Background: Gynaecomastia ranges from localised fibroglandular enlargement to marked ptotic hypertrophy, and reported complication and revision rates after surgical correction vary considerably. 

Methods: This study comprises a random-effects re-analysis of aggregate data from a published systematic review and a retrospective cohort study. Arm-level data from 17 technique-specific subgroups comprising 7,294 patients were analysed using Freeman–Tukey random-effects pooling, subgroup analysis, meta-regression and prediction intervals. Because Simon grade was absent from the source data and inferred from the operative technique used, grade-specific estimates were indirect rather than observed. A consecutive 300-patient single-practice cohort was retrospectively assessed over a 365-day records-based observation period and compared with technique-standardised expected rates. The primary revision endpoint was any further procedure within 365 days. Further procedures were classified by the treating surgeons, without independent review, as elective refinement, residual-tissue re-excision, unplanned complication-related reoperation or true glandular recurrence.

Results: The pooled complication rate was 14.44% (95% CI, 9.04–20.73%), with a 95% prediction interval of 0.11–44.88% and substantial heterogeneity (I² = 96.9%). Complication rates differed across indirectly assigned Simon-grade groups: 14.85% for Grades I–IIa, 10.24% for Grades IIa–IIb, and 32.64% for Grades IIb–III (p < 0.001), with the highest rate observed in the skin-excision subgroup. Glandular excision alone was associated with 2.61 times the complication rate of excision with adjunctive liposuction. In the practice cohort, 11 patients (3.67%) experienced complications. The technique-standardised complication ratio was 0.30, although the observed rate remained within the prediction interval. Of 36 subsequent procedures, 32 were elective refinements, three were residual-tissue re-excisions, and one was an unplanned return to theatre. No true glandular recurrences occurred. The clinically indicated reoperation rate was 1.33%, compared with an overall revision rate of 12.00%.

Conclusions: Complication and reoperation rates should be interpreted as stratified outcomes influenced by gynaecomastia grade, operative complexity, adjunctive procedures and outcome definitions. Without such stratification, comparisons may be confounded, and the true rates of complications and clinically indicated reoperations may be misrepresented.

Gynaecomastia is benign proliferation of true glandular tissue in the male breast, distinct from lipomastia, in which enlargement is predominantly adipose [1,2]. This distinction has direct operative implications. Liposuction can reduce the fatty component and improve chest contour, but it may leave a firm fibroglandular disc untreated when used alone, contributing to residual fullness or subsequent reoperation [3]. Surgical planning must therefore address the relative contributions of gland, fat, skin excess, areolar enlargement, ptosis, and nipple–areola complex position rather than apply a single surgical technique or modality to all grades. Simon's classification remains the most widely used clinical grading system: Grade I denotes minor enlargement without skin excess; Grade IIa, moderate enlargement without skin excess; Grade IIb, moderate enlargement with minor skin excess; and Grade III, marked enlargement with substantial skin excess [1]. Rohrich's system incorporates estimated tissue volume, tissue composition, and ptosis [4], whereas Cordova and Moschella classify severity according to the relationship of the nipple–areola complex to the inframammary fold [5] (Table 1).

Table 1: Commonly used clinical classification systems for gynaecomastia

Simon’s classification was selected for the present analysis because its central variable—skin excess—most directly determines whether treatment can remain skin-sparing or requires formal skin excision. The systems of Rohrich and of Cordova and Moschella could not be applied, because the contributing studies did not report resected tissue volume, tissue composition, degree of ptosis, or the position of the nipple–areola complex relative to the inframammary fold. Lower gynaecomastia grades are usually treated with suction-assisted contouring and limited direct glandular excision. Intermediate disease commonly requires glandular excision combined with liposuction, with selective energy-assisted skin contraction or circumareolar skin reduction/periareolar mastopexy when skin redundancy or areolar enlargement persists. Advanced ptotic disease may require a skin-reducing mastectomy with nipple–areola complex (NAC) repositioning on a vascularised pedicle or, in selected severe cases, as a free graft. Each additional operative step addresses a distinct aspect of the gynaecomastia presentation, but also alters scar placement, dead space, closure tension, NAC perfusion and sensation, and the expected complication profile [2, 6, 7]. Reported complication and reoperation rates vary several-fold. A systematic review of 94 studies found complication rates of 14.87% after aspiration, 30.64% after excision, and 11.76% after combined approaches [7]; another review reported major complication rates from 0% to 33% [8], whereas national database studies using 30-day follow-up periods reported rates of 1.9% and 4.4% [9, 10]. This variation may reflect differences in severity, operative technique, duration and completeness of follow-up, data ascertainment, or outcome definition. Technique labels can be misleadingly broad: “excision” may refer to limited removal of gland, excision combined with periareolar mastopexy, or a more extensive skin-reducing mastectomy. Likewise, “revision” may denote emergency treatment of a complication, re-excision of residual tissue, treatment of true regrowth, or elective refinement of an otherwise acceptable result. These operations and outcomes are clinically distinct but are often pooled.

This study examined complications and revisions as related but clinically distinct outcomes by combining two complementary sources of evidence. A random-effects meta-analysis of 7,294 patients generated reference estimates stratified by surgical technique, adjunctive liposuction, indirectly assigned Simon grade, and operative complexity. A 300-patient cohort from the authors’ practice was then benchmarked against those estimates and re-analysed using progressively narrower definitions of revision. The aim of the study was therefore threefold: to generate technique-, grade- and complexity-stratified reference estimates of complication risk from the available published evidence; to benchmark a consecutive single-practice cohort against those estimates using prediction intervals rather than pooled means; and to quantify the extent to which a reported revision rate is determined by the endpoint definition applied rather than by the clinical events themselves. The design was a systematic review with random-effects meta-analysis, undertaken as a secondary re-analysis of previously published aggregate arm-level data, together with an accompanying retrospective observational cohort study. Because the grade and complexity strata were inferred from technique rather than observed at patient level, the stratified estimates are presented throughout as hypothesis-generating, and the study is intended to define what can legitimately be learned from the existing evidence and what cannot, rather than to establish comparative effectiveness.

