of the International Gynecologic Cancer Society)

• British Gynaecological Cancer Society (BGCS)

• Annual Meeting of European Society of Gynaecological Oncology (ESGO)

Selection of studies

We downloaded all titles and abstracts retrieved by electronic searching to a reference management database (Endnote). After duplicates were removed, we transferred these data to Covidence for study selection. Two review authors (NS and RB) examined the remaining references independently. We excluded those studies which clearly did not meet the inclusion criteria. We obtained copies of the full text of potentially relevant references. Two review authors (NS and RB) independently assessed the eligibility of the retrieved reports/publications. We resolved any disagreement through discussion and we consulted a third person (TL) for a final decision. We identified and excluded duplicates and collated multiple reports of the same study so that each study rather than each report was the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram and 'Characteristics of excluded studies' table (Liberati 2009).

Data extraction and management

Two review authors (NS and KP) independently extracted study characteristics and outcome data from included studies to a piloted data collection form, using Covidence. We noted in the Characteristics of included studies table if outcome data were not reported in a usable way. We resolved disagreements by consensus or by involving a third person (RB). One review author (NS) transferred data into the Review Manager 5 (RevMan 2014) file. We double‐checked that data were entered correctly by comparing the data presented in the systematic review with the study reports. A second review author (KP) spot‐checked study characteristics for accuracy against the trial report.

For included studies, we extracted the following details.

• Author, year of publication and journal citation (including language)

• Country

• Setting

• Inclusion and exclusion criteria

• Study design, methodology

• Study population

  • Total number enrolled

  • Participant characteristics: age, menopausal status at diagnosis, performance status

  • Treatment: type of surgery and chemotherapy

  • Tumour stage, grade and types

• Intervention details

  • All types of HRT: oestrogen alone or combined with progestin, oestrogen agonist/antagonist, progestin, or testosterone, and tibolone

  • Duration of administration of HRT in years

  • Route and doses of HRT

• Comparison

  • Placebo or no treatment

• Risk of bias in study (see Assessment of risk of bias in included studies)

• Duration of follow‐up

• Outcomes: for each outcome, we extracted the outcome definition. For adjusted estimates, we recorded variables adjusted for in analyses.

• Results: we extracted the number of participants allocated to each intervention group, the total number analyzed for each outcome and number of dropouts, including reason for leaving the study.

• Notes: funding for trial and notable conflicts of interest of trial authors.

Outcome data were extracted as follows.

• For time‐to‐event data (survival and disease progression), we extracted the log of the hazard ratio (log(HR)) and its standard error from trial reports. For those studies that they did not report the HR, we attempted to estimate HR, the observed minus expected events (O‐E) and the variance (V) using formula according to Tierney 2007.

• For dichotomous outcomes (e.g. adverse events or deaths) we extracted the number of participants in each treatment arm who experienced the outcome of interest and the number of participants assessed at endpoint, in order to estimate a risk ratio.

• For continuous outcomes (e.g. quality‐of‐life measures), we extracted the final value and standard deviation of the outcome of interest and the number of participants assessed at endpoint in each treatment arm at the end of follow‐up, in order to estimate the mean difference between treatment arms and its standard error.

Where possible, all data extracted would be those relevant to an intention‐to‐treat analysis, in which participants would be analyzed in groups to which they were assigned. We noted the time points at which outcomes were collected and reported.

Assessment of risk of bias in included studies

Two review authors (NS and KP) applied the Cochrane 'Risk of bias' tool independently and resolved differences by discussion or by appeal to a third review author (EM). We judged each item as being at high, low or unclear risk of bias, as set out in the criteria provided by Higgins 2011 (and shown below), and provided a quote from the study report and/or a statement as justification for the judgement for each item in the 'Risk of bias' table. We summarized results in both a 'Risk of bias' graph and a 'Risk of bias' summary. When interpreting treatment effects and meta‐analyses, we took into account the risk of bias for the studies that contributed to that outcome. Where information on risk of bias related to unpublished data or correspondence with a trialist, we noted this in the 'Risk of bias' table.

• Random sequence generation

  • Low risk of bias: e.g. participants were assigned to treatments on basis of a computer‐generated random sequence or a table of random numbers

  • High risk of bias: e.g. participants were assigned to treatments on basis of date of birth, clinic ID number or surname, or there was no attempt to randomize participants

  • Unclear risk of bias: e.g. sequence generation not reported or the information was not available

• Allocation concealment

  • Low risk of bias: e.g. where the allocation sequence could not be foretold

  • High risk of bias: e.g. the allocation sequence could be foretold by participants, investigators or treatment providers

  • Unclear risk of bias: e.g. allocation concealment was not reported

• Blinding of participants and personnel

  • Low risk of bias if participants and personnel were adequately blinded

  • High risk of bias if participants were not blinded to the intervention that the participant received

  • Unclear risk of bias if this was not reported or unclear

• Blinding of outcomes assessors

  • Low risk of bias if outcome assessors were adequately blinded

  • High risk of bias if outcome assessors were not blinded to the intervention that the participant received

  • Unclear risk of bias if this was not reported or unclear

• Incomplete outcome data: we recorded the proportion of participants whose outcomes were not reported at the end of the study. We coded a satisfactory level of loss to follow‐up for each outcome as follows.

