A Statewide Approach to Reducing Re-excision Rates for Women with Breast Conserving Surgery

Abstract

Studies have demonstrated wide variation in re-excision rates for women with breast cancer undergoing breast conserving surgery, identifying it as a high-value improvement target. We tested the effectiveness of benchmarked performance reports using existing discharge data paired with a statewide collaborative intervention on breast re-excision rates. Re-excision significantly decreased for intervention but not control group hospitals (OR=0.69, 95%CI=0.52–0.91). Benchmarked performance reports and collaborative quality improvement can decrease re-excisions, increase quality, and decrease costs, demonstrating the effective use of administrative data as a platform for state-wide quality collaboratives and suggesting a new paradigm that promotes participation across practice settings.

Objective:

Test the effectiveness of benchmarked performance reports based on existing discharge data paired with a statewide intervention to implement evidence-based strategies on breast re-excision rates.

Summary Background:

Breast-conserving surgery (BCS) is a common breast cancer surgery performed in a range of hospital settings. Studies have demonstrated variations in post-BCS re-excision rates, identifying it as a high-value improvement target.

Methods:

Wisconsin Hospital Association discharge data (2017–2019) were used to compare 60-day re-excision rates following BCS for breast cancer. The analysis estimated the difference in the average change pre-to post-intervention between Surgical Collaborative of Wisconsin (SCW) and non-participating hospitals using a logistic mixed-effects model with repeated measures, adjusting for age, payer, and hospital volume, including hospitals as random effects. The intervention included five collaborative meetings in 2018–2019 where surgeon champions shared guideline updates, best practices/challenges, and facilitated action planning. Confidential benchmarked performance reports were provided.

Results:

In 2017, there were 3,692 breast procedures in SCW and 1,279 in non-participating hospitals; hospital-level re-excision rates ranged from 5% to >50%. There was no statistically significant baseline difference in re-excision rates between SCW and non-participating hospitals (16.1% vs. 17.1%, p=0.47). Re-excision significantly decreased for SCW but not for non-participating hospitals (OR=0.69, 95%CI=0.52–0.91).

Conclusions:

Benchmarked performance reports and collaborative quality improvement can decrease post-BCS re-excisions, increase quality, and decrease costs. Our study demonstrates the effective use of administrative data as a platform for state-wide quality collaboratives. Using existing data requires fewer resources and offers a new paradigm that promotes participation across practice settings

INTRODUCTION

Of the nearly 230,000 women presenting with locoregional breast cancer each year, 79% preserve their breast through a combination of breast-conserving surgery (BCS) and radiation therapy.¹ Breast conservation requires an “adequate margin,” or sufficient distance between the resection edge and nearest cancer cells to ensure lower recurrence rates.^2–4 If adequate margins are not achieved, as occurs for 1 in 5 women in the United States, re-excision is required, resulting in adjuvant therapy delays and the added physical, psychological, and economic burdens of an additional operation.^5,6 A 2015 New England Journal of Medicine editorial describes re-excision as the “other breast cancer epidemic,” underscoring the urgent need for interventions to decrease these avoidable repeat operations.⁷

Numerous evidence-based practices exist to decrease the variation in re-excision rates observed nationally (0 to 70%^{5, 8, 9}), yet implementation has been suboptimal.^10–20 In 2014, the Society for Surgical Oncology (SSO) and American Society for Radiation Oncology (ASTRO) jointly released margin guidelines to facilitate decision-making for second procedures, defining a negative margin for invasive breast cancer as “no tumor on ink” and for ductal carcinoma, 2 mm.²¹ In 2015 in order to ensure that surgeons are aware of evidence-based practices to lower re-excision rates, the American Society of Breast Surgeons (ASBrS) convened the CALLER (Collaborative Attempt to Lower Lumpectomy Re-operative Rates) consensus conference to publish a succinct compilation of the relevant published guidelines and best practices.²² Guideline implementation resulted in a 3.7% absolute reduction in re-excision rates in a cohort of engaged surgeons with access to performance metrics through the ASBrS.²³ Further, a meta-analysis of 7 institutional, registry, and national studies demonstrated that while the 2014 guideline was associated with an overall reduction in re-excisions, reductions were not evident in all studies (5 of the 7), there was significant variation in the magnitude of rate decline across studies (range: 1 to 27.5%), and re-excision rates remained above the target rate of 10% (range: 9–18%),²⁴ indicating important opportunities for further intervention.

