Effect of Surgeon Factors on Long-Term Patient-Reported Outcomes After Breast Conserving Therapy in Older Breast Cancer Survivors

Abstract

Background:

The effect of surgeon factors on patient-reported quality of life outcomes after breast conserving therapy (BCT) is unknown and may help patients make informed care decisions.

Methods:

We performed a survey study of women age ≥67 with non-metastatic breast cancer diagnosed in 2009 treated with guideline-concordant BCT to determine the association of surgeon factors with patient-reported outcomes. The treating surgeon was identified using Medicare claims and surgeon factors identified via the AMA Physician Masterfile. The primary outcome was patient-reported cosmetic satisfaction measured by CanSORT Satisfaction with Breast Cosmetic Outcome. Secondary outcomes included BREAST-Q subdomains. All patient, treatment, and surgeon covariables were included in a saturated multivariable linear regression model with backward elimination applied until remaining variables were P<0.1.

Results:

Of 1650 women randomly selected to receive the questionnaire, 489 responded and of these, 289 underwent BCT. Median age at diagnosis was 72 years and time from diagnosis to survey was 6 years. Mean adjusted CanSORT score was higher for patients treated by surgical oncologists than patients treated by non-surgical oncologists (4.01 [95% CI 3.65–4.38] vs. 3.53 [95% CI 3.28–3.77]), P=0.006). Similarly, mean adjusted BREAST-Q Physical Well-Being (91.97 [95% CI 86.13–97.80] vs. 83.04 [95% CI 80.85–85.22], P=0.006) and Adverse Radiation Effects (95.28 [95% CI 91.25–99.31] vs. 88.90 [95% CI 86.23–91.57], P=0.004) scores were better among patients treated by surgical oncologists.

Conclusions:

Specialized surgical oncology training is associated with improved long-term patient-reported outcomes. These findings underscore the value of specialized training and may be useful to patients choosing their care team.

Introduction

Older breast cancer patients face many decisions regarding their treatment, including type of primary surgery, reconstruction (if any), and radiation therapy, including the option to forego radiation therapy in some cases.^1,2 Patients rely on their physicians to discuss the advantages and drawbacks of each treatment option,³ and there is an increasing body of evidence to assist with these discussions.^4–8 Previous studies have shown that the treating surgeon can have a significant impact on the type of therapy a patient receives,^9–11 and that older patients are more likely than younger patients to defer to their physicians for treatment decisions.^12,13

Beyond the influence of treating surgeons on patient treatment decisions, there is evidence to suggest an association between the treating surgeon and patient outcomes. For example, surgeon expertise has been shown to correlate with overall patient satisfaction with their treatment experience.¹⁴ However, little is known about how individual surgeons’ characteristics influence long-term patient-reported outcomes (PROs). Such information may be helpful for patients choosing among multiple potential providers and treatment settings and in targeting quality improvement interventions.

We recently performed a national survey study of older breast cancer patients following definitive treatment to determine the associations between local therapy type and long-term patient-reported QOL outcomes.¹⁵ Within the subgroup of patients treated with breast-conserving local therapy, where optimal cosmetic outcome is both a key goal of therapy and known to vary considerably,¹⁶ we hypothesized that the specialization of the treating surgeon contributes to patient-reported cosmetic satisfaction. To test this hypothesis, we evaluated whether surgeon factors were associated with long-term cosmetic satisfaction and other important PROs in older breast cancer survivors treated with breast-conserving therapy.

Methods

This study was approved by the Institutional Review Board of the University of Texas MD Anderson Cancer Center as well the Centers for Medicare and Medicaid Services.

Cohort Selection

We identified a cohort of potential study participants by applying a validated algorithm to national Medicare claims, which identified women age 67 years of age or older diagnosed with early stage invasive breast cancer in 2009 (N = 42,735).^17,18 Of these, we included only patients who were treated with one of five guideline-concordant local therapy strategies: lumpectomy with whole breast irradiation (Lump+WBI), lumpectomy with brachytherapy (Lump+Brachy, inclusive of both intracavitary balloon-based and interstitial high dose rate procedures), lumpectomy with endocrine therapy alone (Lump alone), mastectomy without radiation (Mast alone), or mastectomy with radiation (Mast+RT). Patients had to be alive as of May 2015, possess continuous Medicare Part A and B coverage from diagnosis through 2011, and be without new diagnosis cancer following breast cancer diagnosis (N = 26,069). Study participants were selected from the remaining patients (eTable 1).

