COA Letter to CMMI Regarding Challenges That Need to Be Addressed in the OCM and Future Payment Reform Models

Author: Community Oncology Alliance

May 31, 2019

Mr. Adam Boehler
Director
Center for Medicare & Medicaid Innovation
Centers for Medicare & Medicaid Services
7500 Security Boulevard
Baltimore, MD 21244

Dear Adam:

On behalf of the Board of Directors of the Community Oncology Alliance (COA), the purpose of this letter is to provide you and the Oncology Care Model (“OCM”) Center for Medicare & Medicaid Innovation (“CMMI”) team with four challenges we believe need to be addressed in the OCM and any future oncology payment reform models. The four areas that the CMMI OCM team need to address are as follows:

1) Price Prediction

2) Risk Adjustment

3) Attribution and Monthly Enhanced Oncology Services (“MEOS”) Payment Recoupment

4) Timeliness of Data/Information

We want to make clear that these are constructive observations about flaws we see in the OCM and provide them in the spirit of improving the OCM and advancing oncology payment reform.

As I believe you know, the Community Oncology Alliance (“COA”) is dedicated to advocating for community oncology practices and, most importantly, the patients they serve. COA is the only non-profit organization focused solely on independent community oncology, the setting where the majority of Americans with cancer are treated. The mission of COA is to ensure that patients with cancer receive the highest quality, most affordable and accessible cancer care in the communities where they live and work. For over 16 years, COA has built a national grassroots network of community oncology practices to advocate for public policies that benefit patients with cancer. Individuals from all perspectives of the cancer care delivery team – oncologists, administrators, pharmacists, mid-level providers, oncology nurses, patients, caregivers, and survivors – volunteer their time on a regular basis to lead COA and serve on its committees.

COA currently has over 80 percent of the OCM participants in an online community and learning network that shares best practices, insights, challenges, solutions, etc. in dealing with the OCM. Both independent community oncology practices and hospital systems participate in regular conference calls and webinars, and on a daily basis, converse via a dedicated, closed listserv to participants. In short, COA is committed to making the OCM a success.

We appreciate working with the CMMI OCM team. They have been very responsive to our requests and have made important adjustments to the OCM as the demonstration project advances. The OCM has made amazing progress given that it is a government program which has regulatory burdens not seen in commercial oncology payment models. Specifically, we appreciate the progressive adjustments and improvements made to the OCM in its first three years. More recently, we applaud CMMI’s release of a new alternative two-sided risk model in the OCM, the announcement of additional risk adjustment for metastatic disease in breast, lung, and intestinal cancers, as well as the announcement regarding the potential uses of available staging and clinical data from the OCM Data Registry for patients with breast, colon and lung cancer. These updates have been well received by the OCM participants and are steps in the right direction to assuage concerns and foster continued participation in the OCM.

However, changes to date have not comprehensively addressed the range of challenges and potential risks that practices face inside the OCM program. With the February 2019 reconciliation, we learned that only 30 percent of practices achieved shared savings in Performance Period 3 despite improvements to the methodology that went into effect. This is the same proportion, but different practices, that earned a Performance-Based Payment in the Performance Period 2 initial reconciliation. Also, in the February 2019 reconciliation, practices experienced unexpected regressions in their Performance Period 2 true-up results, with a significant number of practices losing their shared savings. These trends are particularly concerning for three reasons.

First, in the beginning stages of the OCM, many practices focused on fulfilling program requirements. As the model progressed, practices increasingly focused resources on substantive care management interventions. Despite this, we did not see progressive improvements in performance results as expected in the field.

Second, while we believe that the methodology changes that CMMI has made to the model since its launch have been significant improvements, practices still do not have a comprehensive understanding of their exposures to risk. OCM participants need more time to analyze the impact of the methodology changes, particularly the ones that went into effect for Performance Period 3. Specifically, they need at least three sets of reconciliation results and reconciliation data in order to make an informed decision on their potential risk exposure in a two-sided model.