Study Design and Data Sources

This was a systematic review with random-effects meta-analysis, undertaken as a secondary re-analysis of previously published aggregate arm-level data, combined with a retrospective audit of a single-practice 300-patient cohort. The published-data analysis generated reference complication estimates and explored differences according to operative technique, use of adjunctive procedures such as liposuction, indirectly assigned Simon grade and a four-level operative complexity score. The audit benchmarked observed outcomes against these estimates, described the distribution of operative techniques and adjunctive procedures, and assessed how the definition of revision influenced interpretation. For the re-analysis, arm-level event counts and denominators for 17 technique-specific subgroups were extracted from the largest published systematic review of complications following gynaecomastia correction. Unlike reviews reporting summary percentages alone, it provided event counts and denominators for each technique subgroup across 94 studies comprising 7,294 patients [7]. Each technique subgroup was treated as the unit of analysis. No de novo study-level search, screening, or data extraction was performed, and the 94 primary studies were not independently retrieved or appraised. Because this study re-analysed aggregated arm-level data rather than undertaking de novo study-level collection, the analysis necessarily inherits the methodology, eligibility decisions, data extraction, technique classification and quality appraisal of the source review. Any error, misclassification or bias present in that review is therefore propagated into the present analysis and could not be independently verified. Within-subgroup heterogeneity and primary-study risk of bias could not be evaluated; therefore, the reported heterogeneity may underestimate the full variability across the contributing studies, and no independent risk-of-bias or certainty-of-evidence assessment, such as Grading of Recommendations Assessment, Development and Evaluation (GRADE), could be undertaken. The pooled estimates should therefore be interpreted as a formal random-effects re-analysis of aggregate data from a previous review, rather than as independent confirmation of its findings, and this study remains conditional on the validity of that review. No protocol was prospectively registered and the study was not registered with PROSPERO because it involved re-analysis of an existing published aggregate dataset. Reporting followed applicable PRISMA principles, with these departures explicitly acknowledged [11]. No contributing study reported complications directly according to Simon grade or consistently separated the individual components of combined procedures. Each technique subgroup was therefore indirectly mapped to the Simon grade and level of operative complexity with which it is conventionally associated (Table 2). This mapping represents an assumption imposed by the authors on the source data rather than a property of the data themselves: no patient in the pooled dataset had a recorded Simon grade, and no contributing study assigned one. Accordingly, all grade-stratified results reported below refer to technique strata labelled according to their assumed correspondence with Simon grade. These findings are hypothesis-generating, depend on the validity of that assumed correspondence across the contributing studies, and must not be interpreted as directly observed grade-specific complication rates.

Table 2: Mapping of technique subgroups to operative complexity and Simon grade.

• Multiple event categories may be counted for one patient. Follow-up and reporting thresholds vary. Open excision includes different excisional approaches

• Complexity levels: 1, aspiration only; 2, gland removed by minimal-access route with or without liposuction; 3, open glandular excision with liposuction; 4, open excision encompassing formal skin excision. 

• Crude rates are simple arm-level proportions. Pooled rates are random-effects estimates (Freeman–Tukey double arcsine, DerSimonian–Laird) and differ from the crude figures because pooling weights each subgroup by its precision rather than by its size; the overall crude rate of 19.29% is dominated by the single 2,540-patient open-excision subgroup, whereas the pooled estimate of 14.44% is not.

• Two strata span more than one complexity level: pull-through and fragmentation techniques sit at level 2 within the Grade I–IIa stratum, and open excision with liposuction sits at level 3 within the Grade IIa–IIb stratum. Grade attribution is indirect throughout, inferred from the technique conventionally used rather than recorded at patient level. Suction-assisted liposuction (18 of 21) is the extreme outlier excluded in sensitivity analysis.

Aspiration-only procedures were assigned to complexity level 1; minimal-access excision, with or without liposuction, to level 2; open excision with liposuction to level 3; and open excision involving formal skin reduction to level 4. In clinical practice, level 4 may encompass circumareolar skin excision or periareolar mastopexy, wider skin-reducing mastectomy, pedicled NAC repositioning, or free NAC grafting. However, these individual procedures could not be distinguished in the source dataset. The grade and complexity analyses are therefore indirect, and the highest-risk stratum should not be interpreted as a procedure-specific estimate for any individual mastopexy, skin-resection pattern, or NAC-transfer technique. The complication rate associated with this stratum reflects a heterogeneous grouping of operative techniques rather than a directly observed rate for any specific procedure or for high-grade gynaecomastia itself. The practice cohort comprised all patients coded as undergoing primary gynaecomastia surgery in the authors’ private London practice between 1 January and 30 June 2025. Nine exact duplicate entries were removed from 309 records, leaving 300 unique patients. Repeat procedures were linked by name and confirmed by date of birth. Three same-name records were verified as different patients, thereby avoiding an 8.3% inflation of the repeat-procedure numerator (Figure. 1).

Figure 1: Practice cohort data collection. Nine exact duplicates were removed from 309 records. Three same-name records were confirmed as different patients by date of birth; without that check, the repeat-procedure numerator would have been inflated by 8.3%.

Outcomes were defined a priori and counted per patient rather than per breast. Complications included any documented haematoma, seroma, infection or cellulitis, wound dehiscence, or nipple–areola complex ischaemia, ranging from superficial epidermolysis to partial- or full-thickness necrosis. The primary revision endpoint was any further procedure within 365 days of the index operation, regardless of indication. This inclusive endpoint was chosen for external comparison because it does not depend on the operating surgeon’s judgement about indication. Further procedures were retrospectively classified as unplanned complication-related reoperation, residual-tissue re-excision, elective refinement or “touch-up”, or true glandular recurrence. Residual tissue was defined as palpable tissue remaining after the primary procedure. Recurrence required an initially satisfactory contour followed by documented enlargement. Elective refinement denoted patient-requested aesthetic surgery when the surgeon recorded no complication, residual gland, or regrowth. Clinically indicated reoperation comprised complication-related reoperation and residual-tissue re-excision; recurrence would also have qualified, but none occurred. Complications or further procedures managed elsewhere could have been missed. These secondary classifications were assigned by the operating surgeon without independent or blinded adjudication. In particular, classifying patient-requested surgery as elective refinement relies on the surgeon’s judgement that the result was satisfactory. Misclassification in favour of the practice is therefore possible, and the distinction between elective and clinically indicated procedures cannot be verified from the records. This is an acknowledged limitation of the study.

