Need to Reclassify Etiologies of ESRD on the CMS 2728 Medical Evidence Report

Introduction

The Centers for Medicare and Medicaid Services (CMS) 2728 Medical Evidence Report (2728 form) serves multiple purposes, including Medicare entitlement and registry, as well as documentation of the presence and presumed cause of ESRD. Dialysis facilities complete this form in newly diagnosed patients with ESRD within 45 days of initiation through the CMS’s CROWNWeb system. Guidance and oversight are provided by the regional ESRD Networks. These data are used to justify payment for RRT.

The legislative origin of the Medical Evidence Report began in 1972 when President Nixon signed the Social Security Amendments into law, making ESRD a health care condition covered by Medicare for individuals regardless of age and without other disabilities. The CMS 2728 form was developed to capture these data in 1973 and then revised in 1987, 1995, and 2006 (1). Although accuracy improved with the 1995 revision, substantial inaccuracy remains (2). The CMS shares CROWNWeb data with the US Renal Data System (USRDS) (3). The USRDS collects additional information from the Organ Procurement and Transplantation Network, the Social Security Administration (mortality), and Medicare (hospital payment records) and then reports trends in the costs, causes, and outcomes of ESRD (1). This information is used by policymakers and researchers. The USRDS data remain critically important to document trends in the development of ESRD and age-, sex-, and race-specific data. A search of the term “US Renal Data System” in PubMed identified >4000 manuscripts, showing broad dissemination.

Many who use these data fail to appreciate the lack of accuracy in the 2728 form “Cause of ESRD” field. Available choices in this field are diagnoses on the basis of the International Classification of Diseases 10th Revision Clinical Modification (ICD-10-CM) coding. Although 14 etiologic categories of CKD are provided, three broad diagnoses reportedly account for approximately 80% of patients with incident cases of ESRD: diabetes mellitus, hypertension, and chronic glomerular disease. These forms are signed by a nephrologist. However, nonmedical surrogates often provide clinical information. Herein, we make a case for simplification and standardization of reported causes of ESRD on the 2728 form to improve accuracy. We also suggest periodic reassessments of these etiologies to keep pace with scientific developments.

Limitations of ESRD Classification on the Current CMS 2728 Form

More than 8000 United States nephrologists currently provide information on the 2728 form regarding suspected causes of ESRD in incident patients. Standardized guidelines are lacking. Contributors to inaccuracy include physician bias introduced by patient race and frequent lack of renal histology (4,5). Kidney biopsies are typically not performed in patients with longstanding diabetes or individuals without diabetes who lack heavy proteinuria. Surprisingly, even when kidney biopsies are performed, incorrect diagnoses may still be listed (2). This may relate to physicians’ lack of awareness that biopsy data exist or that nonmedical personnel completed the form.

To show the difficulty in selecting a diagnostic category, consider the following clinical presentation. A patient presents to an emergency department with malaise after not having seen a physician for years. Physical examination reveals only elevated BP, and laboratory work shows markedly reduced kidney function with anemia, acidosis, and hyperkalemia. These patients often require emergent dialysis, and subsequent imaging reveals small kidneys supporting chronicity. How should this disease be classified on the 2728 form? It likely developed over years to decades. Renal sclerosis, unspecified (N26.9) might be most appropriate. However, physicians often report specific causes, such as hypertension (I12.9), after a cursory evaluation. Secondary hypertension is a common consequence of longstanding kidney disease. Layton et al. (2) further reported that causes of kidney failure were missing in up to 57% of patients.

Given recent advances, perhaps future versions of the 2728 form should precisely classify known etiologies of nephropathy but report phenotypic patterns in those with less certain causes. The 2014 USRDS Annual Report lists hypertension as the primary cause of ESRD in 28.6% of incident patients with ESRD; almost 40% of blacks were given this diagnosis (3). However, two definitive National Institutes of Health–sponsored trials in participants with CKD whose clinical features were initially thought to reflect hypertensive nephropathy revealed that mild to moderate hypertension was unlikely to have initiated kidney disease (6,7). Identification of the apolipoprotein L1 gene (APOL1)–associated spectrum of nondiabetic nephropathy in patients with recent African ancestry supports the need to remove “hypertension” as the cause of ESRD in those with mild to moderately elevated BPs. Detection of genes underlying steroid-resistant nephrotic syndrome and autosomal dominant tubulointerstitial forms of nephropathy further supports the need to update the nomenclature, particularly after genetic testing becomes widely available (8,9).

