Risk Standardized Mortality Ratio for Late-Stage Chronic Kidney Disease (CKD) and End Stage Renal Disease (ESRD) (3754)

Bayer US (“Bayer”) appreciates the opportunity to submit comments pertaining to the proposed endorsement of this measure. 

Bayer is a global enterprise with core competencies in the Life Science fields of health care and agriculture with nearly 25,000 employees in 300 sites across the United States. Our products and services are designed to benefit people and improve their quality of life. At the same time, we aim to create value through innovation and are committed to the principles of sustainable development and to our social and ethical responsibilities as a corporate citizen.

Bayer is committed to improving patient outcomes and advancing health by closing gaps in care. We recognize the importance of value-based care and quality measurement and are pleased to submit feedback in response to Battelle’s open comment period on the Spring 2023 measure endorsement cycle.

Bayer supports the endorsement of this measure to drive improvement in CKD outcomes. We emphasize the importance of focusing on assessing care for this population as CKD affects about 37 million Americans and 2 in 1,000 Americans are living with end-stage kidney disease.¹ 

Given there are disparities in CKD and ESRD care, with higher prevalence and severity of both conditions found in Black and Hispanic American populations, we encourage measure stratification by race/ethnicity to identify and close disparities in care. ¹

¹ NIH. Kidney Disease Statistics for the United States. https://www.niddk.nih.gov/health-information/health-statistics/kidney-disease/ Accessed June 1, 2023.

Dear Colleagues:

The National Forum of ESRD Networks (“The Forum” at www.esrdnetworks.org) appreciates this opportunity to offer comments regarding several quality metrics for chronic kidney disease (CKD) and end-stage renal disease (ESRD) that the Centers for Medicare and Medicaid Services (CMS) has offered for review. The Forum noted the current placement of CKD quality metrics under the “Primary Care and Chronic Illness Standing Committee” and hopes that in the future there is the opportunity to utilize a broader mandate under CKD recognizing the patient-centered implications of a continuum of care from CKD to ESRD. The Forum would also very much like to be more proactively engaged in stages earlier than the public commentary period given the representation of all Congressionally mandated and CMS contracted ESRD Networks. Given the limited time available to the Forum to provide comment on the current quality metrics, we would nevertheless like to offer some abbreviated thoughts regarding the following metrics:

• Delay in Progression of CKD Measure (CBE ID 3753): Whereas the Forum remains fully supportive of efforts to slow the progression of CKD and value-based payment models that encourage such process improvement, we are concerned about several areas of the proposed streamlined metric to include the limited exclusionary criteria of advanced or metastatic cancer. The Forum feels that there are several other acute and chronic disease states associated with limited life expectancy that should be accounted for in the denominator and would welcome the opportunity to make further recommendations. Furthermore, notwithstanding the separate mortality metric (see below), death as the competing event for CKD progression to ESRD has not been accounted for in this metric, so that death events may appear to halt progression to ESRD, leading to misleadingly favorable scores. Lastly, the Forum would also like to recognize the fact that delaying CKD progression measures that begin in the latter stages (i.e., stage 4) of CKD limits a provider’s ability to make meaningful change in the trajectory of the patient’s illness. The Forum therefore recommends initiating such quality metrics at earlier CKD stages if possible.
• Risk Standardized Mortality Ratio for Late-Stage CKD and ESRD (CBE ID 3754): The Forum again voices its appreciation regarding efforts to improve the outcomes of persons with CKD; however, the Forum has concerns regarding individual providers’ abilities to have consistent and direct impacts on all-cause mortality. Quality measures that assess outcomes outside of the nephrologist’s control may lead to ineffective strategies for efforts that are unlikely to succeed. The Forum would welcome the opportunity to discuss efforts more directly under the kidney care community’s control and place a focus of incentivizing improvements in these critically important areas.
• ESRD Dialysis Patient Life Goals Survey (PaLS) (CBE ID 3742): The Forum is supportive of efforts to better understand life goals of persons with CKD as well as efforts to better focus on person-centered outcomes with the ultimate goal of living well with kidney disease. While we do support the premise of this quality measure, the Forum has concerns regarding dialysis facility staffing shortages (and the increased burden on limited staff to obtain such important information), as well as the willingness of patients to participate in more surveys without clear indication of how it might benefit them. 