Complication rates were transformed using the Freeman–Tukey double-arcsine method, which stabilises variance and accommodates zero-event strata [12], and pooled with a DerSimonian–Laird random-effects model [13]. Estimates were back-transformed at the harmonic mean sample size [14]. Heterogeneity was assessed using Cochran's Q, τ², and I² [15]. A 95% prediction interval, calculated as t(k − 2)√(τ² + SE²), represented the true complication rate expected for an individual centre and was therefore considered more appropriate for comparison than the confidence interval around the pooled mean. Three prespecified sensitivity analyses were performed: logit pooling, exclusion of strata with fewer than 20 patients, and exclusion of the single stratum with the highest complication rate. Subgroup differences were assessed by Q-between. Meta-regression used the prespecified ordinal operative-complexity score, with residual τ² estimated on fixed weights, and small-study effects were examined by Egger regression [16].

Cohort outcomes were reported with exact Clopper–Pearson intervals [17]. Comparisons with reference estimates used exact binomial and Fisher exact tests, Katz risk ratios, and indirect standardisation for technique mix, with Holm–Bonferroni correction for multiplicity [18]. Robustness was assessed by fragility index and by the benchmark rate below which the cohort comparison would cease to be significant. Tests were two-sided with α = 0.05. Analyses were performed in Python 3.10.12 using custom scripts for exact binomial inference, Fisher’s exact tests, random-effects meta-analysis, estimation of between-study heterogeneity, calculation of prediction intervals, subgroup comparisons, meta-regression and sensitivity analyses.

Pooled Complication Rate and Heterogeneity

Across 17 technique strata and 7,294 patients, 1,407 complications were reported. The random-effects pooled complication rate was 14.44% (95% CI, 9.04%–20.73%), with extreme heterogeneity (Q = 511.4, df = 16, I² = 96.9%, τ² = 0.0225; p < 0.001). The findings were robust to sensitivity analyses: logit-transformed pooling yielded 14.95% (95% CI, 10.24%–21.30%), exclusion of strata with fewer than 20 patients yielded 15.01% (95% CI, 9.54%–21.41%), and exclusion of the extreme outlier yielded 11.79% (95% CI, 6.93%–17.59%); I² remained 96.8%–97.1% throughout (Table 3, Figure. 2).

Table 3: Pooled complication rate and sensitivity analyses

Figure 2: Forest plot of complication rates across 17 technique-specific complication strata and 7,294 patients. The random-effects pooled estimate was 14.44% (95% CI 9.04–20.73; I² = 96.9%), with technique-specific rates from 0% to 85.7%.

The 95% prediction interval extended from 0.11% to 44.88%. Complication rates were 14.85% for indirectly assigned Grade I–IIa, 10.24% for Grade IIa–IIb, and 32.64% for Grade IIb–III including skin resection (p < 0.001). Risk was U-shaped rather than linearly related to operative complexity, with the lowest rates in intermediate combined procedures.

Technique, Operative Complexity, and Adjunctive Procedures

Random-effects pooled complication rates were 18.07% (8.65–29.82) for aspiration-only techniques, 13.15% (0.92–32.96) for excision, and 11.99% (9.74–14.43) for combined excision and liposuction. The between-group test was not statistically significant (Q = 1.56, df = 2; p = 0.46), reflecting substantial within-group heterogeneity. Aggregated arm-level comparisons showed higher complication risks for excision alone than for combined excision–liposuction (crude risk ratio, 2.61; 95% CI, 2.35–2.89; p < 0.001), and for open excision than for open excision with liposuction (crude risk ratio, 2.86; 95% CI, 2.53–3.25). Although the excision-versus-combined comparisons were directionally consistent, the crude analyses used different weighting and did not account for between-stratum heterogeneity or potential case-mix differences. They should therefore be interpreted as descriptive, unadjusted associations rather than causal effects of surgical technique. When strata were regrouped according to whether liposuction was used as an adjunct to excision, single-modality procedures pooled at 16.25% (7.95–26.47), compared with 11.99% (9.74–14.43) for combined procedures (Q = 1.29; p = 0.26 by random effects). The association is clinically plausible because liposuction can debulk the adipose component, improve peripheral contouring, and reduce the extent of open dissection, but the aggregate data do not establish causality. Energy-assisted skin tightening, periareolar mastopexy, wider skin excision, and NAC grafting could not be analysed as independent adjuncts because the source review did not provide separable event denominators. A 2024 review identified limited and heterogeneous evidence for laser-, ultrasound-, radiofrequency-, and other skin-retraction adjuncts [6]. These modalities therefore remain contextual evidence rather than quantitatively pooled comparators in the present analysis.

Simon Grade and Operative Complexity

Under the indirect, technique-based grade mapping, pooled complication rates were 14.85% (9.56–21.02) for Grade I–IIa, 10.24% (6.27–14.88) for Grade IIa–IIb, and 32.64% (30.83–34.47) for Grade IIb–III, the only grade band encompassing skin-excisional surgery. Grade bands differed significantly (Q = 75.3, df = 2; p < 0.001), with the highest band carrying 2.20 times the rate of the lowest (Fig. 3a). These bands were assigned from technique and were not observed at patient level. The comparison is therefore a comparison between technique strata that have been labelled by grade, and the 32.64% figure is an inference from a single technique stratum rather than a direct analysis of patient-level grade data. Its interpretation as evidence of increased complications in high-grade gynaecomastia is therefore hypothesis-generating and requires confirmation in patient series with directly recorded grades.

Figure 3: Risk by indirectly assigned Simon grade band (a) and increasing operative complexity (b). Grade bands differed significantly (Q = 75.3, p < 0.001), with the stratum encompassing open skin-resecting surgery at 32.64%. The complexity pattern was U-shaped (18.07%, 10.20%, 11.39%, and 32.64%), with the lowest rates in intermediate combined-technique strata. Specific mastopexy technique, skin-excision and NAC-repositioning procedures were not reported separately in the source data. 

Meta-regression found no evidence of a linear trend in complication rates across the operative-complexity score. This score ranks procedures by operative extent, from 1 to 4, but is not a validated quantitative measure of clinical complexity. The analysis therefore tested for a trend across categories, rather than for an association between complication rates and independently measured clinical complexity (Table 4).