Statewide surgical quality collaboratives have demonstrated improved quality and decreased cost,²⁵ but require significant resources limiting generalizability. The Surgical Collaborative of Wisconsin (SCW) is a consortium that, as of 2022, includes 85 hospitals and 224 surgeons and quality leaders. SCW’s strategy includes active engagement of surgeons and hospitals and uses existing administrative data to obviate the need for resource-intensive primary data collection by participants. The goal of the project was to ensure that surgeons had 1) knowledge of SSO-ASTRO and CALLER evidence-based practices; 2) confidential benchmarked performance data necessary to recognize the need for improvement; and 3) the support of other surgeons to facilitate clinical practice. In 2017, surgeon-level re-excision rates in Wisconsin ranged from 5 to 50%, demonstrating the need for these interventions. The purpose of the current investigation was to assess the effect of statewide implementation of a breast surgical quality improvement initiative on re-excision rates in Wisconsin 4 years after publication of the guidelines.

METHODS

Intervention Components:

The SCW breast quality improvement initiative contained three components:

1. Benchmarked performance reporting: Participating surgeons were provided with confidential, de-identified performance reports containing risk- and reliability-adjusted^{26, 27} surgeon and hospital-level 60-day re-excision and mastectomy rates benchmarked against other SCW-participating hospitals and state-wide averages. These reports were distributed in November 2018, July 2019, and November 2019.

2. A collaborative learning environment for sharing experiences with practice change, including challenges and successes with approaches to the implementation of guideline recommendations: SCW surgeon members had the opportunity to meet five times, either in person or virtually, during 2018–2019. Each institution had a surgeon champion and administrative lead who attended meetings and spearheaded the practice change team. Surgeon initiative leaders were fellowship trained. Champions leading active planning sessions had breast as a primary practice. Each SCW meeting included breakout sessions specific to breast cancer, with facilitated discussion of the change process. Breakout sessions began with open-ended questions about attempted change strategies as well as the facilitators and barriers to change. The goal of these discussions was for each institution to develop its own quality improvement strategies tailored to meet individual needs based on prior experience, resources, setting, patient population, and culture. Initiative leaders and surgeon champions were presented with deidentified descriptive information about the variation in re-excision across the state. Surgeon champions led in-person group discussions to facilitate the development of action plans. Action plans walked surgeons through the process of prioritizing CALLER best practice components, setting goals, describing characteristics of their practice that could help or hinder implementation, identifying who and how they needed to engage other members of the care team including defining roles, and describing concrete initial tasks to start them toward their goal(s). Example strategies were provided. Scribes were present at each table, and discussion summaries were created and analyzed for use at subsequent meetings.

3. Education on evidence-based best practice: Participating surgeons attended presentations during the 2018–2019 meetings and were provided with educational material in 2018 that included SSO-ASTRO best practices and guidelines that included pre- and intraoperative approaches for reducing re-excision rates, as reported by the CALLER consensus panel (Figure 1).^21,22

Figure 1:

SSO-ASTRO and CALLER guidelines^{21, 22} provided to SCW-participating surgeons and hospitals targeting the reduction of repeat operations following breast conserving procedures

Data Source

Wisconsin Hospital Association (WHA) hospital discharge data from 2017 to 2019 were used. Pursuant to state statute, all Wisconsin hospital and ambulatory surgery centers submit quarterly data on inpatient and outpatient operations to the WHA. A partnership with the Agency for Healthcare Research and Quality (AHRQ) yielded a validated algorithm now used by AHRQ to identify unique patients across facilities and repeat hospital use.²⁸ This algorithm allows for an unbiased assessment of patient-level healthcare utilization over time across facilities in the state. Patients were included if they were female, at least 18 years old, and had an ICD-10-CM code for non-metastatic breast cancer at the time of their index breast procedure. Patients were excluded if they underwent a breast procedure within the 12-month period preceding their performance year.