The Forum would once again like to thank CMS and PQM for the opportunity to comment on the above proposed quality measurements and offer its willingness to become more actively engaged in future discussions. 

Sincerely,

David Henner, DO - President, Forum of ESRD Networks

Kam Kalantar-Zadeh, MD, MPH, PhD - President-Elect, Forum of ESRD Networks

Daniel Landry, DO - Chair, Forum Medical Advisory Council

Preethi Yerram, MD -  Vice Chair, Forum Medical Advisory Council

RISK-STANDARDIZED MORTALITY RATIO FOR LATE-STAGE CKD AND ESRD (#3754)

KCP believes mortality is an important outcome to measure, but has a number of concerns with the  RSMR, detailed below.

Validity 

• Identification of Stage 4 and 5 CKD Cases:  While there was an acceptably high match rate on data element validation, ostensibly suggesting  the use of claims to identify denominator cases can achieve the intended measure cohort, we note again here that inclusion in the denominator only requires a single Stage 4 (N18.4) or Stage 5 (N18.5) code in a single claim during the measurement year.  Given the reality that patients may move between CKD stages, this construct may lead to the inadvertent inclusion of Stage 3 patients in the denominator.  To remedy this issue, KCP recommends the measure should require more than one Stage 4 or 5 code documented during the year.  Ideally, the appropriate number of codes required would be determined empirically. 
• Lack of eGFR and Albuminuria Laboratory Data:  While Face Validity was apparently not assessed with the RSMR, as with Measure #3753, KCP again has concerns about potential cohort misidentification in the absence of eGFR and albuminuria clinical laboratory data. Again, laboratory results are necessary to identify progression of disease and to provide precision in identifying the appropriate patient risk profile for the measure outcome. 
• The international Kidney Disease Improving Global Outcomes (KDIGO) CKD Evaluation and Management Guideline, as well as the US-specific Kidney Disease Outcomes Quality Initiative (KDOQI) commentary on the CKD guideline both emphasize using eGFR and albuminuria in concert to assess risk and group individuals into risk categories for prognostication and treatment.  Current risk equations for kidney failure, such as the Kidney Failure Risk Equation (KFRE) and the CKD Prognosis Consortium equation rely heavily on these two kidney disease markers for risk stratifying among individuals with CKD, achieving c-statistics of ~0.90 in multiple populations worldwide with an equation including only age, sex, eGFR and albuminuria.[1]  Highlighting how impactful eGFR and albuminuria are for risk prediction, for a 70 year-old woman, the 2-year risk of ESRD with an eGFR of 28 mL/min/1.73m² and urine albumin to creatinine ratio (UACR) of 10 mg/g is 1.5% when using the KFRE. I n contrast, if the eGFR is 17 mL/min/1.73 m² and UACR is 750 mg/g, the 2-year risk of ESRD is 31%.  Both of these individuals have CKD Stage 4; however, the risk of ESRD is 20-fold higher in the second individual.  This is recognized in clinical practice, where the first individual would be designated as CKD Stage G4A1 and the second as CKD Stage G4A3, clearly highlighting different risk.  
• The current measure as proposed also may worsen disparities. There are considerable differences in the prevalence of albuminuria by race/ethnicity, with a recent report from the Centers for Disease Control and Prevention Chronic Kidney Disease Surveillance Team demonstrating that more than 13% of non-White individuals with CKD have severely elevated albuminuria (>300 mg/g), compared to 8% of White individuals.  In this report, there was better performance on many CKD care process measures among non-White patients than among White patients, contrasting with the known higher risk of kidney failure among non-White patients.[2]  This finding emphasizes that improving care processes alone may be inadequate for reducing disparities.  Critically, it also highlights the potentially substantial risk differences by ethnicity that need to be accounted for in an equitable quality program, including by incorporation of both albuminuria and eGFR into risk adjustment.
• Nephrology clinicians recognize and rely on eGFR and albuminuria to risk stratify and guide treatment decisions, including planning for kidney failure.  While we acknowledge that eGFR and albuminuria data are not currently available in such a way that would enable seamless incorporation into a risk adjustment process for a quality metric, implementing any CKD Progression metric without these data and without recognizing the heterogeneity within a CKD stage lacks face validity and is discordant with clinical practice. 
• This issue is of particular concern in smaller provider groups, wherein there are not a sufficient number of patients to smooth the impact of discrepant claims and clinical data. As with Measure 3753, because of this issue, we cannot support the measure until there is improvement in Electronic Health Record (EHR) data and coding practices to allow for improved case identification.

