April 2020 HSC Section 4 - Plastic and Reconstructive Problems

Annals of Plastic Surgery • Volume 83, Number 2, August 2019

Keloid Excision and Adjuvant Treatments

extracted the information of interest. These included year of publication, study design, number of patients, demographics of patients such as race, treatment regimen, number of recurrences, previous treatments, location of keloid, and duration of follow-up after each treatment.

range of follow-up, type of adjuvant treatment, or date. Reviews, un- controlled studies, animal studies, conference proceedings, and letters to the editor were excluded. Two authors (C.S., A.U.) independently completed article selection based on eligibility criteria. Discrepancies were arbitrated by the senior author (D.S.C.). A flow diagram of the literature search is presented in Figure 1.

Quality Assessment

The first author (C.S.) assessed the methodological quality of the included studies, by using two tools. The Jadad score 26 was used for ex- perimental studies (eg, trials) and the MINORS (Methodological Index for Non-Randomized Studies) criteria 27 were used for observational studies. For studies that the full-text article was not available, we did not perform a quality assessment.

Data Extraction

We categorized the different regimens into 7 broad categories to ensure sufficient sample sizes in each group and subsequently the power to draw conclusions. Categories were designed in advance to minimize introduction of heterogeneity in our analysis and were as fol- lows: (1) no excision group: all patients who received either medical or radiological therapy but no surgical therapy; (2) excision-only group: patients who received only excision and no adjuvant treatment (neither medical nor radiological); (3) excision + 1 adjuvant drug group: all patients who underwent keloid excision and 1 adjuvant medical treatment (eg, intralesional injections of steroids, interferon [IFN], calcium-channel blockers [verapamil], or colchicine); (4) excision + 2 adjuvant drug group: patients who received excision and 2 different classes of drugs as adjuvant treatment; (5) excision + radiation therapy group: patients who underwent excision and radiation therapy with or without medical treatment; (6) excision + pressure: patients who underwent excision and then pressure therapy; and (7) excision + skin grafting: patient who underwent excision and then skin grafting on top of the excision site. Finally, within the excision + 1 adjuvant drug group, we extracted infor- mation on the drug used and formed 2 groups: one that received steroid injections and one that received IFN. Two authors (C.S., A.U.) independently

Statistical Analysis

Paired meta-analysis was initially performed to evaluate those studies that had directly evaluated the same treatment regimens. We conducted the statistical analysis using Review Manager, version 5.3 (The Cochrane Collaboration, Copenhagen). 28 We preset confidence intervals (CIs) at 95% and the level of statistical significance at P < 0.05. We calculated odds ratio (OR) and 95% CI for all outcomes using the Mantel-Haenszel method random-effects model (REM) because of considerable heterogeneity across the studies' populations. 29 We tested for statistical heterogeneity with the Q statistic, generated by χ 2 test and measured the extent of the heterogeneity based on the I 2 measure- ment. 30 We considered heterogeneity values for I 2 values of less than 50% as low, 50% to 75% as medium, and greater than 75% as high. 30 We did not formally assess publication bias and funnel plot asymmetry

FIGURE 1. Flow diagram of literature search.

www.annalsplasticsurgery.com

© 2019 Wolters Kluwer Health, Inc. All rights reserved.

128