Outlier Practice Patterns in Mohs Micrographic Surgery: Defining the Problem and a Proposed Solution

Key Points

Question  How much variation do physicians demonstrate when performing Mohs micrographic surgery (MMS), and what physician factors are associated with outlier status?

Findings  In this analysis of publicly available Medicare data, marked variation exists in the average stages per case for the treatment of skin cancers with MMS, a specialty society–endorsed quality metric. Physicians in solo practice were 2.35 times likely to be persistent high outliers in all 3 years of the study compared with those in group practice.

Meaning  Wide variation exists in the practice patterns of Mohs surgeons for the treatment of skin cancer; confidential collegial feedback to physicians may reduce unwarranted variation.

Importance  Outlier physician practices in health care can represent a significant burden to patients and the health system.

Objective  To study outlier physician practices in Mohs micrographic surgery (MMS) and the associated factors.

Design, Setting, and Participants  This retrospective analysis of publicly available Medicare Part B claims data from January 2012 to December 2014 includes all physicians who received Medicare payments for MMS from any practice performing MMS on the head and neck, genitalia, hands, and feet region of Medicare Part B patients.

Main Outcomes and Measures  Characteristics of outlier physicians, defined as those whose mean number of stages for MMS was 2 standard deviations greater than the mean number for all physicians billing MMS. Logistic regression was used to study the physician characteristics associated with outlier status.

Results  Our analysis included 2305 individual billing physicians performing MMS. The mean number of stages per MMS case for all physicians practicing from January 2012 to December 2014 was 1.74, the median was 1.69, and the range was 1.09 to 4.11. Overall, 137 physicians who perform Mohs surgery were greater than 2 standard deviations above the mean (2 standard deviations above the mean = 2.41 stages per case) in at least 1 of the 3 examined years, and 49 physicians (35.8%) were persistent high outliers in all 3 years. Persistent high outlier status was associated with performing Mohs surgery in a solo practice (odds ratio, 2.35; 95% CI, 1.25-4.35). Volume of cases per year, practice experience, and geographic location were not associated with persistent high outlier status.

Conclusions and Relevance  Marked variation exists in the number of stages per case for MMS for head and neck, genitalia, hands, and feet skin cancers, which may represent an additional financial burden and unnecessary surgery on individual patients. Providing feedback to physicians may reduce unwarranted variation on this metric of quality.

Mohs micrographic surgery (MMS), a procedure for skin cancer treatment that uses a series of resections to ensure a cancer-free margin while preserving normal tissue, is performed commonly on the head and neck, genitalia, hands, and feet in the United States and is increasingly being used for numerous skin cancers, including melanoma.¹ The effectiveness of MMS has been established, but individual physician practice variation in the mean stages performed per MMS case has not been studied. Unwarranted practice variation can result in unnecessary procedures or poor quality of care for patients.^2,3

The American College of Mohs Surgery (ACMS) is a professional specialty medical association comprised of fellowship trained skin cancer and reconstructive surgeons that recently endorsed its first clinical quality metric: mean stages per case for skin cancers by surgeon. This metric, developed by clinical leaders in the field and recognized by the ACMS, is based on the premise that describing the distribution of practice patterns nationally would elucidate which surgeons have outlier practice patterns, potentially performing too few or too many staged resections for a given tumor, and these physicians could be identified and educated. An excessive number of stages per case may result in overtreatment, increased complications, and cost to patients. Using a novel metric, we evaluated the variation in physician practice patterns to establish a quality of care benchmark.

We conducted a retrospective analysis of all physicians who bill Medicare for MMS in the United States using January 2012 through December 2014 published payment data from the Medicare Provider Utilization and Payment Data: Physician and Other Supplier data set⁴ provided by Centers for Medicare and Medicaid Services (CMS).⁴ We identified physicians performing Mohs resection of a skin cancer of the head and neck, genitalia, hands, feet, or any location with surgery directly involving muscle, cartilage, bone, tendon, major nerves, or vessels (Current Procedural Terminology [CPT] code 17311 for the first stage and 17312 for each additional stage). Mohs surgery of the head, neck, hands, feet, genitalia and deeply invasive tumors accounts for greater than 85% of all Mohs surgery reimbursed by Medicare.

This study was approved by the Johns Hopkins University School of Medicine institutional review board.