Exclusions 

• Cancer Diagnosis Exclusions: The measure excludes patients with metastatic and advanced cancers; however, cases are identified exclusively from ICD-10 codes from an inpatient encounter.  As such, the measure may not capture all appropriate exclusions, given that some such patients may be managed exclusively in the outpatient setting. KCP thus recommends expanding the exclusion definition to capture both inpatient and outpatient ICD-10 codes for metastatic and advanced cancers.
• Palliative Care:  While hospice enrollment is a measure exclusion, this does not account for the significant number of patients who opt out of dialysis in favor of palliative care.  We recognize there is currently no sufficiently valid means of reliably capturing this data point, but we note that patients opting for this treatment approach will likely increase in the coming years—particularly with an increased emphasis on patients’ life goals, as is the intent of the next measure.  As such, we can expect that this issue will increasingly contribute to an inaccurate cohort capture over time.  More importantly, the failure to account for patient choice to value quality over longevity contrasts with the goal of a patient-centered quality metric by incentivizing more intensive care over goal-concordant care.  A metric spanning advanced CKD and dialysis needs to incorporate patient choice for comprehensive non-dialysis, non-hospice medical care.  Again, eventual improvement in EHR data and coding practices would presumably allow for improved case identification in this regard, as well.

Reliability by Provider Size

Overall reliability was sufficient at 0.623, but results for small practices (<25 cases) were again not provided.  Given the drop in reliability from 0.742 when small providers were excluded, it is likely that the reliability for these small groups—nearly 20 percent of all nephrology groups—does not meet the established minimum standard of 0.6.  As illustrated above, prior trends with other CMS risk-standardized ratio measures support this supposition.  Without evidence to the contrary, KCP is concerned that reliability for Measure 3754 is similarly lower for small groups, effectively rendering the metric meaningless for use in performance measurement in this substantial subset of providers.  KCP believes it is incumbent on CMS to demonstrate reliability for all providers by stratifying data by practice size, and we recommend a small group exclusion be incorporated if reliability is in fact inadequate among such providers.

Social Risk Variables

CMS found no statistically significant relationship between dual eligibility, low AHRQ SES, race, and urbanicity and measure scores among nephrology practices with the highest proportion of patients with these social risk factors.  Because of the minimal impact on provider scores and the risk of masking differential care for patients, social risk factors were not included in the measure’s final risk model.  KCP shares CMS’s concern that risk adjustment can mask real differences in care based on sociodemographic variables; however, we believe that stratification can be an appropriate alternative approach to social risk in some measures.  Given well-established variations in mortality rates across sociodemographic groups,[3] we urge CMS to consider measure stratification by known social risk factors to allow providers and other healthcare stakeholders to identify differences in care and outcomes and to develop and implement equity-focused practices to better address disparities. 

All-Cause Construct

Finally, KCP strongly objects to the all-cause construct of the RSMR, believing it is too expansive in scope and will unfairly penalize clinicians and groups for outcomes beyond their control or sphere of influence.  We note that the corollary facility-level mortality measure specifically excludes deaths due to street drugs or accidents unrelated to treatment; we urge CMS to revise Measure 3754 to incorporate these same numerator case exclusions. 

