Unrecognized disease progression is associated with higher health care costs both for patients with end-stage kidney disease and late-stage (stages G4-G5) chronic kidney disease.
Objectives: Many individuals with chronic kidney disease (CKD) are undiagnosed or unaware of the disease and at risk of not receiving services to manage their condition and of “crashing” into dialysis. Past studies report higher health care costs among patients with delayed nephrology care and suboptimal dialysis initiation, but they are limited because they focused on patients undergoing dialysis and did not evaluate costs associated with unrecognized disease for patients “upstream,” or patients with late-stage CKD. We compared costs for patients with unrecognized progression to late-stage (stages G4 and G5) CKD and end-stage kidney disease (ESKD) with costs for individuals with prior CKD recognition.
Study Design: Retrospective study of commercial, Medicare Advantage, and Medicare fee-for-service enrollees 40 years and older.
Methods: Using deidentified claims data, we identified 2 groups of patients with late-stage CKD or ESKD, one group with prior evidence of CKD diagnosis and the other without, and compared total and CKD-related costs in the first year following late-stage diagnosis between the 2 groups. We used generalized linear models to determine the association between prior recognition and costs and used recycled predictions to calculate predicted costs.
Results: Total and CKD-related costs were 26% and 19% higher, respectively, for patients without prior diagnosis compared with those with prior recognition. Total costs were higher both for unrecognized patients with ESKD and unrecognized patients with late-stage disease.
Conclusions: Our findings indicate that costs associated with undiagnosed CKD extend to patients not yet requiring dialysis and highlight potential savings from earlier disease detection and management.
Am J Manag Care. 2023;29(2):e64-e68. https://doi.org/10.37765/ajmc.2023.89323
Based on a large sample of commercial and Medicare enrollees, total all-cause health care costs were higher for patients with late-stage chronic kidney disease (CKD) or end-stage kidney disease who lacked prior diagnosis compared with those with prior recognition.
More than 37 million US adults have chronic kidney disease (CKD), and approximately 800,000 live with end-stage kidney disease (ESKD).1 The majority of individuals with CKD are unaware they have the disease,1 and half or more of adults at increased risk of CKD go untested or undiagnosed,2,3 putting them at risk of receiving inappropriate care, not receiving services and treatments intended to help delay or slow CKD progression, and having little or no nephrologist care prior to dialysis (or “crashing” into dialysis). Suboptimal dialysis initiation (eg, marked by dialysis starts during a hospitalization or without a permanent dialysis access) is not uncommon,4-6 and individuals with such dialysis starts have reduced access to home dialysis and transplantation, poor clinical outcomes, and increased mortality.4,6-8
Several studies have also reported higher health care costs (including higher hospital utilization) among patients with late referral to nephrology and unplanned or emergent dialysis initiation.9-11 These studies, however, are limited because they focused on patients who needed renal replacement therapy (overlooking the potential cost ramifications of undiagnosed or unmanaged CKD for other patients with late-stage CKD), involved small samples, and are dated.
Using a large, deidentified commercial and Medicare claims database, we calculated health care costs for patients with unrecognized progression to late-stage CKD or ESKD immediately before and after late-stage diagnosis and compared them with costs for individuals with late-stage CKD or ESKD who had prior CKD recognition.
Data and Sample
We conducted a retrospective study using 2011-2017 Optum Labs Data Warehouse data.12 Optum Labs Data Warehouse contains deidentified, longitudinal medical and pharmacy claims and enrollment data with linked socioeconomic status and death information for commercial, Medicare Advantage, and Medicare fee-for-service enrollees. Medicare fee-for-service claims files held by Optum Labs, a certified CMS qualified entity, were approved by CMS for reuse. The study obtained an exemption from the New England Institutional Review Board.
The sample included commercial and Medicare enrollees 40 years and older. Two groups were selected: (1) patients with late-stage CKD (stages G4 and G5) or ESKD with prior CKD recognition in the claims data and (2) patients with late-stage CKD or ESKD without prior recognition. Evidence of late-stage CKD/ESKD required at least 1 inpatient claim with a late-stage CKD or ESKD diagnosis (primary position) or 2 claims with a diagnosis (any position) on separate dates at least 30 days apart within a 1-year period. For both groups, the index date was the earliest diagnosis for late-stage CKD/ESKD. Evidence of prior recognition required at least 1 inpatient claim with an early-stage or unspecified CKD diagnosis (primary position) or 2 claims with a diagnosis for early-stage/unspecified CKD (any position) at least 30 days apart in the 12 months before the index date (ie, baseline). If an enrollee had a claim(s) for CKD in the 30 days before the index but no CKD claims earlier during the baseline, the enrollee was considered unrecognized. (See eAppendix Table 1 [eAppendix available at ajmc.com] for diagnosis codes.) Individuals in both groups had continuous medical and pharmacy coverage during the baseline and 12 months following their index date or to death (ie, follow-up). Enrollees with baseline evidence of a kidney transplant, hospice care, or at least 90 days of long-term care were excluded.