Physician characteristics were identified by linking the National Provider Identification number of each physician with the National Plan & Provider Enumeration System National Provider Identification Registry. Characteristics of interest included sex, years in practice (calculated by adding 5 years for residency and fellowship training to the medical school graduation year and then subtracting that value from the year 2014), whether the physician practiced in a solo practice or a group practice (determined by “number of group practice members,” provided in the Physician Compare database), membership in the ACMS, whether the physician practiced at an ACGME (Accreditation Council for Graduate Medical Education) teaching site for MMS, and volume of MMS (volume of CPT code 17311). Volume was then stratified by quartile. Urban or rural practice setting was identified by the physician’s practice ZIP code using data from the American Community Survey for 2013 (United States Census). We identified teaching sites as of 2014 from the ACGME website.⁵ Membership to ACMS is currently obtained by completing a formal ACGME accredited 1-to-2–year fellowship training program in Micrographic Surgery and Dermatologic Oncology following a Dermatology residency, with a minimum 500 MMS case volume including both complex tumors and advanced reconstructions, as well as evidence of an academic research project. Physicians performing fewer than 10 MMS procedures a year were purposely excluded from the Medicare Payment Data by Medicare for privacy concerns.

The mean stages per case per year were calculated from claims codes using the following formula: (volume of 17311 + volume of 17312)/(volume of 17311). Outlier physicians were defined as physicians that had annual mean data that was at the extreme, that is, outside of 2 standard deviations from the mean. We calculated the mean stages per case using volumes from individual years and the total volume of cases from January 2012 to December 2014. Outliers were identified in each year separately, while persistent high outliers were physicians who were outliers in all 3 consecutive years. We also identified low outliers, or physicians who performed an average number of stages per case in the bottom 2.5% of the distribution of physicians, as well as persistent low outliers, who were low outliers in all 3 years of our analysis.

We performed descriptive analysis expressed by percentages and means. Odds ratios (ORs) were calculated using a multivariable logistic regression of persistent high outlier status for an individual year, sex, MMS case volumes, number of physicians in the practice, membership in ACMS, practicing at an ACGME fellowship site, geographic location, and years in practice. All analyses were performed in Stata 12 (StataCorp).

We identified 2305 unique physicians who perform Mohs surgery who met the inclusion criteria (Table 1). This included 2014 physicians in 2012, 2096 physicians in 2013, and 2130 physicians in 2014. A total of 1845 physicians practiced in all 3 years. Physicians who perform Mohs surgery were primarily male (66.8%), and more than half had been in practice for fewer than 20 years (64.7%). Solo practice physicians who perform Mohs surgery comprised 19.0%.

There was marked variation in the mean stages per case per physician for MMS (Figure 1). The mean for all physicians practicing from January 2012 to December 2014 was 1.74 stages per case, the median was 1.69 stages per case, and the range was 1.09 to 4.11 average stages per case with interquartile range 1.51 to 1.89 stages per case and remained consistent in all 3 years investigated. The high outlier status cutoff point was 2.41 stages per case based on the aggregate of data from all 3 years (Table 2). Of the 2305 physicians who perform Mohs surgery from all 3 years, 137 were outliers in at least 1 year; of these, 49 were persistent high outliers in all 3 years. There was similar variation of billing practices in 2 of 140 (1.9%) physicians at ACGME-accredited training locations (Figure 2).

Low outliers were defined as physicians whose average stages per case was in the bottom 2.5% of the distribution. For physicians practicing in all 3 years of our analysis, this cutoff point was calculated to be 1.28 stages per case. There were 92 low outliers in at least 1 year; of these, 20 were persistent low outliers in all 3 years of our analysis.

Practicing MMS in a solo practice was associated with a 2.35-times likelihood of being a persistent high outlier (95% CI, 1.25-4.35) (Table 3). Overall, 4.5% of solo practitioners (17 of 359) were persistent high outliers compared with 2.1% of physicians who perform Mohs surgery in a group practice (28 of 1337). Persistent high outlier status was not associated with sex, practice experience, case volumes, ACMS membership, practicing in an ACGME training site, or geographic location.

Our analysis demonstrates marked individual physician practice variation in the mean stages per case for MMS of head and neck, genitalia, hands, and feet skin cancers. Solo practice was associated with outlier practice patterns for a greater number of MMS stages. Financial motivation may be one of the possible reasons for the variation. The payment model that governs MMS rewards physicians for more resections, or “stages” per case performed—a volume-specific fee-for-service paradigm that has been suggested to drive procedural overuse.^6,7 These charges are directly passed onto Medicare Part B patients, who are expected to pay 20% their health care bill.⁸ The limitation of collegial interaction and peer scrutiny in solo practice may contribute to unawareness of practice variation.

Persistent low outliers were also identified. Possible explanations for low outliers include: (1) incorrect coding, eg, failing to code any subsequent stages (17312) or inappropriate use of the MMS code (17311) instead of standard excision codes when the surgeon is not personally performing the pathologic evaluation; (2) inappropriate tumor selection; or (3) unnecessarily aggressive first stages. Low outliers pose a quality concern because Mohs surgery should be reserved for complex tumors for which cost-effectiveness has been demonstrated.^9-11 Treating simple tumors that could be managed differently or that can predictably be cleared in a single stage undermines the value of Mohs surgery, as does harvesting even appropriate tumors with excessively wide layers that would require larger repairs. Mohs surgery is intended to both provide the highest cure rates for complex tumors in challenging anatomic sites and to necessarily preserve critical tissue to allow the most functional and cosmetic outcomes.^12,13 Low outliers likely negate these benefits.

