The surgical treatment of non-metastatic melanoma in a Clinical National Melanoma Registry Study Group (CNMR): a retrospective cohort quality improvement study to reduce the morbidity rates

Reproducible, high-quality surgery is a key point in the management of cancer patients. Quality indicators for surgical treatment of melanoma has been presented with benchmarks but data on morbidity are still limited. This study presents the quality indicators on morbidity after surgical treatment for non-metastatic skin melanoma in an Italian registry. Data were extracted from the Central National Melanoma Registry (CNMR) promoted by the Italian Melanoma Intergroup (IMI). All surgical procedures (WE, SNLB or LFND) for non-metastatic skin melanoma between January 2011 and February 2017 were evaluated for inclusion in the study. Only centers with adequate completeness of information (> 80%) were included in the study. Short-term complications (wound infection, dehiscence, skin graft failure and seroma) were investigated. Wound infection rate was 1.1% (0.4 to 2.7%) in WE, 1.3% (0.7 to 2.5%) in SLNB and 4.1% (2.1 to 8.0%) in LFND. Wound dehiscence rate was 2.0% (0.8 to 5.1%) in WE, 0.9% (0.2 to 3.0%) in SLNB and 2.8% (0.9 to 8.6%) in LFND. Seroma rate was 4.2% (1.5 to 11.1%) in SLNB and 15.1% (4.6 to 39.9%) in LFND. Unreliable information was found on skin graft failure. Our findings contribute to available literature in setting up the recommended standards for melanoma centers, thus improving the quality of surgery offered to patients. A consensus on the core issues around surgical morbidity is needed to provide practical guidance on morbidity prevention and management.


(Continued from previous page)
Conclusions: Our findings contribute to available literature in setting up the recommended standards for melanoma centers, thus improving the quality of surgery offered to patients. A consensus on the core issues around surgical morbidity is needed to provide practical guidance on morbidity prevention and management.
Keywords: Melanoma, Melanoma morbidity, Skin Cancer, Melanoma morbidity, Melanoma surgical treatment, Melanoma quality improvement Background Worldwide, 287,723 new cases of cutaneous melanoma are diagnosed each year, causing the death of 60,712 patients [1]. In non-metastatic melanoma (i.e. without clinically evident regional lymph node or distant metastasis), surgery is the mainstay of treatment [2]. The surgical strategy involves a combination of wide excision (WE), sentinel lymph node biopsy (SLNB) and radical lymph node dissection (LFND), according to cancer staging. WE can be virtually part of the surgical management of all patients with skin melanoma, SNLB is recommended based on the primary tumor thickness and LFND may be required in patients with positive sentinel node [3].
Reproducible, high-quality surgery is a key point in the management and prognosis of cancer patients [4,5]. When a standardized surgical procedure is established, a cyclic audit (including collection, analysis and feedback of both procedural and outcome data) allows for monitoring and improving the quality of surgery. However, cancer surgery is not always standardized, thus limiting the application of such approach [6].
While quality of surgery has been already investigated in other cancers, the use of measures of quality assurance for surgery is less established in melanoma [6]. Available literature offers substantial heterogeneity in surgical procedures among melanoma centers or even among surgeons within the same center [7]. Although adherence to current standards is part of a quality assurance process, the spreading of clinical practice guidelines is not sufficient per se to warrant homogeneity and quality of surgical treatment [8]. There is a growing interest in the implementation of a quality assurance program that includes a quantitative analysis of a set of quality indicators, such as those that can be extracted from electronic medical records [6].
Since 2014, the Italian Melanoma Intergroup (IMI) has been promoting the standardization and quality control of surgical treatment of stage I-III melanoma in Italy [9]. The IMI achieved an expert consensus on surgical treatment of melanoma, defining a list of quality indicators (detection rate, false negative rate, minimum number of excised lymph nodes, postoperative morbidity and local recurrence) with reference to benchmark values, which could be the basis for a standardized quality assurance program in Italy [10]. While quality indicators for detection rate, false negative rate and minimum number of excised lymph nodes were presented for the Central National Melanoma Registry (CNMR), data on morbidity were still pending and data on recurrence are waiting for adequate follow-up [10].
The present study focuses on evaluation of the quality indicators about morbidity after surgical treatment for non-metastatic skin melanoma.