Table 4: Pooled complication rates by indirectly assigned Simon grade band and operative-complexity level. 

Grade bands differ significantly, but the relationship with operative complexity is U-shaped rather than monotonic. The excess risk is concentrated in the stratum encompassing skin reduction. The source data did not distinguish periareolar mastopexy, wider skin excision, pedicled NAC repositioning, or free NAC grafting within that subgroup. Neither the linear association (slope 0.014, 95% CI −0.080 to 0.109; p = 0.77) nor the quadratic term assessing curvature (p = 0.087) was statistically significant. Nevertheless, the observed rates across levels 1–4 were 18.07%, 10.20%, 11.39%, and 32.64%, suggesting a U-shaped pattern with higher risk at both the aspiration-only and skin-reduction ends of the spectrum (Figure. 3b). Because the published studies combined different open glandular-excision and skin-reducing techniques, the 32.64% estimate cannot be attributed to any single approach — gland excision with periareolar mastopexy, a specific skin-resection pattern, pedicled NAC repositioning, or free nipple grafting. Compared with combined techniques, excision alone had higher rates of seroma (8.32% versus 0.93; ratio 8.95), over-resection (1.30% versus 0.11%; ratio 11.90), pathological scarring (4.70% versus 1.15%; ratio 4.09), haematoma (7.71% versus 2.68%; ratio 2.87), and nipple–areola necrosis or abrasion (1.27% versus 0.44%; ratio 2.89). The pattern reversed for contour-related outcomes: contour irregularity or redundant skin was 3.7 times more frequent after combined techniques, asymmetry 7.3 times more frequent, and wound dehiscence approximately twice as frequent. Aspiration-only techniques had the highest rates of under-resection (2.86%), altered sensation (4.58%), and revision or recurrence (2.97%) (Figure. 4).

Figure 4: Comparative complication profile of aspiration, excision, and combined techniques. 

Complications in the 300-Patient Practice Cohort

Bilateral gland excision was the most commonly performed primary procedure for gynaecomastia correction in our practice (143 patients, 47.67%), followed by VASER-assisted liposuction with gland excision (95, 31.67%) and gland excision and liposuction with either radiofrequency-assisted skin tightening or formal skin reduction/mastopexy, including pedicled NAC repositioning or free NAC grafting when required (42, 14.00%). Gland excision formed part of 291 procedures (97.00%; 95% CI 94.38–98.62), and 137 patients (45.67%) underwent combined excision and liposuction (Table 5). 

Table 5: Single-practice cohort: primary gynaecomastia surgery and outcomes (n = 300)

Outcomes counted once per patient, not per breast. No patient had more than one recorded complication. Confidence intervals are exact binomial (Clopper–Pearson) intervals.

Eleven patients had a recorded complication (3.67%; 1.84–6.47): haematoma in 4 (1.33%), infection or cellulitis in 3 (1.00%), wound dehiscence in 3 (1.00%), and seroma in 1 (0.33%); no ischaemic change of the nipple–areola complex was recorded. Three haematomas were managed by in-office aspiration and one required operative evacuation. No patient had more than one recorded complication, and one patient (0.33%) required an unplanned return to theatre. Because complications could not be linked to individual components of combined procedures, no valid technique-specific complication rate could be calculated for gland excision alone, gland excision with liposuction, energy-assisted skin tightening (including helium plasma–radiofrequency devices such as Renuvion), or skin-envelope management with or without NAC repositioning or grafting.

Benchmarking Practice-Level Complications. The cohort’s 3.67% complication rate was lower than 16 of the 17 published technique-specific strata. The only lower estimate was a zero-event stratum of five transaxillary cases, and the lowest non-zero rate was 5.07% after endoscopically assisted gland excision. The cohort was significantly below every pooled reference estimate: the risk ratio was 0.25 against all techniques, 0.28 against excision, 0.31 against combined excision and liposuction, and 0.11 against the Grade IIb–III sub-group (all p < 0.001 after Holm–Bonferroni correction; Figure. 5, Table 6). 

Table 6: Practice cohort benchmarked against every meta-analytic reference estimate

All comparisons remained significant after Holm–Bonferroni correction. Although the cohort is below every point estimate, it falls within the range expected for an individual centre in this heterogeneous literature, therefore superiority cannot be claimed. Standardised complication ratio for technique mix was 0.30 (95% CI 0.15–0.54). Our practice audit and published literature differ in case mix, gynaecomastia grade, definition of outcome, study design and methodology, duration of follow-up, and how complications were identified and recorded. The apparent advantage observed in the 300-patient cohort may therefore reflect these methodological and case-mix differences rather than a true difference in surgical outcome.

Figure 5: The practice cohort complication rate of 3.67% compared with each meta-analytic reference estimate. It was significantly below all seven pooled point estimates (all p < 0.001; risk ratios 0.11–0.36) but remained within the 95% prediction interval of 0.11–44.88%, so the data do not establish superiority.

Indirect standardisation was applied because the practice cohort was weighted  towards excision. Using raw arm-level rates, 63.30 complications were expected among 291 classifiable patients compared 

with 11 observed, giving a standardised complication ratio of 0.17. Using the more conservative random-effects estimates, 36.68 were expected, yielding a ratio of 0.30 (95% CI 0.15–0.54; exact p < 0.001; Figure. 6a).

Figure 6: Indirect standardisation and detectability analysis. (a) Technique-standardised benchmarking of the practice cohort: 11 complications were observed versus 63.3 expected from raw arm-level rates and 36.7 from random-effects estimates, giving standardised complication ratios of 0.17 and 0.30. (b) Detectability analysis: minimum detectable risk ratios for planned technique comparisons were 3.71–5.21, exceeding the pooled excision-versus-combined risk ratio of 2.61, indicating limited power to test technique-level effects. 

Despite these favourable point comparisons, the cohort estimate lay within the 95% prediction interval of 0.11%–44.88% and was therefore compatible with the range of outcomes expected across comparable centres. The result may reasonably be regarded as favourable within the context of a practice audit, but it does not establish statistical superiority over the published evidence base. In particular, the low observed complication rate cannot be attributed to surgeon skill, the techniques employed, or other characteristics of the practice, because it remains consistent with ordinary between-centre variation in a highly heterogeneous literature. A single-practice series cannot, by design, establish the superiority of a technique or practice, and no such claim is made here.