Determination of Local Therapy

We used International Classification of Diseases (ICD-9) and Common Procedural Terminology (CPT) code sets to determine receipt of surgery and/or radiation. “Lump+WBI” was defined as lumpectomy with 15 or more external beam radiation fractions and no brachytherapy. “Lump+Brachy” was defined as a lumpectomy followed by brachytherapy and no external beam radiation treatment. “Lump alone” was defined as a lumpectomy with a dispensed prescription for endocrine therapy. “Mast alone” was defined as a mastectomy and no radiotherapy in the subsequent year. “Mast+RT” was defined as a mastectomy with 15 or more external beam radiation fractions.

Sampling Strategy

Lump+WBI was the most common local treatment (N = 2,731) and therefore served as the referent group for all analyses. We determined that including 250 patients per local therapy group would result in 80% power to detect a difference of 0.31 points on a 5-point Likert scale, which was based on prior published data from the CanSORT Breast Cosmetic Outcome instrument. Estimating a 75% response rate, we sampled 330 patients per local therapy group. We first randomly sampled 330 patients who received Lump+WBI and subsequently used sampling without replacement with probability proportional to size to sample 330 patients who received each of the other four types of local therapies so that the age distribution resembled the Lump+WBI group.

Survey Collection

The Centers for Medicare and Medicaid Services Beneficiary Contact Service mailed an informational flyer including a description of our study and an opt-out, postage-paid response card to 1,650 potential participants in September 2015. We received 397 (24%) opt-out responses. In November 2015, a study invitation letter, $10 gift card, and questionnaire were sent by mail to 1,253 potential participants and followed up with non-responders by mail and phone through January 2016.

Survey Measures

We developed a paper questionnaire to collect baseline patient information and PROs in breast cancer patients. We piloted the questionnaire in 10 breast cancer patients returning for follow up care at our institution. Cognitive interviews were conducted to obtain feedback on the length and clarity of the questionnaire.¹⁹

The survey assessed demographic characteristics including race/ethnicity, education, and income. Other personal and clinical items included current height and weight, weight at diagnosis, bra cup size, smoking history, subsequent local or distant breast cancer recurrence or mastectomy, and second cancer diagnoses.

The primary outcome, defined a priori, was patient-reported cosmetic satisfaction evaluated using the Cancer Surveillance and Outcomes Research Team (CanSORT) Satisfaction with Breast Cosmetic Outcome instrument.⁴ This validated tool consists of six items scored with a 5-point Likert scale. The total score is the arithmetic mean of the six items, and a higher score indicates a higher level of satisfaction. A difference in mean scores of 0.3 or greater was considered clinically relevant, as our prior data from this cohort indicated that 0.3 is the approximate magnitude of the difference in satisfaction between breast-conserving therapy and mastectomy.¹⁵

Secondary outcomes were measured using up to five independent BREAST-Q modules²⁰ including Satisfaction with Breasts, Physical Well-Being, Adverse Effects of Radiation, Psychosocial Well-Being, and Sexual Well-Being. The BREAST-Q is a validated measure of patient satisfaction and QoL after breast cancer surgery. For each measure, the sum score is converted to an equivalent Rasch transformed score that ranges from 0–100. Higher scores indicate better outcomes. Clinically relevant differences in BREAST-Q scores have been reported for the core modules as follows: satisfaction with Breasts, 8 points; Physical Well-Being, 7 points; Psychosocial Well-Being, 10 points; and Sexual Well-Being, 10 points.²¹

Determination of Surgeon Factors

For each patient, we identified the treating surgeon, defined as the surgeon who performed the initial local breast surgery via Medicare claims.¹⁰ The following surgeon factors were obtained via linkage to the American Medical Association (AMA) Physician Masterfile by National Provider Identification number: board certification (yes versus no), location of training (United States versus outside of the United States), type of medical degree (MD versus DO), decade of medical school graduation (before 1980 versus after 1980), surgical specialty (surgical oncology versus general surgery/other), surgeon sex (male versus female), and Medicare breast surgery case volume (≤40 cases versus 41–100 cases versus >100 cases). Surgeon subspecialty is determined via self-report. Options for self-reporting include surgical oncology and general surgery; breast surgery is not an option. Surgical oncologists have completed additional fellowship training after general surgery residency focusing on the advanced surgical management of cancers. Case volume was defined as the number of Medicare fee-for-service lumpectomy and mastectomy cases performed by the surgeon from 2007–2010.

