The Centers for Medicare & Medicaid Services (CMS) uses the resource-based relative value system to determine payment for physicians and nonphysician practitioners for their professional services. For many surgeries and other types of procedures, Medicare payment also covers a bundle of pre- and post-operative visits delivered during a global period of 10 days or 90 days anchored on a surgery date. In the final rule for the 2015 physician fee schedule, CMS announced that all surgeries with a 10- or 90-day global period would transition to a 0-day global period in 2017 and 2018, respectively. CMS's rationale for scaling back global surgical packages was driven by concerns over the accuracy of the payment for post-operative care. In Section 523 of the Medicare Access and CHIP Reauthorization Act of 2015, Congress directed CMS not to transition all 10-day and 90-day global surgery packages to 0-day global periods (Public Law 114–10, 2015). Instead, Congress mandated that CMS collect data on the "number and level" of visits in the global period from a representative sample of physicians beginning January 1, 2017.
CMS previously asked the RAND Corporation to provide recommendations on how to best collect the number and level of post-operative visits through the use of nonpayment claims. Based on the input of an expert panel, we proposed a new set of codes to CMS that combined scope of services with time, for both inpatient and office-based services. The rationale and description of these codes (nonpayment codes referred to as G-codes in this study) were summarized in a prior RAND report (Mehrotra et al., 2016). In July 2016, CMS issued a proposed rule that included a slightly modified version of the post-operative visit G-codes that RAND had developed for CMS (CMS, 2016a) and proposed to require their use by practitioners for certain services. The G-codes from the proposed rule had never been tested or used by practitioners.
Therefore, CMS asked the RAND Corporation to pilot the G-codes from the proposed rule with the goal of testing whether practitioners (and coding/billing experts if applicable) understood and could accurately apply the codes. After this testing was completed, CMS issued a final rule for calendar year 2017 that requires practitioners to use Current Procedural Terminology (CPT®) code 99024 instead of the proposed G-codes to report post-operative services in the global period (CMS, 2016b). Therefore, the G-codes tested in this study will not be used by practitioners to report services to Medicare. We provide this study, however, to document our findings in case they may help to inform any future discussions about similar nonpayment codes.
Our overall approach was to create a series of vignettes and to test the use of these vignettes using semi-structured interviews with a small set of physicians who perform procedures. We then conducted more-extensive testing through surveys with a larger group of physicians. First, we identified five specialties to test the proposed G-codes: cardiology, dermatology, general surgery, neurosurgery, and ophthalmology. For each of the five specialties, we created three clinical vignettes that described patient visits that were varying combinations of either inpatient or outpatient, and typical or complex. We then conducted interviews with one physician from each of the specialties. The goals of these interviews were three-fold: (1) to assess how physicians understood the codes and would apply them to recent visits; (2) to refine our vignettes for the survey; and (3) to pilot test use of the proposed G-codes on the newly created vignettes.
Based on input from the interviews described above, we developed an online survey to test the proposed G-codes among a larger sample of physicians from the same five specialties. In the interviews, respondents could generally accurately apply the codes to vignettes, and to recent actual visits. In the survey, accuracy of coding was 71 percent for choosing the correct code on the basis of setting and complexity, and 61 percent for choosing the correct time increment.
Comments from both interviews and the survey coalesced around several concerns with the proposed G-codes: the burden of reporting nonpayment codes, keeping track of time spent, the definitions of "typical" and "complex," and how the codes capture work done by multiple practitioners. Our testing uncovered valuable insights that could be useful if CMS considers similar new codes in the future.