Third, the OCM methodology states that practices can remain in one-sided risk if they achieve at least one shared savings payment by Performance Period 4. Practices that initially thought they were exempt from downside risk but lost their shared savings through the true-up process are now facing significant uncertainties and unanticipated exposures to risk and the complex decision of whether or not to stay in the model.

In order for practices to make informed decisions regarding their risk exposure related to staying in the OCM and taking on downside risk, participants particularly need additional clarity on the accuracy and volatility of episode price setting and risk adjustment. They will also benefit from clarity on how CMMI may evolve the model to ensure it is flexible enough to account for the complexities that are innately associated with cancer conditions. Additionally, given that the model is approaching its final phase, practices are seeking guidance on what that means if they were to take on two-sided risk. We encourage CMMI to provide information, as specific as possible, regarding the OCM, its relationship to the CMS Quality Payment Program – and after the OCM is currently scheduled to end.

What follows are some more detailed points relating to the four areas of concern we summarized above. For each area, we will summarize the related issues and then describe specific example cases. We examine each issue in more detail and conclude each section with a suggested solution(s) that will help improve the model’s predictability and flexibility.

In the near term, given these complexities, our immediate recommendation is to delay the October 2019 deadline on two-sided risk to April 2020 in order to give practices a third set of reconciliation data and results post-implementation of the updated the OCM risk adjusters for breast, prostate and bladder cancer. Making the decision on whether to accept downside risk based on only two sets of reconciliation results/data after these key corrections to the methodology were made exposes practices to significant uncertainty rather than quantifiable risk. This potentially threatens their viability and in turn, access to care for Medicare beneficiaries. The additional results and time to decide whether or not they will take on downside risk will enable participants to gain critical insight into their performance and move to risk in a more viable pathway. Delaying this deadline will keep practices in the model and give participants the additional flexibility necessary to evaluate the methodology changes and the impact on their practice.

Finally, we would like to set up a meeting or call with the CMMI OCM team to discuss these issues and other noted requests in greater detail. Although our concerns may seem overly detailed as presented by the cases that follow, we use the cases, which are real-life OCM experiences, to illustrate the larger issues and challenges.

Price Prediction

SUMMARY: Applying price adjustments at the level of the practice, rather than the episode, leads to underpricing for practices that deviate from average national distributions; e.g., of population-level characteristics or cancer types (Cases 1-3). Practices with atypical Part D enrollment are unable to meet their target amount, which assumes average Part D enrollment rates, especially given steady increases in oral drug prices (Case 1). Finally, the novel therapy adjustment does not help practices which use a novel therapy for a cancer type with a small number of episodes (Cases 2-4).

Case #1: A male in his late 60s has multiple myeloma. He begins a Part D treatment of lenalidomide, ixazomib, and dexamethasone. Over the six-month course of treatment, the cost of these Part D oral drugs is $95,000, nearly twice his target price of $51,000. He has no other utilization in his six-month episode besides monthly office visits.

There are many episodes that are over target, with very little hospitalization use, emergency department, or post-acute spending. Pharmaceutical costs are driving much of the cost in oncology, even when patients have little or no utilization of facilities or emergency services. In particular, Part D drugs have steadily increased in price over the past five years. Because Part D coverage rates vary across OCM participants, as does access to particular drugs like lenalidomide, many practices are particularly vulnerable to these target overages. Practices have no way to control the price of prescription drugs in their patients’ episodes, especially when these prices increase frequently.

Suggested solution: Part D agents should be accounted for directly in price prediction, at the episode level.

Disease prevalence and practice population

Case #2: An elderly man triggers an episode with a treatment of cabozantinib. He receives cabozantinib every month for the duration of his episode. In the fifth month of his episode, he is admitted through the emergency department (“ED”) with shortness of breath and has a $13,000 hospital stay that lasts one week. He has no other utilization outside of office visits during the episode. Overall, his episode is $58,000 over target.