A previous study¹⁴ of Mohs surgery cases on Medicare beneficiaries in 2009 described an average of 1.75 stages per case by physician among 1777 providers and characterized variation in the case volumes of physicians who perform Mohs surgery. Our report reinforces this statistical result through more current, comprehensive data analysis and further describes wide variations in the practice of MMS. This novel quality metric endorsed by the ACMS is recommended to benchmark physician performance and to inform quality improvement efforts.

In other areas of medicine, it has been shown that a financial incentive may alter practice patterns and physicians may be unwilling to change patterns.^15,16 A study of 50 urology practices showed that urologists who owned linear accelerators for radiation therapy tended to self-refer patients to their own radiation services at no definitive benefit to the patients.¹⁵ Physicians may not follow clinical practice guidelines due to a range of reasons that include financial benefits, skepticism or lack of awareness of guidelines, and unwillingness to change practice habits.¹⁶ To improve appropriate case selection for MMS, Appropriate Use Criteria (AUC) have been established for Mohs surgery^12,13 but have not been widely embraced by insurers. These AUC criteria and other clinical guidelines such as those established for skin cancer by The National Comprehensive Cancer Network,^17,18 American Joint Commission on Cancer¹⁹ and American Society of Dermatologic Surgery^20,21 should serve to inform treatment decisions for skin cancer.

Unnecessary surgery financially burdens both insurers and patients. Given the direct cost to patients, we suggest that patients would benefit from knowing if their physician has a typical practice pattern (within 2 standard deviations from the mean) or is an extreme outlier. New publicly available data released by Medicare poses a novel ethical dilemma: now that extreme outliers can be identified in publicly available data, does the patient have a right to know? And what role does organized medicine play in addressing variation in care? The ACMS believes that a professional specialty medical association has a duty to notify, engage, educate and help extreme outliers that demonstrate practice patterns believed to represent unwarranted variation by peer experts.

One approach to the problem of unwarranted practice variation is internal data transparency. The ACMS in 2016 initiated a national quality collaborative as part of a broader multispecialty national quality improvement endeavor called Improving Wisely,² funded by the Robert Wood Johnson Foundation and based at Johns Hopkins University. The project uses a physician performance metric developed and endorsed by the respective specialty association to inform physicians of their individual performance data in a confidential and nonpunitive data report. Improvement among outliers in this project is based on the conceptual model that (1) outliers may be unaware that they are outliers, (2) no one wants to be an outlier, and (3) a civil, confidential, peer-to-peer data sharing dialogue in the spirit of improvement is an effective way to engage physicians.²² These reported data serve as a baseline assessment for subsequent analyses and as a benchmark group for quality improvement efforts. Furthermore, for the clear majority of physicians who are not outliers, the data transparency helps to positively reinforce nonoutlier practice patterns. Preliminary survey data from the annual ACMS meeting suggested that the members were favorable to these efforts, with 94% of respondents saying they were strongly or moderately in favor of learning how their practice patterns compared with those of their peers. Most (87%) of the surveyed physicians felt it would be beneficial for them to see their comparative performance data, and 71% believed that a data transparency program among physicians could lower unnecessary health care costs. Given the estimated magnitude of health care waste based on the 2012 Institute of Medicine report, Best Care at Lower Cost, the Improving Wisely peer-comparison method could be a practical and useful tool to reign in health care costs.²³

An earlier study of 2851 primary care physicians treating Medicare beneficiaries showed that physician spending patterns mirrored those of their residencies, regardless of ultimate practice location.²⁴ Given these results, fellows and residents may be imprinted with the practice tendencies of the physicians that mentor them at training sites. Thus, addressing wide variation particularly at training sites could help prevent avoidable variation among the next generation of Mohs surgeons. Notably, there were 2 persistent high outliers among the 140 Mohs surgeons practicing at ACGME accredited training locations, and 1 persistent low outlier. Education in proper billing and coding, appropriate case selection, and conservative Mohs surgery technique is a required component of ACGME training and could explain this result.

