Criteria for considering studies for this review
Types of studies
We will search for randomized controlled trials (RCTs), as they provide the highest level of evidence.
If no RCTs are found, we will expand our criteria to include nonrandomized comparative studies (cohort and case-control studies).
Types of participants
Women with a diagnosis of ductal carcinoma in situ or invasive breast cancer according to the histopathological criteria of the World Health Organization (Tavassoli 2003), regardless of age, time of onset, or disease stage.
Types of interventions
Skin-sparing mastectomy (ie, removal of all breast tissue and the nipple-areola complex and preservation of the entire skin) compared with radical modified mastectomy (ie, removal of the skin that overlies the breast including the nipple and areola) for the treatment of ductal carcinoma in situ or invasive breast cancer, regardless of adjuvant therapy.
In this review, only women who have had breast reconstructive surgery will be included.
Types of outcome measures
Local recurrence incidence rate and time-to-event for recurrence during the follow-up period, and considered separately for ductal carcinoma in situ and early breast cancer.
Adverse events: local surgical complications or systemic surgical complications such as thromboembolic events.
Explantation of implant/expander.
Skin necrosis (nipple, areola, or flap necrosis).
Skin necrosis with revisional surgery.
Cosmetic results: using the BREAST-Q (Breast-Q) questionnaire. If no validated instruments are included in the studies, participant and professional opinion will be documented.
Quality of life and participant satisfaction: using the EORTC QLQ-BR23 (Fayers 1999) and BREAST-Q (Breast-Q) questionnaires. If studies do not include validated instruments, a narrative description of participant opinion will be provided.
Search methods for identification of studies
We will search the following databases.
Searching other resources
We will try to identify further studies from reference lists of identified relevant trials and reviews. A copy of the full article for each reference reporting a potentially eligible trial will be obtained. When this is not possible, attempts will be made to contact study authors to request additional information.
Data collection and analysis
Selection of studies
Two review authors (BS and MDR) will independently examine the titles and abstracts of articles identified in the search as potentially relevant trials. After this initial assessment, we will obtain full versions of all potentially relevant articles. We will consult a third review author (JLBB) to help resolve disagreements. Excluded studies will be recorded in the "Characteristics of excluded studies" table. Studies published only in English will be included.
Data extraction and management
We will extract and record data on data extraction forms, which we will develop. Two review authors (BS and MDR) will independently undertake full data extraction and will consult a third review author (JLBB) to help resolve disagreements. For those studies with more than one publication, we will extract data from all publications, and the version with the longest follow-up will be considered. We will seek unpublished data concerning outcomes of interest from authors by letter, as stated above. We will enter references into Review Manager 5 software (RevMan).
We will include the following information from individual studies on data extraction forms.
Study design, study setting, inclusion/exclusion criteria, methods used to control for confounders, adjusted and unadjusted outcome measures.
Participant population (eg, age, type of surgical procedure, tumor stage, grade of tumor, molecular subtypes).
Details of interventions.
Withdrawals, length and method of follow-up, and number of participants followed up.
We will enter quantitative data into RevMan software (RevMan) to carry out the analyses.
Assessment of risk of bias in included studies
Assessment of the methodological quality of each study will be undertaken independently by two review authors (BS and RR), as per the data extraction process described in the Cochrane Handbook for Systematic Reviews of Interventions, version 5.1.0 (Higgins 2011).
Appraisal will follow The Cochrane Collaboration's risk of bias (RoB) tool (Higgins 2011), the EPOC risk of bias criteria (Cochrane EPOC 2013), and recommendations by Norris et al (Norris 2013). Risk of bias will be assessed under the following domains: selection bias, performance/detection bias, attrition bias, reporting bias, and other bias. For each RoB domain and its associated specific questions outlined below, review authors will assign "high risk," "low risk," or "unclear risk."