The OCM faces an important challenge of accurately pricing emergent therapies and pharmaceutical innovation. Rarely, if ever, are past prices predictive of future costs. In these instances, the trend factor and novel therapy adjustment operate at too high of a level: they are applied at the practice level rather than the episode level. Practices may not be able to achieve a novel therapy adjustment for several reasons:

  • They may treat a wide range of cancer types, many of which are treated with generic or otherwise non-novel chemotherapy. The novel therapy adjustment does not account for a practice’s cancer mix, despite some cancers having more, or fewer, agents on the OCM novel therapy list. For practices using extremely expensive novel agents in low-prevalence cancers, it is virtually impossible to earn a novel therapy adjustment.
  • Due to regional variations or other specialists in their region who treat patients with existing cancer, a practice may see a smaller proportion of patients who are eligible for novel therapies compared to the average non-OCM practice.
  • Patients’ staging data is not used or updated to account for whether episodes are eligible for novel therapy utilization.

Suggested solution: Adjustments for emergent therapies should occur at the episode-level, based on which drugs are actually prescribed to the patient, using staging data to determine whether the episode is eligible, or ineligible, for novel therapy utilization.

Case #3: An elderly woman has low-risk breast cancer. During her episode, she has a month-long skilled nursing facility (“SNF“) stay for heart failure. After discharge from the SNF, she is hospitalized once for heart failure and a second time for cirrhosis of the liver. Finally, she has a colonoscopy and has several lesions removed. Her target price for low-risk breast cancer is $6,000, but her actual episode cost is nearly $50,000.

For outlier episodes in low-risk cancer types, it takes a huge number of under-target episodes to “average out” these expensive cases. In this case, it would take two low-risk breast cancer episodes averaging $1,000 below target in order to make up the difference for this one outlier. Especially for episodes with low target prices, like low-risk breast cancer and low-risk prostate cancer, it isn’t possible to “average out” the cost of catastrophic cases by having multiple patients who are below target. Practices with even a handful of outlier episodes are unable to meet their target prices for low-risk cancer types.

Suggested solution: CMMI should improve pricing for outlier patients so that practices are not penalized, or significantly over-target, due to a few outlier episodes. Episode-level pricing of novel therapies and chronic conditions would help improve episode pricing for practices who cannot absorb the risk of having excessively expensive episodes, especially for patients with low-risk cancers.

Case #4: A practice treats an elderly patient with bladder cancer. After being treated with other chemotherapy drugs, the patient is prescribed atezolizumab. The patient’s treatment continues for the entire 6-month episode with no hospitalization, ED, or post-acute facility stays. Nonetheless, the patient’s drug costs alone exceed $60,000, more than twice the target price of $30,000. As part of a quality improvement initiative, the care team reviews this case and a series of similar cases and finds that 11 out of 12 cases have appropriate use of atezolizumab. The practice does not earn a novel therapy adjustment.

In addition to problems with cancer prevalence and risk adjustment (discussed in the following Risk Adjustment section), which is presented in this case, the practice fails to earn a novel therapy adjustment because atezolizumab was listed on the novel therapy list for nine months before it was assigned a distinguishing HCPCS code. OCM Practices around the country failed to have their atezolizumab costs counted toward their total novel therapy adjustment between July 1, 2016, and December 31, 2017, because atezolizumab was listed under a generic novel therapy code of J9999.

Suggested solution: The novel therapy adjustment should be updated to include atezolizumab before December 31, 2017.

Risk Adjustment

SUMMARY: Risk adjustment for global outlier cases is not adequately addressed, as illustrated in Cases 5-11 below. Risk adjustment in OCM also fails to account for many cancer-related surgeries, which are relevant for determining the complexity of a patient’s case. There are limited exclusion criteria in the OCM for high-intensity medical conditions not related to cancer, such as strokes and orthopedic procedures. Finally, accounting for risk related to enhanced disease severity, late presentation of illness, progression of disease, and metastases is inadequately accounted for in the current methodology. While the planned updates announced on May 1, 2019 will use clinical data to add to the risk adjustment methodology, significant additional challenges still exist. Given the complexity and broad-ranging impact of metastatic disease on the severity of cancer and the expected resource utilization, improvements are needed to equitably shift risk to practices while still retaining patients’ access to care.

