Following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines, a comprehensive research of several databases from each database's inception to May 11, 2020, for studies in both English and Spanish languages, was conducted. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, and Scopus. The search strategy was designed and conducted by an experienced librarian, with input from the study's principal investigator. Controlled vocabulary supplemented with keywords was used to search for studies of de-transition and regret in adult patients who underwent gender confirmation surgery. The actual strategy listing all search terms used and how they are combined is available in Supplemental Digital Content 1. (See Supplemental Digital Content 1, which displays the search strategy. https://links.lww.com/PRSGO/B598.)
Search results were exported from the database into XML format and then uploaded to Covidence. The study selection was performed in a 2-stage screening process. The first step was conducted by 2 screeners (V.P.B. and S.S.B.), who reviewed titles and abstracts and selected those of relevance to the research question. Then, the same 2 screeners reviewed full text of the remaining articles and selected those eligible according to the inclusion and exclusion criteria (Figure 1). If disagreements were encountered, a third reviewer (O.J.M.) moderated a discussion, and a joint decision between the 3 reviewers was made for a final determination. Inclusion criteria were all the articles that included patients aged 13 years or more who underwent GAS and report regret or de-transition rates, and observational or interventional studies in English or Spanish language. Exclusion criteria were letter to the editors, case series with <10 patients, case reports correspondences, and animal studies.
After selecting the articles, we assessed study characteristics. We identified year of publication, country in which the study was conducted, population size, and number of transmasculine and transfemenine patients with their respective mean age (expressed with SD, range, or interquartile range if included in the study). In addition, we extracted information of the method of data collection (interviews versus questionnaires), number of regrets following GAS, as well as the type of surgery, time of follow-up, and de-transition procedures. We classified the type of regret based on the patient's reasons for regret if they were mentioned in the studies. We used the Pfäfflin and Kuiper and Cohen-Kettenis classifications of regret (Table 1).[20,23]
To assess the risk of bias within each study, the National Institute of Health (NIH) quality assessment tool was used. This tool ranks each article as "good," "fair," or "poor," and with this, we categorized each article into "low risk," "moderate risk," or "high risk" of bias, respectively.
Our primary outcome of interest was the prevalence of regret of transgender patients who underwent any type of GAS. Secondary outcomes of interest were discriminating the prevalence of regrets by type gender transition (transfemenine and transmasculine), and type of surgery.
Data Analysis and Synthesis
The binominal data were analyzed, and the pooled prevalence of regret was estimated using proportion meta-analysis with Stata Software/IC (version 16.1). Given the heterogeneity between studies, we conducted a logistic-normal-random-effect model. The study-specific proportions with 95% exact CIs and overall pooled estimates with 95% Wald CIs with Freeman-Turkey double arcsine transformation were used. The effect size and percentage of weight were presented for each individual study.[25,26]
To evaluate heterogeneity, I2 statistics was used. If P < 0.05 or I2 > 50%, significant heterogeneity was considered. A univariate meta-regression analysis was performed to assess the significance in country of origin, tools of measurement, and quality of the studies.
To assess publication bias, we used funnel plot graphic and the Egger test. If this test showed us no statistical significance (P > 0.05), we assumed that the publication bias had a low impact on the results of our metanalysis. To assess the impact of the publication bias on our missing studies, we used the trim-and-fill method.
A sensitivity analysis was conducted to assess the influence of certain characteristics in the magnitude and precision of the overall prevalence of regret. The following characteristics were excluded: <10 participants included, and the presence of a high risk of bias.
Plast Reconstr Surg Glob Open. 2021;9(3):e3477 © 2021 Lippincott Williams & Wilkins