Variable definitions:

Outcome Variables:

The primary outcome variable was 60-day re-excision (yes/no), defined as having a breast procedure performed within 60 days of a patient’s index breast conserving surgery (BCS). Index BCS procedures were defined using ICD10 procedure codes (0HBT0ZZ, 0HBU0ZZ, 0HBV0ZZ, 0HBW0ZZ, 0HBX0ZZ, 0HBY0ZZ, 0HBT0ZX, 0HBU0ZX, 0HBV0ZX, 0HBW0ZX, 0HBX0ZX, and 0HBY0ZX) and current procedural terminology (CPT) codes (19120, 19125, 19160, 19301, 19162, and 19302). Second procedures could be BCS or mastectomy procedures (ICD10: 0HTT0ZZ, 0HTU0ZZ, 0HTV0ZZ, 0HTWXZZ, 0HTXXZZ, 0HTY0ZZ; CPT: 19140, 19180, 19182, 19200, 19220, 19240, 19300, 19303, 19304, 19305, 19306, 19307). A secondary outcome variable was constructed as a balance measure (mastectomy as the first procedure) to assess the extent to which the intervention to reduce re-excisions had the unintended consequence of increasing mastectomy procedures.

Explanatory Variables:

Our primary explanatory variable was participation in the SCW breast quality initiative in 2018–2019 (yes/no). Participation in the intervention was defined at the hospital level based on where the index BCS was performed. Hospitals that participated over the time period 2018–2019 were defined as those with an SCW member agreement as of January 2018, with one or more surgeons from the hospital attending a minimum of one SCW meeting and receiving educational material during the calendar year 2018. The level of participation varied among hospitals, ranging from receiving only educational materials to participating in all SCW events and having a representative on the SCW executive committee. Two hospitals began participating in SCW in the last quarter of 2019; these hospitals were included in the comparison group. Other covariates included patient age in years, payer type (Medicaid, medical assistance or uninsured; Medicare or other government insurance; and private insurance), re-excision rates, and breast procedure volume during the baseline period (2017).

Analytic Approach:

Chi-square tests for proportions and t-tests for continuous measures were used to determine intervention and comparison group differences in patient case mix and hospital factors during the baseline time frame (2017). Repeated measures ANOVA was used to assess the statistical significance of the within-group change in patient outcomes between the baseline and post-intervention time points. The adjusted analysis estimated the difference in the average change pre-to post-intervention between SCW participating and non-participating hospitals using a multivariable logistic mixed effects model with repeated measures. The intervention effect in the model was represented by an interaction between SCW participation and the intervention period (pre-or post-intervention). Models adjusted for age, payer, baseline year (2017) re-excision rates, and hospital-level breast procedure volume. Continuous measures (patient age and procedural volume) were standardized to facilitate model interpretation. A vector of hospital random effects were added to the model to account for the natural clustering of patients within hospitals.²⁹ This modeling strategy, also referred to as a difference-in-difference approach,^{30, 31} was selected to facilitate the ready communication and translation of research findings into practice, with results expressed as estimated changes in outcome rates (re-excision or mastectomy) owing to the SCW intervention relative to the non-participating hospitals. All analyses were performed using R statistical software v.4.1.2. This quality improvement project using deidentified discharge data was considered exempt from the Institutional Review Board of the Research Medical Center.