The problem of risk adjusting the highest cost patients is a widely recognized problem in health care and the OCM pricing model consistently underestimates the cost of care for high-risk, outlier patients. Practices who treat higher risk patients are penalized by this effect. The cases below describe these issues in greater detail.

Suggested solution: More sophisticated pricing models would allow for risk adjustment for high-risk patients, so that practices are not penalized for serving high-risk populations. In addition, the OCM pricing methodology should be more clinically motivated. Additional clinical acumen needs to be married with statistical modeling to better predict clinical risk. One step would be to include price adjustments for specific chronic conditions, rather than simply counting the number of HCCs for each patient.

Case #5: An elderly woman is being treated for malignant melanoma. During the course of her care, she is admitted through the ED for secondary malignant neoplasm of the small intestine, a non-reconciliation eligible cancer. During her hospitalization, she receives surgery to remove her small intestine. However, there is no risk adjustment for the non-reconciliation eligible cancer surgery, so her target price is set at $58,000. Her actual episode cost is $150,000.

Currently, there is no risk adjustment for surgeries on non-reconciliation eligible cancer types.

Suggested solution: We recommend adding surgeries related to all cancer types to the surgery list so that if patients have surgery for any type of cancer, they will have an increased target price reflecting the increased complexity of their cancer episode.

Exclusion Criteria

Case #6: A man in his late 70s has prostate cancer and triggers an episode with a dose of leuprolide. The patient has hemophilia A, also known as Hereditary Factor VIII deficiency, and receives multiple infusions of factor VIII throughout his episode, running up a total episode cost of $2.8 million. Despite having his episode winsorized down, the episode is $23,000 over target.

Case #7: A woman in her mid-70s has pancreatic cancer. During her episode, she is admitted through the ED for Guillain-Barre syndrome and has a nearly two-week hospital stay, followed by a month in a SNF. Her target price is $26,000, but given the complications from Guillain-Barre syndrome, her actual episode cost is $85,000, more than three times the target price.

Patients with comorbidities, such as blood clotting factor deficiencies and high-complexity autoimmune disorders, fall outside the scope of cost containment that can be achieved by oncology practices and have typically been excluded from advanced payment models.

Suggested solution: Similar to other payment models, such as the Bundled Payments for Care Improvement (“BPCI”) Initiative, patients with disorders such as hemophilia and Guillain-Barre syndrome should be excluded from the OCM.

Case #8: A patient with low-risk breast cancer triggers an episode with no evidence of bone metastasis. Within a week of the start of her episode, the patient falls and is hospitalized for a non-pathological femur fracture. She is discharged to a SNF for an extended stay. Her target price was $6,500, but the actual cost of her episode is approximately $40,000.

Case #9: A patient with low-risk breast cancer triggers an episode with a dose of letrozole. Four days after the start of her episode, she is admitted for elective knee replacement surgery at a cost of $12,000. She continues on letrozole and has physical therapy for her knee. She has no other cancer treatment during her episode but is $15,000 over target.

Case #10: A patient with high-risk breast cancer triggers an episode via treatment with fulvestrant. In the last month of her episode, she is hospitalized twice within three days with a diagnosis of cerebral infarction. Her hospital costs are $11,500. With no other cancer treatment during her episode, she is $16,000 over target.

Oncologists can’t control traumatic events that occur during a six month total-cost-of-care episode. Based on practice location, access to care, referral patterns, etc., some practices systematically have more patients with high-cost non-oncology traumatic events than others. Oncologists should not be held to account for elective knee or hip replacement surgeries, which are unrelated to cancer care. Smaller practices with few episodes may be particularly hard hit by a few such episodes in a given time period.