Statistical Analysis

Baseline patient characteristics including age, Charlson Comorbidity Index, geographic region, and receipt of axillary surgery or chemotherapy were obtained from Medicare enrollment data, Medicare claims, and self-report (see Survey Measures). Baseline patient characteristics between different local therapy groups were compared using the chi-square test or Fisher’s exact test when cell sizes were small.

Primary and secondary outcomes were analyzed using multivariable linear regression. We included all patient, surgeon, and treatment variables in a multivariable model and applied backward elimination to identify patient, treatment, and surgeon factors associated with each outcome at P<0.1. Type of local therapy was retained in the models regardless of P value. The effect of type of local therapy on PROs is reported elsewhere.¹⁵ As a sensitivity analysis, we retained all surgeon factors in the final multivariable models regardless of their significance.

In the study population, 258 surgeons treated one patient each and only 13 surgeons treated two patients each, thus we used a traditional single-level modeling approach for all outcomes and a multilevel model of patients nested within surgeons was created only for the primary outcome as an exploratory analysis. We created non-response weights to account for differential sample selection across treatment groups and age categories. Non-response weights were scaled based on the ratios of numbers of responders versus the total samples in each cell. All analyses used these weights to compensate for different probabilities of sample selection and non-response so that results were more representative of the sampled population.⁴ Statistical analyses were performed in SAS v9.3 with significance level of two-sided P<0.05.

Results

Patient Characteristics

A total of 489 women responded (30% response rate). In the current study, we included the 289 responders treated with one of three breast-conserving local therapies, including 108 (37%) Lump+WBI, 103 (36%) Lump+Brachy, and 78 (27%) Lump alone. Compared to non-responders, responders were more likely to be younger, white, have not received Lump alone, have not received chemotherapy, have lower Charlson Comorbidity Index, and live in the South geographic region (P<0.05 for all) (eTable 2). The median age at breast cancer diagnosis was 72 years (range, 67–87) and elapsed time from diagnosis to survey completion was about 6 years for all respondents. Other patient characteristics can be found in Table 1.

Table 1.

Patient demographic and clinical characteristics (n=289)

Characteristic	No. (%)a
Local therapy
Lump+WBI	108 (37.4)
Lump+Brachy	103 (35.6)
Lump alone	78 (27.0)
Age, years
Median (range)	72 (67–87)
67–74	200 (57.5)
≥75	89 (42.5)
Patient-reported race
White	250 (86.7)
Non-white	39 (13.3)
Charlson Comorbidity Index
0	197 (68.5)
1	71 (24.8)
2+	21 (6.6)
Patient-reported baseline BMI
Underweight / Normal	93 (35.7)
Overweight	77 (21.9)
Obese	103 (34.8)
Unknown	16 (7.7)
Patient-reported current bra cup size
A-B	89 (31.0)
C	88 (31.4)
D+	81 (25.8)
Unknown	31 (11.8)
Axillary surgery
Axillary dissection/axillary surgery NOS	94 (36.2)
Sentinel lymph node biopsy	126 (45.7)
No axillary surgery	69 (18.0)
Patient-reported endocrine therapy
No	95 (39.5)
Yes	194 (60.5)
Chemotherapy
No	254 (84.7)
Yes	35 (15.3)
Region
Northeast	50 (22.2)
Midwest	68 (20.8)
South	119 (39.3)
West or unknown	51 (17.7)
Patient-reported education
High school or less	95 (32.9)
Associate degree	97 (33.6)
College or higher	82 (28.4)
Unknown	15 (5.2)
Patient-reported current household income
<$40K	120 (44.5)
$40–60K	45 (12.5)
>$60K	67 (22.8)
Unknown	57 (20.2)
Has smoked 100 cigarettes in lifetime
No	154 (53.8)
Yes	119 (40.0)
Unknown	16 (6.2)

Abbreviations: Lump+WBI = lumpectomy plus whole-breast irradiation; Lump+Brachy = lumpectomy plus brachytherapy; Lump alone = lumpectomy plus endocrine therapy alone; NOS = not otherwise specified.