Revisions and Subsequent Procedures in the Practice Cohort

The primary revision endpoint—any further procedure within 365 days—was met by 36 of 300 patients (12.00%; 8.55–16.22). Under the secondary, surgeon-assigned sub-classification, 32 of these were recorded as elective refinements (10.67%; 7.41–14.72), 3 as residual-tissue re-excisions (1.00%; 0.21–2.89), and 1 as an unplanned complication-related return to theatre (0.33%). No true glandular recurrence was documented (exact upper 95% limit 1.22%). The composite clinically indicated reoperation rate — residual-tissue re-excision, complication-related return to theatre, or true recurrence — was therefore 4 of 300 (1.33%; 0.36–3.38), and elective refinements accounted for 88.9% of patients who returned for further treatment (Fisher exact p < 0.001). The same records consequently yield revision estimates of 12.00%, 10.67%, 1.33%, 1.00%, 0.33%, or 0.00%, depending on how revision is defined (Figure. 7).

Figure 7: 300 practice patients under six definitions of revision, producing rates from 12.00% to 0.00%. Elective refinements accounted for 88.9% of patients returning for further treatment (Fisher exact p < 0.001); no true glandular recurrence was documented within 365 days. The variation in revision rates reflects differences in endpoint definition rather than differences in the underlying clinical events.

Robustness and Limitations of the Cohort Findings

The practice complication rate remained significantly lower than benchmarks exceeding 6.45%; the corresponding thresholds were 3.43% for haematoma and 1.25% for recurrence. The comparison of complication rates had a fragility index of 14 - an additional 14 unrecorded events would have been required for the comparison with the 11.76% combined-technique benchmark to lose statistical significance. Even doubling the recorded number of complications to 22 of 300 yielded a rate of 7.33% (95% CI, 4.65–10.89%; p = 0.015). By contrast, the clinically indicated reoperation estimate had a fragility index of 1, the four observed cases already lying close to the upper limit of the published range (1.4%; Fig. 8). Thus, whereas the complication comparison appears robust, the 1.33% reoperation estimate is sensitive to minimal changes in event detection or classification—for example, an unrecorded reoperation performed elsewhere or reclassification of one refinement as clinically indicated. It should therefore be interpreted cautiously and not regarded as a practice benchmark.

Figure 8: Robustness of the cohort benchmarks. The complication comparison had a fragility index of 14 and remained statistically significant even after doubling the recorded event count (7.33%, p = 0.015). By contrast, the clinically indicated reoperation estimate was highly fragile (fragility index 1): a single additional externally managed or reclassified case would move the estimate beyond the published range. Single-centre complication and reoperation estimates may therefore be biased in opposite directions.

Procedures and complications were reported independently, preventing individual adverse events from being linked to specific techniques or adjuncts. The cohort therefore could not directly compare glandular excision alone with excision plus liposuction, periareolar mastopexy, formal skin reduction, or other adjunctive procedures. Moreover, the low event rate limited statistical power even if patient-level linkage had been available. The minimum detectable risk ratios were 3.71 for excision alone versus excision with liposuction, 4.68 for comparison with the 

radiofrequency stratum, and 5.21 between the two liposuction-containing strata, all exceeding the pooled arm-level risk ratio of 2.61. Detecting an effect of that magnitude between the two liposuction-containing strata would require approximately 278 patients per group (Figure. 6b).

Patient-Reported Outcomes

No contributing study reported validated patient-reported outcomes stratified by Simon grade, and none was collected in the practice cohort (Figure. 9).

Figure 9: Reported outcomes by Simon grade. Complication data are available indirectly through technique, whereas direct grade-stratified complication data and validated patient-reported outcomes by grade are absent.

Existing reviews identify widespread use of non-validated and inconsistent satisfaction measures [19, 20]. The BODY-Q Chest Module is validated for this population [21]. Published studies have reported favourable postoperative chest and psychological scores [22], an association between revision surgery and poorer chest satisfaction [23], and greater satisfaction after endoscopic transaxillary access than after conventional approaches [24]. These findings suggest a plausible relationship between grade, revision burden, and patient experience, but the available evidence does not permit that relationship to be quantified.

By integrating a random-effects meta-analysis with an audit of 300 consecutive patients treated in a single practice, this study examines associations among Simon grade, operative complexity, adjunctive procedures, complications and reoperations in gynaecomastia surgery. The pooled complication rate was 14.44%, but extreme heterogeneity (I² = 96.9%) precluded its use as a universal benchmark.