Methods
This is a prospective multicenter study on postoperative morbidity after surgical treatment for non-metastatic skin melanoma. Data were extracted from the Central National Melanoma Registry (CNMR), which is a prospectively maintained national database for melanoma treatment promoted by the IMI in 43 melanoma centers in Italy [9]. The study was conducted according to Helsinki Declaration principles and was approved by the Ethics Committee of the Central National Melanoma Registry (CNMR). All patients gave their consent to have their data collected for scientific purpose.
The following short-term (within 30 days after surgery) complications were investigated: infection, dehiscence, skin necrosis after WE; infection, dehiscence and seroma after SLNB or LFND. Complications were defined using the National Cancer Institute Common Terminology Criteria for Adverse Events v4.0 [11]. Wound infection was defined as a disorder characterized by an infectious process involving the wound. Wound dehiscence was defined as a finding of separation of the approximated margins of a surgical wound. Skin necrosis was defined as a death of cells in living tissue caused by external factors such as infection, trauma, or bacterial agents. Seroma was defined as finding of a collection of serum in the tissues. Patient demographics were also collected.
All surgical procedures (WE, SNLB or LFND) for nonmetastatic skin melanoma between January 2011 and February 2017 were evaluated for inclusion in the study. Only centers with adequate completeness of information (> 80%) were included in the study (11 centers for analysis of morbidity after WE; 14 centers for analysis of morbidity after SNLB; 11 centers for analysis of morbidity after LFND). Within selected centers, patients with pending information on morbidity and those with unfilled form were excluded from the analysis. Patient selection is shown in Supplementary Figs. 1, 2 and 3.
Statistical analysis was conducted using R 3.5 (R Foundation for Statistical Computing, Vienna, Austria) [12]. Categorical data were expressed as number and percentage, and continuous data as median and interquartile range (IQR). The morbidity rate was pooled among centers using a generalized linear mixed-effects model (with the center as random effect). The I 2 value was used to investigate heterogeneity, and meta-regression was used to explore the role of center size and groin site on heterogeneity. A p-value less than 0.05 was considered statistically significant.
Wound infection occurred in 44 patients and dehiscence in 100. The information on skin graft failure was not reliable due to the non-negligible number of unfilled forms (21%).
The proportion of patients developing early wound complications ranged from 0.007 to 0.831 among centers. Summary estimate of proportion of patients developing early wound complications after SLNB was 0.069 (95% CI 0.030 to 0.150) (Fig. 2). Heterogeneity was high (I 2 = 94%); neither center size (p = 0.11) or rate of groin sites among SLNBs (p = 0.84) were not found to have a significant contribution to heterogeneity.
The proportion of patients developing wound infection ranged from 0.005 to 0.104 among centers. Summary estimate of proportion of patients developing wound infection after SLNB was 0.013 (95% CI 0.007 to 0.025) (Fig. 2). Heterogeneity was moderate (I 2 = 60%); neither center size (p = 0.25) or rate of groin sites among SLNBs (p = 0.60) were not found to have a significant contribution to heterogeneity.
The proportion of patients developing dehiscence ranged from 0.005 to 0.101 among centers. Summary estimate of proportion of patients developing dehiscence after SLNB was 0.009 (95% CI 0.002 to 0.030) (Fig. 2). Heterogeneity was high (I 2 = 88%); neither center size (p = 0.98) or rate of groin sites among SLNBs (p = 0.89) were not found to have a significant contribution to heterogeneity.
The proportion of patients developing seroma ranged from 0.004 to 0.792 among centers. Summary estimate of proportion of patients developing seroma after SLNB was 0.042 (95% CI 0.015 to 0.111) (Fig. 2). Heterogeneity was high (I 2 = 95%); neither center size (p = 0.10) or rate of groin sites among SLNBs (p = 0.45) were not found to have a significant contribution to heterogeneity.
The proportion of patients developing early wound complications ranged from 0.024 to 0.969 among centers. Summary estimate of proportion of patients developing early wound complications after LFND was 0.195 (95% CI 0.068 to 0.447) (Fig. 3). Heterogeneity was high (I 2 = 94%); neither center size (p = 0.08) or rate of groin sites among LFNDs (p = 0.41) were not found to have a significant contribution to heterogeneity.
The proportion of patients developing wound infection ranged from 0.006 to 0.58 among centers. Summary estimate of proportion of patients developing wound infection after SLNB was 0.041 (95% CI 0.021 to 0.080) (Fig.  3). Heterogeneity was moderate (I 2 = 54%); neither center size (p = 0.07) or rate of groin sites among SLNBs (p = 0.53) were not found to have a significant contribution to heterogeneity.
The proportion of patients developing dehiscence ranged from 0.006 to 0.244 among centers. Summary estimate of proportion of patients developing dehiscence after SLNB was 0.028 (95% CI 0.009 to 0.086) (Fig. 3). Heterogeneity was high (I 2 = 80%); neither center size (p = 0.38) or rate of groin sites among SLNBs (p = 0.14) were not found to have a significant contribution to heterogeneity.
The proportion of patients developing seroma ranged from 0.021 to 0.969 among centers. Summary estimate of proportion of patients developing seroma after SLNB was 0.151 (95% CI 0.046 to 0.399) (Fig.  3). Heterogeneity was high (I 2 = 95%); neither center size (p = 0.18) or rate of groin sites among SLNBs (p = 0.29) were not found to have a significant contribution to heterogeneity.