^aAll percentages are weighted percentages except for local therapy; weights accounted for differential selection and response rates.

Surgeon Factors

Treating surgeon data were available for 284 out of 289 patients (98%), representing 271 unique surgeons (including 13 surgeons who had treated 2 patients each). Most were US-trained (88%), had an MD degree (93%), were male (61%), and did not have specialized surgical oncology training (86%). Surgical oncologists had higher breast surgery case volume compared to non-surgical oncologists from 2007–2010 (61.6% vs. 38.4% with >100 cases, P<0.001). Additional surgeon characteristics are listed in Table 2.

Table 2.

Treating surgeon characteristics (n=271^a)

Characteristic	No. (%)b
Board certification
Yes	244 (87.5)
No	27 (12.5)
Location of training
United States	238 (85.7)
Outside of United States	22 (11.5)
Unknown	11 (2.8)
Type of medical degree
MD	251 (91.2)
DO/Unknownc	20 (8.8)
Decade of medical school graduation
Before 1980	91 (38.1)
1980 and after	169 (59.1)
Unknown	11 (2.8)
Surgical specialty
General surgery/other	233 (86.7)
Surgical oncology	38 (13.3)
Gender
Male	164 (64.2)
Female	96 (32.9)
Unknown	11 (2.8)
Case volumed
≤40	70 (29.8)
41–100	97 (36.3)
>100	104 (33.9)

^aSurgeon data were available for 284 out of 289 patients (98%), and there were 271 unique surgeons.

^bAll percentages are weighted percentages; weights accounted for differential selection and response rates.

^cDue to small cell sizes, “DO” and “Unknown” categories were combined to avoid compromising provider identity in accordance with the Data User’s Agreement with Centers for Medicare and Medicaid Services.

^dNumber of Medicare fee-for-service cases from 2007–2010

Multivariable Analysis of CanSORT Satisfaction with Breast Cosmetic Outcome and BREAST-Q Satisfaction with Breasts

In multivariable analysis adjusting for type of local therapy, CanSORT Satisfaction with Breast Cosmetic Outcome was higher among patients treated by surgical oncologists compared to non-surgical oncologists (effect size 0.48, [95% CI 0.14–0.83]; P=0.006) (Table 3). This finding was stable in an exploratory multilevel model of patients nested within surgeons (effect size 0.49, [95% CI 0.14–0.83]; P=0.006). Sensitivity analysis with all surgeon factors retained in the final model regardless of their significance showed similar results (eTable 4).

Table 3.

Multivariable linear regression models for CanSORT Satisfaction with Breast Cosmetic Outcome and BREAST-Q Satisfaction with Breasts

	CanSORT (n=282)				BREAST-Q Satisfaction with Breasts (n=273)
	Estimate Coefficient	SE	95% CI	P Value	Estimate Coefficient	SE	95% CI	P Value
Intercepta	3.62	0.12	3.40 to 3.85	<0.001	60.69	3.28	54.26 to 67.11	<0.001
Surgical specialty
General surgery/other	Referent
Surgical oncology	0.48	0.18	0.14 to 0.83	0.006
Local therapy
Lump+WBI	Referent				Referent
Lump+Brachy	0.35	0.13	0.10 to 0.61	0.007	7.06	3.25	0.70 to 13.42	0.031
Lump alone	0.56	0.24	0.09 to 1.02	0.02	8.54	4.84	−0.95 to 18.03	0.08
Charlson Comorbidity Index
0	Referent				Referent
1	0.02	0.18	−0.34 to 0.38	0.91	−0.61	4.06	−8.57 to 7.34	0.88
2+	−1.23	0.30	−1.81 to −0.64	<0.001	−19.42	5.52	−30.25 to −8.60	<0.001
Region
South					Referent
Northeast					2.42	5.25	−7.86 to 12.70	0.65
Midwest					8.89	4.44	0.18 to 17.60	0.046
West					−4.73	4.70	−13.95 to 4.48	0.31

Abbreviations: Lump+WBI = lumpectomy plus whole-breast irradiation; Lump+Brachy = lumpectomy plus brachytherapy; Lump alone = lumpectomy plus endocrine therapy alone; SE = standard error; CI = confidence interval.

^aIntercept is the expected mean value of the outcome when all covariables are referent.

All analyses incorporated weights to adjust for differential sample selection and response rates.