Stratified analyses identified the greatest morbidity in the indirectly assigned Grade IIb–III group, which encompassed skin-resecting procedures. Intermediate combined procedures had lower complication rates than both aspiration-only approaches and the most extensive operations. At the aggregate treatment-group level, excision alone was associated with a higher complication rate than excision with adjunctive liposuction, although the complication profile shifted from wound- and scar-related morbidity towards contour irregularity and asymmetry. The practice cohort had a favourable technique-standardised complication ratio but remained within the meta-analytic prediction interval. Thirty-six of the 300 patients (12.00%) underwent a further procedure within 365 days, and this inclusive figure is the appropriate one for comparison with other series. Of these 36, the operating surgeons classified 32 as elective refinements rather than clinically indicated reoperations. Depending on the endpoint applied, the same records yield 12.00% for any further procedure, 1.33% for clinically indicated reoperation, 0.33% for unplanned return to theatre and 0% for true glandular recurrence. This variation illustrates the extent to which revision rates depend on endpoint definition. The 95% prediction interval is essential when interpreting practice-level complication rates. Although the cohort's 3.67% rate was below every pooled point estimate and its technique-standardised complication ratio was 0.30, it remained within the 0.11–44.88% range expected for an individual centre. A confidence interval quantifies uncertainty around the pooled mean. It does not describe the range of results plausibly expected in another practice. The prediction interval is therefore the more relevant comparator, although its extraordinary width also highlights the limitations of the evidence base. The meta-analytic synthesis relied on aggregated technique-specific groups from a single systematic review, meaning that within-group heterogeneity, primary-study risk of bias and study-level publication bias could not be assessed reliably. The prediction interval consequently reflects both genuine clinical variation and inconsistencies in outcome definitions, follow-up periods and reporting thresholds. The cohort is therefore consistent with the favourable end of the range reported in current practice, but it does not establish superiority. Indirect analyses suggested that gynaecomastia grade and operative extent were associated with both the frequency and type of complication, although the relationship did not follow a simple linear increase with complexity. The Grade IIb–III group, which encompassed formal skin-excisional techniques, had a pooled complication rate of 32.64%—more than twice the rate in the Grade I–IIa group and more than three times that in the Grade IIa–IIb group. By contrast, minimal-access excision and open gland excision combined with liposuction occupied the lower portion of the observed U-shaped pattern. Greater morbidity at the upper end is clinically plausible because skin reduction may involve additional undermining, dead space, closure tension, longer scars and potential compromise of nipple–areola complex (NAC) perfusion or sensation. Nevertheless, the grade-based analysis was indirect. Simon grade was inferred from the technique conventionally used rather than recorded at patient level, and the high-grade estimate was derived from a single large open-excision group. The findings therefore support the broader conclusion that procedures requiring skin reduction carry greater morbidity, but they do not quantify the independent risk associated with gland excision and periareolar mastopexy, any specific skin-resection pattern, pedicled NAC repositioning or free NAC grafting. Adjunctive procedures should be interpreted according to the problem they are intended to address rather than regarded as uniformly risk-increasing additions. In the aggregate treatment-group data, excision alone was associated with 2.61 times the overall complication rate of excision combined with adjunctive liposuction. It was also associated with higher rates of seroma, haematoma, abnormal scarring, over-resection and NAC compromise. Liposuction-containing approaches, however, were associated with more contour irregularity and asymmetry, demonstrating that adjunctive treatment changes the complication profile rather than increasing the overall risk. At the lower end of the treatment spectrum, aspiration-only techniques had the highest rates of under-resection and revision or recurrence, consistent with the limited ability of suction alone to remove a fibroglandular disc [3]. These findings support the use of adjunctive liposuction as a potentially tissue-sparing contouring adjunct, while recognising that this non-randomised aggregate comparison cannot establish causation. The clearest lessons from the 300-patient cohort concern the classification of subsequent procedures. Of the 36 patients who underwent a further procedure within 365 days, 32 were recorded as elective refinements, three underwent re-excision for residual tissue and one returned to theatre because of a complication. No true glandular recurrence was documented. A broad “revision” label can therefore conflate patient-requested optimisation, incomplete initial excision, complication-related reoperation and breast tissue regrowth—events with fundamentally different causes, clinical implications and potential preventability. However, simply adopting a narrower definition is not sufficient. Because the sub-classification was assigned by the operating surgeon without independent review, and because the elective-refinement category requires the surgeon to judge as satisfactory a result that the patient has sought further surgery to change, a narrow reoperation rate derived in this way is itself a definitional artefact and is systematically liable to understate clinically meaningful dissatisfaction. Nothing in the present data can establish that any individual refinement was genuinely elective. The defensible position is to report the inclusive rate as the headline figure, to report the sub-classification alongside it with its criteria stated, and to have the categories assigned prospectively by an assessor independent of the operating surgeon. Although the limitations of unstratified comparisons are well recognised, this analysis quantitatively demonstrates, within the same dataset, the magnitude of endpoint dependence: revision rates ranged from 12.00% to 0.00% solely according to the definition applied. Within this framework, residual tissue is considered an endpoint of index-procedure completeness, whereas recurrence requires new glandular growth following an initially satisfactory result. The absence of documented recurrence should be regarded as an early observation rather than evidence of long-term durability. Follow-up was limited to 365 days, which is insufficient to assess recurrence in a condition in which regrowth is typically described over several years; treatment received elsewhere may have been missed; and late recurrence after gynaecomastia surgery has been reported [25]. A 0% recurrence rate at one year therefore reflects the follow-up period rather than the durability of the operation, and the exact upper 95% confidence limit of 1.22% should be interpreted accordingly. Similarly, the clinically indicated reoperation rate of 1.33% was based on only four events and was fragile to a single additional case, whereas the overall complication comparison was substantially more robust. The distinction between residual tissue and true recurrence is based primarily on clinical history rather than a validated measurement threshold. Residual tissue is palpable during the early postoperative period and is characteristically described by the patient as having persisted since surgery. True recurrence requires an interval of satisfactory chest contour followed by documented enlargement of the retroareolar disc. Its appearance should prompt investigation for potential precipitating factors, including drug exposure or endocrine disease [26]. Findings reported as recurrence have frequently been attributed to incomplete removal or postoperative scar formation rather than genuine regrowth, and the incidence of recurrence remains uncertain [27]. In practice this distinction is frequently not clear-cut. It depends on the patient’s recollection of whether the chest was ever satisfactory and on an unblinded clinical assessment by the operating surgeon, neither of which is verifiable from a retrospective record, and no imaging or histological standard was applied in the present cohort. When the postoperative interval is short, or the patient history is uncertain, residual tissue and early recurrence may be difficult to distinguish both conceptually and clinically. The observed three-to-zero split between residual tissue and recurrence should therefore be regarded as provisional and as a further illustration of the endpoint dependence that underpins the central argument of this article.

No quantitative definition of adequate glandular resection has been established. The literature provides neither a residual-mass threshold expressed in grams nor a validated imaging criterion. Instead, adequacy is often described in terms of retained retroareolar thickness, an endpoint that remains contested. Conventional teaching favours subtotal excision to reduce the risk of contour depression, with one series recommending preservation of a 3–5-mm retroareolar disc [28]. By contrast, a single-surgeon series of 567 patients advocated complete glandular removal with layered closure of the subareolar dead space and reported no revisions for recurrence, contour depression or persistent nipple prominence [27]. That series warrants the same critical appraisal that has been applied to the present cohort. It is a retrospective, uncontrolled single-surgeon series subject to the same selection, data collection and reporting biases. Outcomes were assessed by the operating surgeon without independent review or validated patient-reported measures, and the reported absence of revisions is itself a surgeon-assessed endpoint of the kind this article argues should be interpreted cautiously. The proposition that complete glandular removal is both achievable and necessary should therefore be regarded as a clearly described technical approach supported by low-level evidence rather than as established evidence. It is cited here as one side of an unresolved technical debate, not as its resolution. These approaches represent fundamentally different operative endpoints. Tissue intentionally preserved under the first approach may be regarded as residual tissue under the second. Neither necessarily implies histological completeness, because microscopic hormonally responsive ductal tissue may remain after subcutaneous mastectomy. “Complete excision” should therefore be understood as a clinical rather than pathological endpoint [27]. In the absence of a validated alternative, this study adopts a pragmatic operational definition rather than implying histological clearance. For the present study, complete excision was defined as the absence of a palpable retroareolar disc at routine follow-up, with the areolar platform lying flush with the surrounding chest wall on inspection and palpation. This definition was adopted for internal consistency but is unvalidated, assessor-dependent, and unblinded, and it is not equivalent to histological clearance. It therefore permits interpretation of residual-tissue rates within this series but limits comparison with studies using retained-thickness or imaging-based criteria. Until a consistent, preferably measurement-based definition is adopted, comparisons of residual-tissue rates across series will remain unreliable.