Discussion
This study reported the morbidity rates in the largest series of non-metastatic melanoma patients in Italy and one of the largest series worldwide. The present work allowed to evaluate standardization and quality of surgical treatment of cutaneous melanoma within the frame of the Italian Melanoma Intergroup, which is the largest Italian scientific organization dedicated to the management of patients with this disease [10].
The occurrence of wound dehiscence and seroma after SLNB or LFND were in broad agreement with previous studies [13][14][15] (Table 1), thus suggesting limited opportunities for further improvements now. Such rates can be used as morbidity benchmarks in addition to available quality indicators for surgical treatment of melanoma [10].
Interestingly, we found lower occurrence of wound infection after WE or LFND when compared to available literature [13] (Table 1). On one hand, this difference might be partially due to different definitions of wound infection, i.e. presence of fever, only skin redness, suppuration versus cellulitis, isolation of bacteria from wound. On the other hand, some factors might have led to such difference, including: i) recording of infections occurring only during hospital stay or also at the time of outpatient department visits; ii) different frequency/type of antibiotic prophylaxis; iii) different surgical techniques (i.e., use, type and duration of permanence of drainages in the surgical wound).
Postoperative infections account for around one out of four complications associated with hospitalrelated health care procedures, and can impair patient prognosis [16]. Beyond clinical importance, a low infection rate can also have an economic impact, since the management of hospital-related infections requires about 0.8% of gross domestic product (GDP) in Italy [17].
Unfortunately, the information on skin graft failure was not reliable because one out of five forms did not report such complication. This situation likely occurred because skin graft failure required a reconstructive Of note, morbidity rates showed high heterogeneity across melanoma centers, underlying the role of the center itself on this matter. Our data suggested an association between higher morbidity rate and small-volume centers, thus confirming the relationship between patient outcome and hospital surgical volume [18].
The present study contributes to the definition of quality indicators for surgical treatment for nonmetastatic skin melanoma, by adding morbidity indicators that can be used as the basis for a standardized quality assurance program [10]. The importance of this topic relies on the large number of surgical procedures for non-metastatic skin melanoma [2], thus patient management and prognosis can benefit from quality control and standardization of such procedures [4,5]. The strengths of this study included the prospective collection of data in a national registry, the multicenter design and the standardized definitions of the complications [11]. The study has some limitations. First, a considerable number of centers were excluded due to poor completeness of data. Although this approach allowed limiting the impact of low-quality data on the study results, the representativeness of the included centers may be limited. Future developments will aim to achieve adequate completeness of data in the excluded centers and will implement regular audits. Second, the occurrence on skin graft failure after WE could not be evaluated due to the nonnegligible number of unfilled forms. This limitation can be addressed by future improvements regarding information exchange among surgical centers involved in patient care.

Conclusion
Our findings contribute to available literature in setting up the recommended standards for melanoma centers, thus improving the quality of surgery offered to patients. Such quality indicators can be used by other hospitals to direct quality improvement efforts.