Attribution 

• Late Referrals:  We note that there is oftentimes a lack of appropriate care coordination between providers, such that many patients are not referred for nephrology care in a timely manner.  Attribution of patient mortality to the nephrologist in such instances does not provide an accurate representation of the care provided by that physician/group and is both inappropriate and unfair.  This concern could be remedied by requiring more patient nephrology encounters than the currently proposed two, although the “appropriate” number has not been identified and would need to be clearly and transparently delineated. 
• Nephrology Group Identification:  Additional detail is needed on how a nephrology group is identified for the purposes of this measure.  While submitted materials indicate measured entities are any clinician group billing for nephrology services to Medicare FFS patients, grouped by taxpayer identification number (TIN), we note that many nephrologists/groups provide both Internal Medicine and Nephrology care.  The point at which a patient receiving care from such a provider transitions from primary care to specialized care is not always clear, which may further compromise accurate cohort identification.  

[1] Tangri N, Grams ME, Levey AS et al.  CKD Prognosis Consortium.  Multinational assessment of accuracy of equations for predicting risk of kidney failure: A meta-analysis.  JAMA.  2016 Jan 12;315(2):164-74. doi: 10.1001/jama.2015.18202. Erratum in: JAMA. 2016 Feb 23;315(8):822. PMID: 26757465; PMCID: PMC4752167.

[2] Chu CD, Powe NR, McCulloch CE et al.  Centers for Disease Control and Prevention Chronic Kidney Disease Surveillance Team. Trends in chronic kidney disease care in the US by race and ethnicity, 2012-2019.  JAMA Netw Open.  2021 Sep 1;4(9):e2127014.  doi: 10.1001/jamanetworkopen.2021.27014. PMID: 34570204; PMCID: PMC8477264.

[3] United States Renal Data System.  2022 USRDS Annual Data Report: Epidemiology of Kidney Disease in the United States.  National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD, 2021.  (See Figure 6.1b.)

The Renal Physicians Association (RPA) is the professional organization of nephrologists whose goals are to ensure optimal care under the highest standards of medical practice for patients with kidney disease and related disorders. RPA acts as the national representative for physicians engaged in the study and management of patients with kidney disease. 

Specifications

RPA is concerned that the use of a single ICD-10 code (N18.4 or N18.5) for inclusion in the denominator does not reflect the reality that patients may move between Stage 3 and 4. RPA believes the measure should require more than one ICD-10 code during the performance year to be included in the denominator.

Further, RPA has concerns about potential cohort misidentification in the absence of eGFR and albuminuria clinical laboratory data. Laboratory results are necessary to identify progression of disease and to provide precision in identifying the appropriate patient risk profile for the measure outcome. This issue is of particular concern in smaller provider groups, wherein there are not a sufficient number of patients to smooth the impact of discrepant claims and clinical data.

Exclusions

RPA objects to the all-cause construct of the RSMR, believing it is too expansive in scope and will unfairly penalize clinicians and groups for outcomes beyond their control or sphere of influence. We note that the corollary facility-level mortality measure specifically excludes deaths due to street drugs or accidents unrelated to treatment; we urge CMS to revise Measure 3754 to incorporate these same numerator case exclusions. 

Additionally, RPA believes that the measure exclusions should be expanded to exclude all patients for whom dialysis is ultimately not the goal of care based on shared decision-making between the clinician and patient. As written, this measure penalizes clinicians for having open conversations with their patients and supporting their patients who make the difficult decision to not pursue dialysis. Many patients want palliative care but refuse hospice because it requires them to stop dialysis or ESAs or office visits with the nephrologist. Therefore, the measure should be expanded to include those patients who choose palliative care.

Finally, while RPA supports the measure exclusion for patients with metastatic and advanced cancers, the requirement of specific cancer-related ICD-10 codes from an inpatient encounter do not reflect the realities of care, which may be managed exclusively as outpatient care so these codes would not be submitted. RPA recommends this be expanded to include both inpatient and outpatient ICD-10 codes for metastatic and advanced cancers.

See attachment