Unlike residual retroareolar tissue, for which no agreed quantitative threshold exists, nipple–areola complex dimensions can be compared with published reference values. Normative male NAC dimensions were established prospectively in 158 men aged 18–90 years, with mean areolar and nipple diameters of 26.6 mm and 6.9 mm, respectively. Inter-nipple distance, sternal-notch–to-nipple distance and their ratios varied with body mass index but not ethnicity [29]. Areolar enlargement and persistent nipple prominence are common presenting concerns in gynaecomastia and important sources of dissatisfaction after otherwise successful surgery [27]. Reporting preoperative and postoperative areolar dimensions would provide an objective measure of aesthetic change and allow what is described as persistent “puffy nipple” deformity prompting many refinement requests to be documented quantitatively rather than subjectively. These normative values were not, however, applied prospectively in the present practice: areolar and nipple dimensions were not measured before or after surgery in this cohort, and no anthropometric criterion was used either to plan the operation or to adjudicate the refinement requests reported here. Their adoption is therefore proposed as a recommendation for future practice and not offered as a description of the present series. This distinction also explains why “under-resection” may be an unsatisfactory label for the most common indication for further glandular surgery. The term implies a technical shortfall. When a surgeon has deliberately preserved a retroareolar disc in accordance with accepted teaching and the patient subsequently requests its removal, however, the event may represent a mismatch between operative intent and patient expectation rather than an error of execution. Patients seeking gynaecomastia correction may explicitly request complete excision, an expectation reinforced by information shared within patient communities. This observation rests on a single surgical series [27] and on the authors’ own secondary-referral experience. No systematic study has established the role of preoperative expectations in gynaecomastia surgery. It should therefore be regarded as a hypothesis requiring formal investigation with validated expectation measures rather than as established evidence. A technically satisfactory result may consequently still be considered inadequate by the patient. This effect may be amplified in secondary practice, where patients presenting after surgery elsewhere are disproportionately those for whom retained tissue proved unacceptable. Recording the indication for every further procedure—elective refinement, residual tissue following incomplete excision, complication-related reoperation, or documented recurrence—is therefore more than a matter of terminology. It is essential for distinguishing technical failure from mismatches in consent or expectation setting and for determining whether a reported revision rate reflects surgical performance or differences in patient acceptance of retained retroareolar tissue. Recent evidence further highlights the importance of documenting the specific components of each procedure. Studies of circumareolar skin excision suggest that periareolar approaches can address skin redundancy and areolar enlargement while limiting visible scarring, although outcomes remain dependent on operative technique and closure tension [30]. In a 2025 retrospective comparison of 118 patients with Grade IIb–III gynaecomastia, the addition of radiofrequency energy to power-assisted liposuction produced greater objective improvement in chest contour and NAC position and reduction in nipple–areola complex (NAC) area than liposuction alone; however, the non-randomised design did not establish superiority in complication or reoperation rates [31]. A separate Grade IIb series combining circumareolar mastectomy, an interlocking periareolar suture, inframammary-fold release, and waterjet-assisted liposuction reported a complication rate of 7.7% together with favourable BODY-Q outcomes [32]. In patients with marked skin redundancy and ptosis, particularly severe or post-weight-loss cases, high-grade series continue to employ formal skin-reduction techniques and, in selected patients, free NAC grafting [33]. Collectively, these findings support a tailored, multimodal approach while underscoring why broad procedural labels such as “excision”, “mastopexy”, and “skin tightening” should not be treated as equivalent in comparative outcome studies. The audit also identified important limitations in practice-level data collection. Its retrospective, single-practice design lacked independent outcome review and patient-level covariates including body mass index, smoking status, Simon grade, skin quality, ptosis, and comorbidity, precluding risk adjustment beyond technique mix. The favourable standardised complication ratio may therefore partly reflect an unmeasured lower-risk case mix. Outcomes recorded within the treating practice were also susceptible to observer and reporting bias, while complications or reoperations managed elsewhere may have been missed. In addition, individual complications could not be reliably linked to specific operative components, preventing direct technique-level comparisons. Conversely, regular outpatient follow-up may have increased detection of minor wound-related events, making under-detection alone an unlikely explanation for the low complication rate. The observed results may therefore reflect genuine practice performance, favourable case mix, or both. This uncertainty should inform interpretation of the 1.00% infection or cellulitis rate, and comparisons that lost significance after multiple-testing adjustment should not be overinterpreted. With only 11 recorded complications, the cohort had very limited statistical power to detect clinically meaningful differences among operative techniques, The estimated minimum detectable risk ratios (3.71–5.21) exceeded the pooled excision-versus-combined risk ratio of 2.61. Under the assumptions of the power calculations, the cohort would therefore not have achieved the target power to detect an association of this magnitude, even had patient-level linkage been available. These limitations preclude reliable conclusions about comparative safety. The priority for future research is prospective multicentre data collection linking gynaecomastia grade, patient risk factors, operative complexity and individual procedure components to the type, timing, severity, management and outcome of subsequent complications, while distinguishing clinically indicated reoperations from elective aesthetic refinements. Such work would require substantial resources, sustained coordination, agreed grading and event definitions, appropriate consent and data governance, independent outcome adjudication, and follow-up extending beyond one year to assess late recurrence. A staged approach, beginning with consensus definitions and a feasibility study involving a small number of committed centres, may be more achievable than immediately establishing a large multicentre registry.