RESULTS

During the 2017 baseline period, 3,692 patients underwent BCS in 63 SCW-participating hospitals and 1,279 patients in 54 non-participating hospitals. During the baseline 2017 period, patients at SCW-participating hospitals were, on average, younger (62.5) than those at non-participating hospitals (63.4) (p=0.03) (Table 1). In addition, patients in SCW-participating hospitals were more likely than patients in non-participating hospitals to have private insurance (47.0% and 43.7%, respectively, p=0.02)).

Table 1:

Descriptive information on SCW-participating and non-participating hospitals (2017)

	SCW-participating Hospitals n=63	Non-Participating Hospitals n=54	p-value
Patient Age
Mean (SD)	62.5 (12.6)	63.4 (12.7)	0.03
Payer			0.02
Medicare/Other Government	47.5%	51.9%
Medical Assistance/Badger Care/Uninsured/Unknown	5.5%	4.4%
Private Insurance	47.0%	43.7%
Hospital Volume
Mean breast surgery volume in 2017 (SD)	58.6 (77.1)	23.7 (38.6)	<0.01
Median breast surgery volume in 2017 (IQR)	30.0 (10–79)	10.5 (5–22)	<0.01
Outcome Measures at Baseline
60-Day Re-excision Rate	16.1	17.1	0.46
Mastectomy Rate	29.8	31.0	0.43

SCW-participating hospitals had significantly higher breast surgery volumes in 2017 on average (mean=58.6; SD=77.1) than non-participating hospitals (mean=23.7; SD=38.6) (p<0.01). During the intervention period, 64.4% of SCW-participating hospitals had surgeons that received educational materials only or participated in 1 activity, 16.9% participated in 2–3, and 18.6% participated in 4 or more activities.

There was no statistically significant difference in 60-day re-excision rates between SCW-participating and non-participating hospitals during the baseline 2017 frame (16.1% and 17.1%, respectively) (p=0.46). However, there was a statistically significant difference in re-excision rates during the post-intervention period (2018–2019), with re-excision rates of 13.9% in SCW-participating and 18.2% in non-participating hospitals (p<0.001). These findings were consistent in the difference-in-differences model that accounted for secular trends between the comparison groups and adjusted for patient age, insurance type, and baseline breast procedural volume (OR=0.69, 95% confidence interval [CI] = 0.52–0.91) (Table 2). In addition, older patients had lower odds of re-excision within 60 days than younger patients (OR=0.92, 95% CI=0.85–1.0).

Table 2:

Logistic mixed models predicting re-excision within 60 days and mastectomy as initial procedure at SCW-participating as compared to non-participating hospitals

	Re-excision			Mastectomy
	Odds Ratio	95% CI	p-value	Odds Ratio	95% CI	p-value
Intervention by Time interaction	0.69	0.52–0.91	0.008	1.28	0.93–1.78	0.13
Covariates
Age (years)*	0.92	0.85–1.0	0.04	1.48	1.40–1.56	<0.01
Insurance
Private	REF	--	--	REF	--	--
Medicare/Other Government	1.04	0.89–1.22	0.61	0.87	0.79–0.97	0.01
Medical Assistance/Badger Care/Uninsured/Unknown	1.18	0.91–1.51	0.23	0.91	0.78–1.07	0.27
Hospital Volume*	0.95	0.78–1.16	0.63	0.82	0.63–1.07	0.14

^*Parameter estimates for age and volume are standardized and can be interpreted as the effect of a 1 standard deviation change.

There was no statistically significant difference in mastectomy rates between SCW-participating and non-participating hospitals during the baseline 2017 time frame (29.8% and 31.0%, respectively, p=0.43), nor was there a statistically significant difference in the degree of pre- to post-intervention change between SCW-participating and non-participating hospitals in mastectomy after adjustment (OR=1.28, 95% CI=0.93–1.78) (Table 2). Older patients and those with Medicare insurance were more likely to undergo mastectomy than younger patients or those with private insurance. There was no statistically significant difference in either re-excison or mastectomy rates over time based on the level of involvement.