Score "low risk" if a random component in the sequence generation process was described (eg, referring to a random number table). Score "high risk" when a nonrandom method was used (eg, performed by date of admission). Nonrandomized studies should be scored "high risk." Score "unclear risk" if not specified in the paper.
Score "low risk" if participants and investigators enrolling participants could not foresee assignment (eg, because a centralized randomization scheme, an on-site computer system, or sealed opaque envelopes were used). Nonrandomized studies should be scored "high risk." Score "unclear risk" if not specified in the paper.
For nonrandomized studies the following questions should also be assessed.
Score "low risk" if performance or participant outcomes were measured before the intervention was provided and no important differences were present across study groups. Score "high risk" if important differences were present and were not adjusted for in the analysis. Score "unclear risk" if not specified in the paper.
Score "low risk" if baseline characteristics of the study and control providers were reported and similar. Score "unclear risk" if this was not clear in the paper (eg, characteristics were mentioned in the text, but no data were presented). Score "high risk" if no report of characteristics was provided in text or tables or if differences between control and intervention providers were noted.
Score "low risk" if appropriate methods were used to adjust for potential confounding. Score "Unclear risk" if the methods used to adjust for confounding were not reported in the paper. Score "High risk" if potential confounding from the following variables was not addressed: adjuvant radiotherapy, age, surgical techniques, stage of disease, ductal carcinoma in situ or invasive breast cancer, and chemotherapy.
Score "low risk" if the study authors stated explicitly that the primary outcome variables were assessed blindly or if the outcomes were objective (eg, length of hospital stay). Score "high risk" if the outcomes were not assessed blindly. Score "unclear risk" if not specified in the paper.
Score "low risk" if missing outcome measures were unlikely to bias the results (eg, reasons for missingness unlikely to be related to true outcome, missing outcome data balanced in numbers across intervention groups with similar reasons for missing data, or missing data imputed using appropriate methods). Score "high risk" if missing outcome data were likely to bias the results. Score "unclear risk" if not specified in the paper.
Score "low risk" if no evidence suggests that outcomes were selectively reported (eg, the study had a protocol prespecifying the outcomes, all relevant outcomes in the methods section are reported in the results section), Score "high risk" if some prespecified outcomes were subsequently omitted from the results. Score "unclear risk" if not specified in the paper.
Score "low risk" for each outcome if no evidence suggests that analyses were selectively reported (eg, analyses were defined in the methods section of the protocol or paper). Score "high risk" if evidence suggests selective analysis reporting (eg, multiple adjusted analyses have been carried out and only one reported, unusual cut-points have been used for categorizing an outcome). Score "unclear risk" if unclear from the paper.
Classification of study designs—study designs that may be included and defined as follows
Prospective cohort study: groups of exposed and nonexposed individuals that have been followed over time to compare incidence (or rate of death from disease) between groups (Gordis 1996). In prospective cohort studies, recruitment, exposure/intervention, and outcomes must all have occurred after the study was set up.
Retrospective cohort study: groups of exposed and nonexposed individuals that have been followed over time to compare incidence (or rate of death from disease) between groups (Gordis 1996). In retrospective cohort studies, outcomes must have occurred before the study was set up or must have been collected afterward, or both.
Case-control study: a study that compares people with a specific outcome of interest (cases) versus people from the same source population but without that outcome (controls), to examine the association between outcome and prior exposure.
Measures of treatment effect
We will report time-to-event outcomes (eg, overall survival, local recurrence) as hazard ratios with 95% confidence intervals (CIs). If necessary, hazard ratios will be estimated using the methods of Parmar et al (Parmar 1998). If it is not possible to estimate hazard ratios from all studies, the number of events (eg, deaths, recurrences) over five years and ten years of follow-up will be treated as dichotomous outcomes.