Adjusted mean CanSORT cosmetic satisfaction score was 4.01 (95% CI 3.65–4.38) for patients treated by surgical oncologists versus 3.53 (95% CI 3.28–3.77) for patients treated by non-surgical oncologists (P=0.006) (Table 4 and Figure 1). Compared to Lump+WBI, patients undergoing Lump alone (effect size 0.56, [95% CI 0.09–1.02]; P=0.02) and Lump+Brachy (effect size 0.35, [95% CI 0.10–0.61]; P=0.007) reported higher CanSORT cosmetic satisfaction. The only patient characteristic associated with CanSORT cosmetic satisfaction was Charlson Comorbidity Index, with higher comorbidity score associated with worse CanSORT cosmetic satisfaction.

Table 4.

Adjusted mean scores for CanSORT Satisfaction with Breast Cosmetic Outcome, Breast-Q Physical Well-Being and Adverse RT Effects

	CanSORTa (n=282)			Breast-Q Physical Well-Beingb (n=275)			Breast-Q Adverse RT Effectsc (n=201)
	n	Mean (95% CI)	P Value	n	Mean (95% CI)	P Value	n	Mean (95% CI)	P Value
Surgical oncologist	36	4.01 (3.65 to 4.37)	0.006	35	91.97 (86.13 to 97.80)	0.006	26	95.28 (91.25 to 99.31)	0.004
Non-surgical oncologist	246	3.53 (3.28 to 3.77)		240	83.04 (80.85 to 85.22)		175	88.90 (86.23 to 91.57)

^aFinal model for CanSORT was additionally adjusted for local therapy and Charlson Comorbidity Index.

^bFinal model for Breast-Q Physical Well-Being was additionally adjusted for local therapy and geographic region.

^cFinal model for Breast-Q Adverse Radiation Effects was additionally adjusted for local therapy and bra cup size.

All analyses incorporated weights to adjust for differential sample selection and response rates.

Figure 1.

Mean adjusted patient-reported outcome scores by surgical specialization

No significant difference in BREAST-Q Satisfaction with Breasts mean score was detected among patients of surgical oncologists compared to non-surgical oncologists in unadjusted analysis (67.4 vs. 64.2, P=0.42) and no surgeon factors were retained in multivariable analysis. In multivariable analysis, Lump+WBI, higher Charlson Comorbidity Index score, and living in the South geographic region were associated with lower Satisfaction with Breasts (Table 3).

Multivariable Analysis of BREAST-Q Physical, Psychosocial, and Sexual Well-Being and Adverse Radiation Effects

In multivariable analysis for BREAST-Q Physical Well-Being and Adverse Radiation Effects, specialization in surgical oncology was associated with improved outcomes (Table 5). The adjusted mean score for BREAST-Q Physical Well-Being was 91.97 (95% CI 86.13–97.80) for patients treated by surgical oncologists versus 83.04 (95% CI 80.85–85.22) for patients treated by non-surgical oncologists (P=0.006). Likewise, the adjusted mean score for BREAST-Q Adverse Radiation Effects was 95.28 (95% CI 91.25–99.31) for patients treated by surgical oncologists versus 88.90 (95% CI 86.23–91.57) for patients treated by non-surgical oncologists (P=0.004) (Table 4 and Figure 1). Compared to Lump+WBI, Lump+Brachy was associated with better Physical Well-Being and better Adverse Radiation Effects. Living in the South geographic region was associated with worse Physical Well-Being and D+ bra cup size was associated with worse Adverse Radiation Effects (Table 5).

Table 5.

Multivariable linear regression models for BREAST-Q Physical Well-Being and Adverse Radiation Effects

	Physical Well-Being (n=275)				Adverse Radiation Effects (n=201)
	Estimate Coefficient	SE	95% CI	P Value	Estimate Coefficient	SE	95% CI	P Value
Intercepta	77.46	2.34	72.87 to 82.05	<0.001	90.73	1.93	86.94 to 94.52	<0.001
Surgical specialty
General surgery/other	Referent				Referent
Surgical oncology	8.93	3.23	2.61 to 15.26	0.006	6.38	2.17	2.13 to 10.63	0.004
Local therapy
Lump+WBI	Referent				Referent
Lump+Brachy	5.72	2.19	1.42 to 10.03	0.01	4.26	2.12	0.11 to 8.42	0.046
Lump alone	0.28	2.67	−4.96 to 5.51	0.92	.	.	.	.
Patient-reported current bra cup size
A-B					Referent
C					−0.34	2.27	−4.79 to 4.12	0.88
D+					−8.49	3.49	−15.33 to −1.65	0.02
Region
South	Referent
Northeast	8.38	3.06	2.39 to 14.39	0.007
Midwest	2.10	3.59	−4.92 to 9.13	0.56
West	3.81	3.57	−3.17 to 10.80	0.29