The absence of validated patient-reported outcomes is a major limitation: complications and reoperations capture harms but inadequately describe the intended benefits in appearance, confidence and psychosocial wellbeing. This also limits interpretation of elective refinements, for which surgeon classification cannot substitute for direct patient assessment. Future studies should prospectively collect the BODY-Q Chest Module and report results by gynaecomastia grade and operative components. The meta-analysis and practice audit support eight reporting domains (Table 7): Simon grade distribution; specific operative components; explicit complication thresholds; separate reporting of elective refinement, residual tissue, true recurrence and unplanned reoperation, with clear distinguishing criteria; a specified follow-up window, including procedures managed outside the treating centre; patient-level denominators; validated patient-reported outcomes; and benchmarking against prediction intervals alongside pooled estimates.

Table 7: Minimum reporting set for future gynaecomastia surgery outcome studies

The proposed framework supplements PRISMA, STROBE and Clavien–Dindo with gynaecomastia-specific reporting items to improve clinical comparability. It addresses four barriers identified in this analysis: missing patient-level grade; aggregation of distinct operative components under single technique labels; inconsistent definitions of revision; and benchmarking against pooled estimates without considering prediction intervals. Derived solely from the present analysis, the framework has not undergone formal consensus development. A Delphi process involving surgeons, biostatisticians and patients, followed by testing against existing datasets, would be needed before recommending global adoption.

Strengths include the large pooled dataset (7,294 patients across 17 technique strata from 94 primary studies), prediction intervals for practice-level benchmarking, and prespecified sensitivity, technique-mix standardisation, detectability and fragility analyses. Benchmarking did not establish superiority. The cohort was consecutive and unselected, duplicate records were reconciled, and a source tabulation discrepancy was disclosed. Reporting six revision-related outcome rates from the same dataset demonstrates the influence of endpoint definition (Fig. 7).

Limitations include reliance on one previous systematic review without independent verification of its search, extraction, classification or appraisal; heterogeneity estimates representing a lower bound; and no additional risk-of-bias or certainty-of-evidence assessment. Grade and operative complexity were inferred from technique, making grade-stratified findings hypothesis-generating. The retrospective, single-practice cohort lacked independent outcome review, patient-level covariates and validated patient-reported outcomes. Revision classification relied on unblinded surgeon judgement, follow-up was limited to 365 days, and the clinically indicated reoperation estimate rested on four events. Technique comparisons were substantially underpowered. The cohort therefore illustrates how stratification and endpoint definitions alter reported rates, without establishing superiority of the practice or its techniques.

Implications

Clinical counselling should consider the initial gynaecomastia presentation and planned operative components, including the higher complication rates observed in the skin-excision subgroup. The likelihood of further procedures should be discussed separately from complications, distinguishing elective refinement from clinically indicated reoperation. The intended extent of glandular excision, including preservation of retroareolar tissue, should be agreed and documented before surgery. These findings do not establish a preferred technique. Future research should prospectively link recorded Simon grade and individual operative components to outcomes, incorporate the BODY-Q Chest Module, independently review indications for further procedures, and extend follow-up substantially beyond one year. The present stratified estimates remain hypothesis-generating; consensus outcome definitions are an immediate priority for meaningful comparisons. Quality assessment and guideline development should account for case mix, operative complexity and endpoint definitions. Unstratified rates may disadvantage practices treating more complex or referred cases or comprehensively recording elective refinements. Comparative use of revision rates requires prespecified definitions, patient-level denominators and benchmarking against prediction intervals alongside pooled estimates.

Complication rates varied across operative techniques and indirectly assigned gynaecomastia grades, with the highest rate in the skin-excision stratum and no clear linear relationship with operative complexity. Adjunctive liposuction was associated with lower aggregate complication rates than excision alone. However, reliance on aggregate data from one previous review and inference of grade from technique make the grade-specific findings hypothesis-generating.

In the 300-patient cohort, 12.00% underwent a further procedure, while 1.33% underwent surgeon-classified clinically indicated reoperation; no true glandular recurrence was documented within 365 days. Most further procedures were classified as elective refinements. These findings illustrate how endpoint selection affects reported rates, but the retrospective design, limited follow-up, absent patient-reported outcomes and lack of independent review limit interpretation. The cohort’s complication rate remained within the meta-analytic prediction interval and did not establish superiority of the practice or its techniques. Meaningful comparisons require recorded disease grade, clearly specified operative components and consistent definitions distinguishing elective refinement, residual-tissue re-excision, complication-related reoperation and true recurrence. Prospective patient-level reporting, validated patient-reported outcomes and benchmarking against prediction intervals are priorities for improving the evidence base.

Funding: No external funding was received for this work.

Conflict of Interest: The authors declare that they have no conflicts of interest to disclose. 

Statement of Human and Animal Rights All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The cohort component was a retrospective review of routinely collected practice records, pseudonymised before analysis, and did not require institutional review board approval. It involved no change to patient care, no additional intervention and no identifiable data, and under applicable national research-governance criteria such work is classified as service evaluation or clinical audit rather than research requiring research ethics committee review. The authors confirm that this position is consistent with the journal’s policy for retrospective studies of routinely collected data, and documentation can be supplied to the Editors on request. The evidence-synthesis component used published aggregate data only. This article does not contain any studies with animals performed by any of the authors.

Informed Consent For this type of retrospective study, formal informed consent is not required. Written informed consent for publication of any clinical images was obtained from the patients concerned.

Data Availability: The aggregate arm-level data re-analysed in this study are published in full in the source systematic review [7]. The derived analytic dataset, the pooled and stratified estimates, and the analysis code are available from the corresponding author on reasonable request. The practice cohort consists of pseudonymised individual patient records that cannot be shared in full; data underlying all reported cohort results are available from the authors on reasonable request, subject to applicable data-protection requirements.

Protocol and Registration No protocol was prospectively registered for this work, and the review was not registered with PROSPERO. The meta-analytic component re-analyses an existing published aggregate dataset rather than undertaking a new primary literature search, and the cohort component was a retrospective service evaluation; neither meets the criteria for prospective registration. The analytical plan for both components—including the outcome definitions, the technique-to-grade mapping, the prespecified sensitivity analyses and the standardisation approach—was fixed before analysis and is reported in full in the Materials and Methods.

Acknowledgements: None.