  • Low risk of bias if fewer than 20% of participants were lost to follow‐up and reasons for loss to follow‐up were similar in both treatment arms

  • High risk of bias if more than 20% of participants were lost to follow‐up or reasons for loss to follow‐up differed between treatment arms

  • Unclear risk of bias if loss to follow‐up was not reported

• Selective reporting of outcomes

  • Low risk of bias: e.g. the study reported all outcomes specified in the protocol

  • High risk of bias: e.g. it was suspected that outcomes had been selectively reported

  • Unclear risk of bias: e.g. it was unclear whether outcomes had been selectively reported

• Other biases

  • Low risk of bias: if we did not suspect any other source of bias and the trial appeared to be methodologically sound

  • High risk of bias: if we suspected that the trial had been prone to an additional bias

  • Unclear risk of bias: if we are uncertain whether an additional bias might had been present

Measures of treatment effect

We used the following measures of the effect of treatment.

• For time‐to‐event data, we used the hazard ratio.

• For dichotomous outcomes, we analyzed data based on the number of events and the number of people assessed in the intervention and comparison groups. We used these to calculate the risk ratio and 95% confidence interval (CI).

• For continuous outcomes, we analyzed data based on the mean, standard deviation and number of people assessed for both the intervention and comparison groups to calculate mean difference between treatment arms, with a 95% CI. If the mean difference was reported without individual group data, we planned to use this to report the study results. If more than one study measured the same outcome using different tools, we planned to calculate the standardized mean difference and 95% CI using the inverse variance method in RevMan 2014.

Unit of analysis issues

If any trials with multiple treatment groups had been identified, we would have divided the ‘shared’ comparison group into the number of treatment groups and comparisons between each treatment group and treated the split comparison group as independent comparisons.

Dealing with missing data

We did not contact study authors to obtain missing data (participant, outcome or summary data). We reported on the levels of loss to follow‐up and assessed this as a source of potential bias. We explored the impact of including studies with high levels of missing data in the overall assessment of treatment effect by using sensitivity analysis. We did not impute missing outcome data for the primary outcome.

Assessment of heterogeneity

We assessed the degree of heterogeneity among trials using I² (Higgins 2003) and Chi² statistics (Deeks 2001). We regarded heterogeneity to be substantial if I² was greater than 50% and either Tau² was greater than zero, or there was a low P value (less than 0.10) in the Chi² test for heterogeneity. If substantial heterogeneity was found, we would have used subgroup and sensitivity analyses to explore the causes of heterogeneity.

Assessment of reporting biases

If we identify more than ten studies in future updates of this review, we will examine funnel plots corresponding to meta‐analysis of the primary outcome to assess the potential for small‐study effects such as publication bias. We plan to assess funnel plot asymmetry visually, and if asymmetry is suggested by a visual assessment, we will perform exploratory analyses to investigate it.

Data synthesis

We carried out statistical analysis using Review Manager 5 (RevMan 2014). We used a fixed‐effect model for combining data if there was no substantial heterogeneity. If substantial statistical heterogeneity was detected, we used a random‐effects meta‐analysis to produce an overall summary and the results were presented as the average treatment effect with 95% CIs (DerSimonian 1986).

• For time‐to‐event data, we pooled hazard ratios using the generic inverse variance facility of RevMan 2014.

• For dichotomous outcomes, we calculated the risk ratio for each study and then pooled them.

• For continuous outcomes, we pooled the mean differences between the treatment arms at the end of follow‐up, where all trials measured the outcome on the same scale; otherwise we planned to pool standardized mean differences.

When we were unable to pool the data statistically using meta‐analysis we conducted a narrative synthesis of results. We presented the major outcomes and results, organised by intervention categories according to the major types or aims (or both) of the identified interventions. Depending on the assembled research, we may also in future explore the possibility of organising the data by population. Within the data categories we will explore the main comparisons of the review.

We presented the overall quality of the evidence for each outcome listed below, using the GRADE approach, which took into account issues not only related to internal validity (risk of bias, inconsistency, imprecision, publication bias) but also to external validity, such as directness of results (Langendam 2013). We created a 'Summary of findings' table based on the methods described Chapter 12.2 of the Cochrane Handbook for Systematic Reviews of Interventions (Schunemann 2011), using GRADEpro GDT. We used the checklist to maximize consistent GRADE decisions and the GRADE Working Group quality of evidence definitions (Meader 2014). We downgraded the evidence from 'high' quality by one level for serious limitations (or two levels for very serious limitations) for each outcome, and outlined our rationale in the footnotes.

• High quality: further research is very unlikely to change our confidence in the estimate of effect.

• Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.

• Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

• Very low quality: we were very uncertain about the estimate.

We included the following outcomes included in the 'Summary of findings' table.

1. Overall survival

2. Quality of life assessment

3. Progression‐free survival

4. Adverse events

   1. Incidence of breast cancer

   2. Incidence of thromboembolic events (DVT/ PE, stroke, MI)

   3. Incidence of gallstones

Subgroup analysis and investigation of heterogeneity

We did not identify any substantial heterogeneity; therefore subgroup analysis were not conducted. In future updates, if more studies are included and substantial heterogeneity is identified, it will be of clinical interest to investigate the safety (risk and benefits) for the prespecified outcomes in this protocol for the following factors.

• Menopausal status at diagnosis: premenopausal versus postmenopausal. If menopausal status is not extractable from studies, we will instead analyze by age (under 50 years versus 50 years or older).

• Hysterectomy versus no hysterectomy.

• Stage of cancer: stage I to II versus stage III to IV

• Tumour types: endometrioid versus non‐endometrioid

• Positive or negative oestrogen receptor, progesterone receptor or BRCA mutation status.

Sensitivity analysis

If necessary in any future update, we will use sensitivity analyses to assess the cause of substantial heterogeneity by omitting the studies with at a high risk of bias.