This study’s primary limitations include that the analysis could not account for clinical patient factors and/or referral bias. If outliers were treating a severely high-risk patient population, the challenging tumors on high-risk anatomic sites may appropriately require more stages of Mohs surgery to obtain clear margins.¹¹ Because neither the Mohs codes nor the database provide information regarding resection diameter or depth, case severity is difficult to assess. In addition to lacking risk stratification for tumor severity, an adjustment for reconstructive effort is absent, which directly impacts the ultimate cost and value equation. Our data does not include third-party private payers and physicians belonging to the Veterans Administrations, United States Armed Services, or group practices like Kaiser Permanente who do not generate direct billings to Medicare. We report data from Medicare’s most recently available years, up to 2014. Practice patterns may have changed in the 3 years since the public availability of this data, which would not be captured by our analysis. Because the CMS database excludes those codes billed fewer than 10 times per year, we were unable to account for a small percentage of physicians who may have performed MMS with an average of 1.0 stages per case, an extreme category of low outliers. Transparency to the public may have provided a unique, external, nonfinancial incentive to motivate physicians to reduce their prior variation. More real-time data would better describe the current state of practice variation and allow for more actionable data for quality improvement, an effort that is currently under way with CMS by the Improving Wisely program.

Physician characteristics that tend to produce outliers in medical practice are not well understood. Furthermore, case stratification needs to be considered in any discussion of practice patterns. Our analysis found that there is notable variation in practice patterns in MMS and that practicing in a solo practice is associated with a greater risk of producing outliers. We believe our Improving Wisely initiative can help educate both isolated and less experienced outlier physicians. We suggest that physicians who bill for Mohs surgery for appropriate tumors should have a mean of 1.7 stages per case, with a maximum variation of greater than 1.3 to less than 2.4, recognizing that risk adjustment is necessary prior to any recommendations. Furthermore, there is wide variation among physicians practicing at ACGME teaching sites for MMS that may contribute to the national variation.

While the effectiveness of MMS has been well demonstrated, outliers may negate this benefit for their patients.^11,20,21,25 Because of both the cost and morbidity that outliers and overuse in Mohs surgery can impose on patients and the health care system, physicians must proactively address the underlying causes contributing to these issues. Recognizing that outliers represent only a minute fraction of practitioners and may be providing appropriate care to a higher-risk population, in our experience, we feel that peer physicians within a specialty society represent the best option to rectify inappropriate outlier practices.

Corresponding Author: John G. Albertini, MD, The Skin Surgery Center, 1450 Professional Park Dr, Ste 150, Winston-Salem, NC 27103 (john.albertini@skinsurgerycenter.net).

Accepted for Publication: April 3, 2017.

Correction: This article was corrected on August 30, 2017, for an incorrect P value in Table 3.

Published Online: April 28, 2017. doi:10.1001/jamadermatol.2017.1450

Author Contributions: Messrs Krishnan and Xu had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Krishnan, Xu, Hutfless, Stasko, Vidimos, Leshin, Bennett, Marks, Brandt, Makary, Albertini.

Acquisition, analysis, or interpretation of data: Krishnan, Xu, Hutfless, Park, Stasko, Leshin, Coldiron, Bennett, Makary, Albertini.

Drafting of the manuscript: Krishnan, Xu, Coldiron, Makary, Albertini.

Critical revision of the manuscript for important intellectual content: Krishnan, Xu, Hutfless, Park, Stasko, Vidimos, Leshin, Bennett, Marks, Brandt, Makary, Albertini.

Statistical analysis: Krishnan, Xu, Hutfless, Park, Bennett, Makary.

Obtained funding: Makary, Albertini.

Administrative, technical, or material support: Xu, Leshin, Coldiron, Brandt, Makary.

Study supervision: Xu, Vidimos, Makary, Albertini.

American College of Mohs Surgery Improving Wisely Study Group: John G. Albertini, MD (The Skin Surgery Center and the Department of Plastic and Reconstructive Surgery, Wake Forest University School of Medicine, Winston-Salem, North Carolina); Thomas Stasko, MD (Department of Dermatology, University of Oklahoma, Oklahoma City, Oklahoma); Allison T. Vidimos, MD, RPh (Cleveland Clinic, Cleveland, Ohio); Barry Leshin, MD (The Skin Surgery Center, Winston-Salem, North Carolina); Brett M. Coldiron, MD (University of Cincinnati Hospital, Cincinnati, Ohio); Richard G. Bennett, MD (Bennett Surgery Center, Santa Monica, California); Victor J. Marks, MD (American College of Mohs Surgery, Milwaukee, Wisconsin); Rebecca Brandt (American College of Mohs Surgery, Milwaukee, Wisconsin).

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported in part by a grant from the Robert Wood Johnson Foundation, (grant No. 73417 to M.A.M.) and by funding from The American College of Mohs Surgery.

Role of the Funder/Sponsor: The funder/sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Meeting Presentation: This study was presented at the 49th Annual Meeting for the American College of Mohs Surgery; April 28, 2017; San Francisco, California.

Additional Contributions: We would like to thank the Robert Wood Johnson Foundation and the American College of Mohs Surgery for their support of this work.