Suggested solution: In other CMS models, such as the Comprehensive Care for Joint Replacement Model and BPCI Advanced, exclusions are listed for high-utilization events that are unrelated to the triggering diagnosis. For example, CMMI has prior excluded clotting disorders, such as hemophilia, from these prior payment models. Given the presence of hematologic disease in many OCM practices, such exclusions are particularly important in the OCM. Patients with blood clotting disorders, strokes, non-pathologic fractures, and unrelated high-cost conditions should have those costs excluded from their OCM episode cost.

Cancer Attribution

Case #11: An elderly man triggers into the model with a diagnosis of malignant neoplasm of the prostate and treatment with bicalutamide. The patient is also receiving denosumab. The patient has extensive imaging done with nuclear medicine for secondary malignant neoplasm of the bone. He has multiple hospitalizations for a variety of different conditions, consistent with progressive respiratory distress and passes away. His attributed cancer is low-risk prostate, but his episode is $48,000 over target.

Many patients have denosumab treatment on their inpatient claims that can be used to infer bone metastases. Unfortunately, the diagnosis of secondary malignant neoplasm of the bone may not be coded in the E&M. We expand on these problems in the next section.

Suggested solution: The OCM risk adjustment algorithm should incorporate all relevant information from across payment systems. With this information, CMMI should develop new algorithms across payment systems to more accurately model risk.

Attribution and MEOS Payment Recoupment

SUMMARY: Patient attribution continues to be challenging for patients on oral chemotherapy regimens (Cases 12-13). While many episodes are accurately attributed to cancer, a disproportionate number likely have the wrong cancer attributed for the episode (Cases 14-15). Episodes involving multiple cancers or sequelae need to be addressed in the risk adjustment methodology.

Patient attribution:

Case #12: A patient with low-risk breast cancer is on an infrequent dose of Tamoxifen. She sees an oncologist, associated with her health care system, who prescribes a year of Tamoxifen. She continues to have ongoing office visits, which are focused on managing her chronic diabetes and hypertension and do not list a breast cancer diagnosis. About five months into her episode, she gets a second opinion from a second oncologist, who does not prescribe any drugs but codes the visit for breast cancer. The patient is attributed to the oncologist who gave a second opinion, since that was the only evaluation and management (“E&M”) visit with a cancer diagnosis after the patient’s trigger. The trigger was the Tamoxifen prescription fill date.

When patients are exclusively on a Part D oral chemotherapy regimen, it is difficult for practices to determine their episode trigger date. Oncologists have no way of knowing whether their patients are seeing other practices for second opinions or additional E&M visits. If a patient does not fill their Part D chemo prescription on the date it was subscribed, there is a very likely chance of missing this attributed patient.

Suggested solution: Practices should be notified as quickly as possible after an episode is triggered so that patients can be monitored and their care can be managed. This would also enable physicians to schedule E&M visits prospectively with their OCM patients and make sure that they are coded correctly to ensure attribution. Another possibility would be for patients to be attributed to an oncologist based on the triggering chemotherapy claim. This would allow practices to know which patients were their “active episodes” with a prospective view, rather than relying on existing delayed retroactive attribution methodology. Finally, MEOS bills themselves could be utilized to determine patient attribution, with additional tie-breaking factors if a patient is being seen in multiple locations.

Case #13: A patient with low-risk breast cancer is taking anastrozole every two months. She sees her oncologist in November but forgets to refill her Part D medication in January. When she does refill it at the pharmacy, it is March, and it has been four months since she has seen her oncologist. The refill triggers a second episode in March, and within two weeks she is admitted through the ED for secondary malignant neoplasm of the bone, with a fractured vertebra. After five days in the hospital, she is discharged to a SNF for a month with generalized weakness and finally sees her oncologist in May. At that point, she is immediately started on palbociclib and fulvestrant, since her cancer is now metastatic. It has been five and a half months since her last E&M visit with her oncologist.