Abbreviations: Lump+WBI = lumpectomy plus whole-breast irradiation; Lump+Brachy = lumpectomy plus brachytherapy; Lump alone = lumpectomy plus endocrine therapy alone; SE = standard error; CI = confidence interval

^aIntercept is the expected mean value of the outcome when all covariables are referent.

All analyses incorporated weights to adjust for differential sample selection and response rates.

In multivariable analysis for BREAST-Q Psychosocial Well-Being and Sexual Well-Being, there were no significant differences in scores by surgeon factors. Local therapy type was not predictive of Psychosocial Well-Being or Sexual Well-Being. Patient characteristics including white race, higher Charlson Comorbidity Index, lower income, and absence of axillary surgery correlated with worse Psychosocial Well-Being. D+ bra cup size and living in the South geographic region correlated with worse Sexual Well-Being (eTable 3).

Discussion

In this nationally representative cohort of older breast cancer survivors treated with breast-conserving local therapy strategies, we found that long-term patient-reported cosmetic satisfaction, physical well-being, and adverse radiation effects were significantly better among patients treated by surgical oncologists compared to those treated by non-surgical oncologists. To the best of our knowledge, this is the first study to directly link surgeon specialization to long-term PROs in cancer patients. For the primary outcome of CanSORT Satisfaction with Breast Cosmetic Outcome, the adjusted effect size of surgical oncology specialization was 0.48, which is even greater than the difference in satisfaction between breast-conserving therapy and mastectomy of 0.3 found in prior data from this cohort, demonstrating the clinical relevance of these findings.¹⁵

Specialization in surgical oncology may influence PROs through a variety of mechanisms. First, surgical oncologists are more likely than general surgeons to practice within cancer centers with expertise across the diagnostic and therapeutic continuum, inclusive of breast imaging, breast pathology, medical oncology, and radiation oncology.^22,23 A practice environment inclusive of other specialists focused on breast cancer is likely to synergize with surgeon expertise to promote the best patient outcomes. Aside from the multidisciplinary expertise associated with practice in a cancer center, there may also be better medical facilities and availability of support staff. Cancer centers tend to be located in metropolitan areas with easily accessible medical care, which has been shown to impact health outcomes.²⁴

Second, surgical oncologists may be more likely to employ techniques that could improve cosmetic outcome, such as oncoplastic reconstruction or local tissue transfer with or without an accompanying plastic surgeon.^25,26 Similarly, it is plausible that excision volumes may be lower among surgical oncologists compared to general surgeons. Third, patients of surgical oncologists may be less likely to undergo re-excision, as re-excision rates vary inversely with surgeon expertise^27–29, and re-excision increases the likelihood of adverse cosmetic outcome.^30,31

In addition, there is evidence to suggest a positive surgery case volume-outcome relationship among multiple surgical procedures and conditions.³² Within the realm of breast cancer, surgeons with higher case volume are more likely to adhere to the latest margin re-excision guidelines³³ and treat patients with breast-conserving surgery.³⁴ In our study, although surgical oncology subspecialization was associated with higher breast surgery case volume, it was surgical specialization and not case volume that was consistently associated with improved PROs in multivariable and sensitivity analyses, which suggests that additional specialty training may add value beyond what experience and case volume offer. Conversely, it is conceivable that surgeons completing competitive³⁵ surgical oncology fellowships are selected for in terms of their prior qualifications and general surgery training, which has been shown to be related to surgical outcomes.³⁶

Finally, surgeon-patient communication may vary between general surgeons and surgical oncologists, leading to long-term differences in PROs. For example, patients treated by surgeons specializing in breast disease are more satisfied with their surgeon-patient relationship,¹⁴ and patients who are satisfied with their surgeon-patient relationship are more likely to be satisfied with their cosmetic outcome.^37,38 In a survey study of patients treated with mastectomy and reconstruction, Ho et al satisfaction with pre-operative counseling and with the surgeon-patient relationship predicted higher found that patient satisfaction with cosmetic outcome.³⁸ Similar correlations between physician-patient relationships and PROs have been observed in prostate cancer³⁹ and medical oncology.⁴⁰ These data underscore the importance of patient-centered care, and highlight areas where surgeons – including those without specialty training in surgical oncology – may be able to impact patient outcomes outside of the operating room.