DISCUSSION

This study is the first with a statewide implementation approach to reduce unnecessary repeat operations in women with breast cancer following the initial publication of the CALLER guidelines in 2015. Following implementation of a comprehensive initiative that incorporated active surgeon engagement and the provision of performance data, we found a statistically significant reduction in 60-day re-excision rates for women undergoing BCS at SCW-participating hospitals as compared to non-participating hospitals. Importantly, this reduction was not accompanied by a change in mastectomy rates over time, suggesting that improvements in re-excision rates were not accompanied by more extensive initial breast procedures.

The majority of studies assessing changes in re-excision rates over time report that the publication of the SSO-ASTRO guidelines in 2014 resulted in declines in re-excision rates.^{23, 32–35} In a meta-analysis including these studies, Havel (2019) documented a 35% reduction in re-excision rates following publication of the CALLER guidelines, with overall re-excision rates ranging from 11.9–36.5% during the pre-guideline and 9–18% during the post-guideline period.²⁴ In this study of an intervention to further reduce re-excision following guideline publication, we demonstrated that statewide implementation of a surgeon-focused quality initiative was able to further reduce rates from 16.1% in 2017 to 13.9% in 2019, with no significant change in re-excision rates for hospitals not participating in the initiative.

Avoiding re-excision has important implications for patients, providers, and healthcare systems. The need for a second operation creates significant challenges for individual patients, including increased physical, financial (e.g., due to time off work, transportation, childcare needs) and psychological burdens (increased anxiety), in addition to delays in the start of adjuvant treatment.^5,6 Re-excision can be frustrating and demoralizing for individual surgeons and requires precious resources, including operative time, which could be used to treat other patients. As we increasingly recognize the importance of improving value in healthcare, by addressing these issues, decreasing the rates of re-excision will both increase quality (by making care more timely, effective, efficient, and patient-centered) and decrease costs for the overall healthcare system.

This study also has important implications for the development of broad population-based strategies to improve the delivery of high-quality surgical care following the publication of new evidence and/or guidelines. Studies suggest it takes an average of 17 years from the generation of new evidence to clinical practice or policy change.³⁶ We have demonstrated the effectiveness of collaborative learning - paired with benchmarked performance reports using existing administrative data - in disseminating information, sharing best practices, and ultimately improving care. Novel approaches such as this that allow a broad spectrum of hospitals to participate are critical if we are to tackle the major challenges facing our healthcare system.

This study has several important limitations. First, hospital discharge data do not include details of a patient’s cancer diagnosis, including stage at diagnosis, precluding the ability to limit our cohort to patients with stage I-II breast cancer, the patient population for which the initial SSO-ASTRO margin guidelines are primarily based, nor is it possible to assess SCW-participating and non-participating hospital differences in diagnosis or treatment factors. This limitation is unlikely to influence results, as it primarily impacts the decision for the initial operation rather than re-excision. Confidence in findings is further increased by our ability to successfully validate re-excision rates at two participating hospitals (using primary record abstraction) and by the fact that no significant difference was found in mastectomy rates over time. Second, it was not possible to randomize surgeons or hospitals to an intervention arm, leaving open the potential for unobserved confounding factors to influence results. We minimized the potential for this bias by using the methodological approach used (difference-in-differences analysis) and the use of a secondary outcome measure, although it could not be ruled out completely. Finally, this study is limited to the state of Wisconsin. The extent to which these findings are applicable to other areas of the country is unknown. However, it is important to note that all 50 states have access to hospital discharge data, and the infrastructure used for performance reporting and evaluation employed here could be modeled in other states. The study team utilized R open-source statistical software to generate both the performance reports and evaluation structure, both of which could be used by collaboratives in other states.

Despite these limitations, our study demonstrated that benchmarked performance reports and collaborative quality improvement can decrease post-BCS re-excision, increasing quality and decreasing the cost of care for women with breast cancer. Importantly, our study demonstrates the effective use of administrative data as a platform for statewide quality collaboratives. The use of existing data requires fewer resources and offers a new paradigm for statewide quality collaboratives that promotes participation across all types of practice settings.