Patients with diabetes and advancing nephropathy are frequently labeled with diabetic kidney disease (DKD), despite awareness that they may have diabetes coincident with another etiology of kidney disease (10). Many patients with progressive DKD manifest heavy proteinuria, whereas others have early declines in eGFR without proteinuria. It is possible that these presentations result from a different pathogenesis, and perhaps, a revised descriptive classification system should reflect this.

Future Opportunities

We propose that the CMS consider convening a Technical Expert Panel or team of nephrologists and renal pathologists to update etiologic categories of kidney disease on the 2728 form on a regular basis. Verification can be considered on the basis of chart reviews in a random sample during review cycles. Ideally, a revised form could include simple descriptive categories when specific diagnoses are unclear. A checkbox should be added to indicate whether a kidney biopsy has been performed. Patients can inform their current nephrologist if a biopsy was previously done, and this would prompt a search for the precise histologic diagnosis. Absent a clear clinical scenario, kidney biopsy, presence of a known genetic kidney disease, or Mendelian form of polycystic kidney disease on the basis of imaging (where precise diagnostic categories would remain as currently listed), we propose using broad phenotypic categories. In this fashion, homogeneous subsets of patients would be grouped to permit genetic and epidemiologic analyses, providing more descriptive classifications without excess work for nephrologists on the basis of similar numbers of classifications.

Instead of imprecise (hypertension) or nonspecific (GN) terminology for ESRD, we propose categorizing patients without diabetes and known causes of CKD as “nondiabetic with proteinuria <3 g/d” or “nondiabetic with proteinuria >3 g/d.” Related glomerular diseases may present with or without heavy proteinuria. For example, APOL1-associated kidney disease can present as focal global glomerulosclerosis or subnephrotic FSGS in the low-proteinuria category and as FSGS with nephrotic-range proteinuria in the high-proteinuria category. Similar classifications could be applied to patients with diabetes: “diabetes with proteinuria >3 g/d” and “diabetes with proteinuria <3 g/d.” In individuals whose diabetes developed before nephropathy, it would be helpful to include diabetes duration before ESRD in Section 17 (“Co-Morbid Conditions”) of the 2728 form. Shorter diabetes durations before onset of ESRD and lack of nonrenal microvascular complications (included in Section 17) suggest higher likelihoods of nondiabetes-associated ESRD (10).

Modern medicine and the electronic medical record strive for precise disease categorization on the basis of ICD-10-CM nomenclature. This leads to great advances when diagnoses are certain. Forcing poorly specified diseases into ICD-10-CM–type nomenclature has had adverse effects on the epidemiology of CKD. After “hypertension I12.9” and “diabetes E11.22” are published in large databases for 75% of incident patients with ESRD, it is difficult to convince researchers and clinicians that elevated BP and blood sugar may not have been causative in a significant percentage of patients. However, this is the current reality.

There would be limitations to changing the current CMS 2728 ESRD classification. New diagnostic categories might make it more difficult to determine trends in incidence rates when contrasting them with prior categories. However, the proposed category “nondiabetes-associated ESRD” (with or without proteinuria >3 g/d) would be roughly equivalent to the current “hypertension” combined with “GN”-associated kidney disease categories. Similarly, the proposed diabetes-associated ESRD category on the basis of severity of proteinuria would still equate to the existing percentage reported with DKD. Finally, we recognize that it is unlikely that kidney biopsy data will become available in a large percentage of incident patients with ESRD. Nonetheless, precise diagnoses in the subset of patients with renal histology remain critical. Selecting “Nephrotic syndrome with focal and segmental glomerular lesions” (N04.1) is far more meaningful in patients with biopsy evidence than in those with clinically presumed diagnoses.

This proposal for simplifying and providing more descriptive categorization of etiologies of ESRD on the CMS 2728 form is not meant to be the final word. We view it as opening a dialogue to advance renal epidemiology and improve diagnostic accuracy. If the CMS 2728 form is simply an instrument for defining appropriate payment for ESRD-related services, there is no need for revision. However, we believe these widely disseminated data are used by researchers studying kidney disease to devise new therapies. As such, the present classification has major deficiencies. When genetic testing becomes common in the near future, the CMS 2728 form should list replicated genetic causes of ESRD. We envision a day when codes denoting APOL1-, UMOD-, and NPHS2-associated ESRD exist. This is the dream of precision medicine and our best hope for curing kidney disease.

Disclosures

Wake Forest University Health Sciences (Winston-Salem, NC) and B.I.F. filed for a patent related to APOL1 gene testing. B.I.F. is a consultant for Ionis and AstraZeneca Pharmaceuticals.