Oncologists have difficulty managing patients when they have no insight into utilization events outside their practice. This is particularly true for patients on Part D drug regimens for maintenance therapy who, according to guidelines, should not have frequent visits with their oncologist. Many hospitalizations could be avoided if oncologists had better insight into which of their patients were failing to fill their prescriptions or going to the emergency room for issues that could be related to their cancer. More timely delivery of claims data would allow practices to identify their patients, especially those with whom they do not have frequent visits, using all available information.

Suggested solution: Release of monthly reports, rather than quarterly reports, would improve practices’ ability to manage patients’ care. These reports would be very helpful in reducing attribution discrepancies in existing methodology reports. For the purposes of managing patients’ cases, final action claims are not necessary; practices would be better able to coordinate care for their patients from simply knowing that a claim has been filed.

Cancer Attribution

Case #14: An elderly gentleman triggers a prostate cancer episode with an infusion of trastuzumab and docetaxel. Three days later, he has two imaging visits that show malignant neoplasm of the parotid gland. While taking trastuzumab every three weeks throughout his episode, he has 11 imaging claims with codes for parotid gland cancer, secondary malignant neoplasm of the brain, and secondary malignant neoplasm of the bone. In addition, the patient has a single ED visit for specified diseases of the anus and rectum. His E&M visits variably list prostate cancer, neoplasm of unspecified behavior of the digestive system, secondary malignant neoplasm of the brain, and malignant neoplasm of unspecified male breast. There are also five doses of denosumab, further supporting an inference of bone metastasis. Together, these claims suggest a patient with prostate cancer, with multiple non-specific malignancies which are poorly differentiated.

Despite having access to multiple diagnoses and claims, the OCM methodology only uses the diagnoses from E&M visits to determine cancer type. Practices have also provided staging data for their patients.

Suggested solution: The OCM should use the full set of information from a patient’s claims and staging data, meaning cancer type could be more accurately assigned. Patients should be assigned to the most resource-intensive cancer type from their various data sources, which would capture the complexity of care required to manage these patients.

Case #15: An elderly gentleman receives a single dose of leuprolide to trigger his episode of low-risk prostate cancer. Three months into his episode, he is hospitalized with secondary malignant neoplasm of the cerebrum. He returns to the hospital for surgery to remove the brain lesion, which is followed by long recovery stays in both an Inpatient Rehabilitation Facility (“IRF”) and a SNF. His actual episode cost for low-risk prostate cancer is $52,000 over target.

The current model fails to incorporate claims data that indicates metastatic disease. Moreover, current coding practices mean that even after a patient experiences metastatic cancer or a second cancer type, the treating oncologist will still code the initial primary cancer type as the primary diagnosis.

Suggested solution: Risk adjustment should include evidence of metastatic disease and adjust prices accordingly. Incorporating diagnoses from drug administration claims, hospitalizations, non-plurality E&M diagnosis codes, and other types of claims would help this aim, as would incorporating patient diagnosis codes from outside the episode boundaries. It is a great advance that CMMI is using the staging data to differentiate patients with breast, lung, and intestinal cancer who have metastatic disease, but more can be done with the staging data, as well. Practices are still exposed to highly varied risk depending on a range of factors related to metastatic disease, including the scope, extent, pattern of progression, location, chronicity, and underlying biology. In addition, metastatic disease is a widespread issue across all cancers, and not only the three types identified.

MEOS Recoupment:

Case #16: A patient with lung cancer begins treatment with their primary oncology practice. After five months of treatment and care management services by their oncology practice, they switch to a Medicare Advantage plan. The oncology practice has billed five MEOS payments for this patient for services rendered but now owes all of that money back in recoupments.