We will report dichotomous outcomes (eg, recurrence, distant disease, explantation of implant/expander, hematoma, seroma, rehospitalization, skin necrosis [nipple, areola, or flap necrosis], skin necrosis [nipple, areola, or flap necrosis] versus revisional surgery, infection, and cosmetic results) as risk ratios (RRs) and risk differences (RDs) with 95% CIs. We will pool the data for meta-analysis using the pooled log-RR, when appropriate. We will report continuous outcomes (eg, quality of life) as mean differences (MDs) with 95% CIs.
Given the current approach for ductal carcinoma in situ or invasive breast cancer, it seems reasonable to consider the treatment effect and reading the primary outcome as a noninferiority question. Thus, the noninferiority bound for the hazard ratio will be 1.13 (based on a 10-year survival rate of 24% for conventional mastectomy with a cutoff value of 27% for skin-sparing mastectomy). For local recurrence, the noninferiority bound for the hazard ratio will be 2.67 (based on 3% 10-year local recurrence for conventional mastectomy with a cutoff value of 8% for skin-sparing mastectomy).
Unit of analysis issues
We anticipate that the appropriate unit of analysis will be the individual participant, rather than the surgical unit, hospital, or center.
Dealing with missing data
When data are missing or are unsuitable for analysis (eg, intention to treat is not used), we will contact the study authors to request further information. When data are missing to the extent that the study cannot be included in the meta-analysis and attempts to retrieve data have been exhausted, we will present the results in the review and will discuss the context of the findings.
Assessment of heterogeneity
If appropriate, we will assess statistical heterogeneity using the Chi2 statistic (Cochran 1954). We will also assess heterogeneity between studies using the I2 statistic to examine the percentage of total variation across studies due to heterogeneity rather than chance. An I2 value of 30% to 60% may represent moderate heterogeneity, and values greater than 50% may be considered to show substantial heterogeneity (Higgins 2003).
We will investigate factors as potential causes of heterogeneity in the included studies using the framework below (χ2 or Chi2).
Clinical diversity: includes study location and setting, full characteristics of participants, comorbidities, and treatments that participants may be receiving on trial entry. We will consider how outcomes were measured, the definitions of outcomes, and how they were recorded. Depending on the extent of clinical diversity, we will analyze studies separately or will present the results using a narrative approach.
Methodological diversity: includes assessment of the randomization process, study quality, and analytical method.
Assessment of reporting biases
We will search for protocols of included trials using PubMed (National Library of Medicine) and other trial registries, when possible. We will contact study authors to attempt to establish a full data set or reasons for the nonreporting of certain outcomes.
We will synthesize data using RevMan (RevMan). We will base the selection of fixed-effect or random-effects models for data synthesis on a number of factors. When significant clinical or methodological heterogeneity exists, we will use a random-effects model. Otherwise, we will use the fixed-effect model.
We will combine data using the inverse variance method on the log-HR scale for time-to-event outcomes, on the log-RR scale for dichotomous outcomes, and on the MD scale for continuous outcomes. For random-effects meta-analysis, we will use the DerSimonian and Laird method.
When the data are too diverse to permit combining of effect sizes in a meaningful or valid manner, we will present the results of individual studies in table and graphical formats and will use a narrative approach to summarize the data.
Subgroup analysis and investigation of heterogeneity
As we expect a small number of published studies, we do not envisage that it will be possible to perform subgroup analyses. However, we would like to consider the following subgroups.
Participants who received adjuvant radiotherapy versus participants who did not receive radiotherapy.
Younger (< 50 years of age) versus older (≥ 50 years of age) women.
Cancer stage as per TNM (size and/or extent of the primary tumor [T], amount of spread to nearby lymph nodes [N], and presence of metastasis [M] or secondary tumors) classification system (NCI 2013).
Systemic therapy (chemotherapy or endocrine therapy) versus no systemic therapy.
We will conduct sensitivity analyses by excluding trials (1) of low methodological quality, (2) with a small sample size, and (3) with short follow-up. We will consider and discuss the results of these analyses in comparison with overall findings.