Interestingly, treatment by a surgical oncologist was also associated with improved patient-reported adverse radiation effects. This is consistent with prior studies that have associated poor post-surgery, pre-radiation cosmetic outcome with late breast fibrosis.⁴¹ As described previously, surgical oncologists are more likely than non-surgical oncologists to be working in a multidisciplinary setting with subspecialized radiation oncologists. These patients may therefore also benefit from the expertise of the treating radiation oncologist.

From a patient perspective, there is surprisingly little data to guide patients toward high quality providers. Although national movements are ongoing to promote outcome and patient satisfaction reporting, such data are not yet widely available. By demonstrating the relevance of surgeon specialization to outcomes experienced by cancer survivors after treatment, this study provides key information needed for patients to understand the implications of selecting providers with specialization in surgical oncology. For surgeons-in-training, our findings document how additional training beyond traditional residency can add value to a surgical practice by promoting optimal patient outcomes. For policymakers and professional societies, our study findings document the value of additional subspecialization within general surgery and underscore needs for ongoing medical education for general surgeons whose practice includes breast surgery. While it may be difficult for practicing surgeons who are not surgical oncologists to return for additional training, there are many opportunities for continuing medical education available at the annual meetings of the American College of Surgeons, American Society of Breast Surgeons, and the Society of Surgical Oncology. Continuing education is widely regarded as a critical component of physician practice and has been shown to impact both physician performance and health care outcomes.⁴²

One of the limitations of the current study is that CanSORT Breast Cosmetic Outcome, but not BREAST-Q Satisfaction with Breasts was found to be significantly associated with surgical specialization. While both are validated instruments for cosmetic outcome that share similarities¹⁵, we hypothesize that more granular scoring of CanSORT Breast Cosmetic Outcome (scored on a Likert scale of 1–5, whereas BREAST-Q Satisfaction with Breasts is scored on a scale of 1–4) and overlap of physical well-being traits in CanSORT Breast Cosmetic Outcome may have led to increased ability of the CanSORT score to detect underlying differences in outcome. Finally, BREAST-Q Satisfaction with Breasts is a longer instrument (11 questions versus 6 questions) and fewer patients completed it compared to CanSORT Breast Cosmetic Outcome (282 versus 273), which suggests some respondents may have fatigued in this portion of the questionnaire.

Other study limitations include a modest response rate, which may be attributed to the older age of the study population (median age 72 at diagnosis and 78 at the time of survey) and the length of time (6 years) elapsed since diagnosis and treatment. There were significant differences in baseline characteristics between responders and non-responders, including age at diagnosis (median age 72 for responders, 75 for non-responders) and local therapy (women who received lumpectomy alone were less likely to respond compared to women who received other local treatment types) that we attempted to control for in statistical analysis with non-response weights. There were no significant differences between responders and non-responders regarding surgeon factors and therefore these differences are unlikely to have caused bias in our primary study endpoint associating surgeon factors with PROs. Our analysis did not account for relevant treatment factors that may impact cosmetic outcome such as tumor size, multifocal disease, requirement for re-excision, and radiation dose and fractionation schedule. Surgical subspecialty was provided via self-report in the AMA Physician Masterfile and breast surgery was not an option – however, it is reasonable to assume that fellowship-trained breast surgeons would self-select surgical oncology as the most appropriate specialty given the options.

In summary, older breast cancer survivors treated by surgical oncologists reported considerably better cosmetic satisfaction, physical well-being, and adverse radiation effects approximately 6 years after breast-conserving surgery. The robust association between surgeon specialization and multiple patient-reported outcome domains is highly relevant to patients seeking best care and to quality improvement efforts to optimize long-term patient outcomes.

Synopsis.

In this survey study of breast cancer survivors following breast-conserving therapy, treatment by surgeons with specialized training in surgical oncology was associated with better patient-reported cosmetic outcome, physical well-being, and adverse radiation effects compared to treatment by non-surgical oncologists.