Identifying OCM patients prospectively is extremely challenging. Oncologists have no control over several disqualifications, including patients losing Medicare Part A and B coverage, enrolling in a Medicare Advantage program, switching to a different primary payer, or becoming an end-stage renal disease patient. Practices are often unable to identify when patients lose eligibility for the OCM and to halt MEOS billing for those patients. Moreover, the patient attribution list is released to practices more than a year after the beginning of a performance period. The MEOS billing window is often closed by the time practices are given the final list of their patients. This means practices are unable to use the attribution list to capture MEOS payments for their attributed patients. The time lags and uncertainty surrounding MEOS recoupments are challenging for practices and mean that oncologists may not be able to spend MEOS money on patient care until they are sure those funds will not be recouped. Further, practices in some regions are particularly impacted by turbulent Medicare Advantage markets and frequent Medicare Advantage enrollments that occur mid-episode.

Suggested solution: CMMI has the data to determine patient eligibility and patient attribution much sooner after the performance period has ended. By providing that data to practices sooner, practices could focus their resources on care management activities, rather than figuring out patient attribution, tracking oral fill dates across multiple pharmacies, and tracking disqualifications from the OCM. Furthermore, allowing practices to keep MEOS funds for patients who have received additional services but lost OCM eligibility mid-episode, would reduce the unpredictability practices currently experience.

Case #17: Many practices have reported inconsistencies and frustration in MEOS recoupment efforts by CMS. CMMI specified that OCM practices would receive a written letter of MEOS monies that are to be refunded. Practices would then be given 30 days to repay the amount in full. Many practices have reported they did not receive this letter. Recoupment began without warning and was initiated through patient line items on remittances. These remittances are posted electronically.

Adherence to CMMI guidelines through the Medicare Adherence Carriers has not been consistent. Additionally, the process to recoup overpaid MEOS amounts has created patient balances that are for an unrecognized service that was at least 12 months earlier. This has caused patient anxiety, as well as practice confusion and frustration. It also increases administrative burdens while reducing goodwill with the OCM program.

Suggested solution: The collection letters should be generated and sent through CMMI to assure consistency in following their own policy. In addition, the CMMI OCM Project Officer for assigned practices should send a note to practices when the letter is sent. Then, if the practice fails to make payment in full within 30 days, recoupment should be made a summary transaction against an electronic remittance – and not as a patient account line item.

Timeliness of Data/Information (for Improved Care Coordination)

SUMMARY: Practices need faster access to claims data in order to manage their patients’ utilization in a total cost of care model.

Case #18: A man in his early 80s triggers a prostate cancer episode with bicalutamide. After two months with very little utilization, he has two ED visits for pain in the knee and secondary malignant neoplasm of the bone. For the final four months of his episode, he had eight admissions through the ED, an IRF stay, and two SNF stays. His six-month episode ends up costing $140,000. His oncologist has no knowledge of this because he is in a community oncology practice and has no information via claims until well after the patient’s episode has ended.

Oncologists have difficulty managing patients when they have no information on patient treatment utilization outside their practice. Many hospitalizations could be avoided if oncologists had better information on which of their patients were failing to fill their prescriptions, going to the ED, or being admitted for issues that could be related to their cancer. More timely delivery of claims data would allow practices to monitor patients, especially those with whom they do not have frequent visits, using all available information.

Suggested solution: Faster release of data to participants would enable quality and cost improvements that would benefit practices and Medicare, as well as patients with cancer. As mentioned above, for managing patients’ cases, adjudicated claims are not necessary. The sooner practices have information on their patients’ entire treatment utilization across all facilities, the better they can manage patients’ care.

Conclusion

CMMI has made a huge leap forward in oncology payment reform. Addressing the challenges and issues described in this letter has importance beyond the success of just the OCM, especially as CMMI is considering next-generation oncology models.

We underscore the importance of fixing the four problem areas we have identified in this letter. As a result, we request that CMMI delay the October 2019 decision on two-sided risk until April 2020. We assume you saw the recent Avalere analysis that more than 50% of the OCM participants could end up owing CMS funds under two-sided risk.1 Without addressing fundamental model flaws identified in this letter, and without more accurate, timely data, OCM participants are at significant financial risk under two-sided risk.

We look forward to discussing these issues in greater detail with you and the CMMI team.

Sincerely,

Michael Diaz, MD
President

Ted Okon
Executive Director