We compared total all-cause health care costs and CKD-related costs in the first year following late-stage CKD/ESKD diagnosis between the 2 groups. Total all-cause costs included combined medical and pharmacy costs, and CKD-related costs included medical claims with a CKD/ESKD diagnosis and claims for dialysis or transplantation (eAppendix Table 2). Outcomes included plan- and patient-paid amounts and were adjusted for consumer price index to reflect inflation between 2011 and 2017. We winsorized the cost variables at 99% to reduce the effect of extreme outliers.
Cost outcomes were analyzed in their original form (ie, individuals’ values were not transformed) using generalized linear models with a γ distribution and log link. The primary covariate of interest was prior recognition status. Others included patients’ sex, age, race/ethnicity, Census region, type of health insurance at index, and index year. Community-level demographic data included American Community Survey 5-year estimates for 2014-2018 (English proficiency and poverty). These variables are specific to a patient’s zip code tabulation area and were translated to national deciles. A variable summarizing a patient’s zip code tabulation area as mostly urban, mostly rural, or completely rural (2010 decennial Census indicator) was also included. Binary flags for baseline cardiovascular disease, cerebrovascular disease, peripheral artery disease, diabetes, hypertension, and obesity were included based on the Agency for Healthcare Research and Quality Clinical Classifications Software.13 Finally, the analysis included covariates indicating whether a patient had an emergency department visit or inpatient stay during the baseline. Using recycled predictions, we calculated point estimates of mean predicted costs by prior recognition status for the overall sample and by ESKD status at index. Based on the estimated model, the recycled predictions method treats sample members as if they are in the same group of interest (ie, prior recognition status) and calculates the mean predicted values for each group value while holding constant all other model covariates.14 This yields predicted values as if each group had the same underlying distribution of covariates. The analysis was stratified by ESKD status at index date because descriptive analyses showed that more individuals in the unrecognized group than in the recognized group (21.4% vs 10.7%; P < .001) (data not shown in tables) had an ESKD diagnosis instead of a diagnosis of stage G4 or G5 at index. Analyses were conducted in SAS version 9.4 (SAS Institute Inc) using the PROC GENMOD procedure.
Table 1 presents sample members’ baseline characteristics and costs. Slightly fewer patients in the unrecognized group were male, in the oldest age group, and living in the Midwest. As expected, patients with prior CKD recognition had higher baseline health care utilization and costs on average.
Unadjusted mean total all-cause and CKD-related costs in the follow-up period are shown in the first set of columns in Table 2, overall and by ESKD status at index. In contrast to baseline costs, unadjusted total costs in the 1-year follow-up period were higher overall among patients without prior recognition. This was true for total medical costs and CKD-related costs, although total pharmacy costs were lower for unrecognized patients. When stratified by ESKD status at index, the higher unadjusted mean total costs among unrecognized patients held only for individuals with ESKD. Unadjusted mean CKD-related costs in the follow-up period were lower for unrecognized patients regardless of ESKD status at index.
Table 2 also presents the results of the multivariate generalized linear models for total and CKD-related costs for the overall sample and stratified by ESKD status at index. The table shows the exponentiated cost ratios and CIs for the primary covariate of interest, prior recognition status (reference group: patients with prior recognition). (The full model results are available in eAppendix Tables 3 and 4.) After controlling for the other covariates, total all-cause costs as well as CKD-related costs in the 1-year follow-up were higher for patients without prior recognition. Specifically, their total costs were 26% higher, and CKD-related costs were approximately 19% higher than those of patients with prior recognition. A breakdown of total costs by cost category reveals that the higher costs observed among the unrecognized group were driven by medical costs, with pharmacy costs in the follow-up period lower among patients without prior recognition. Based on these models, mean predicted total costs in the first year following late-stage or ESKD diagnosis amounted to $69,188.89 for patients without prior recognition and $54,993.10 for those with prior recognition. Predicted CKD-related costs totaled $34,617.86 for the unrecognized group compared with $29,219.55 for those with prior recognition (see last set of columns in Table 2).
Similar findings are observed for sample members whose initial late-stage diagnosis was stage G4 or G5, although the cost ratios for this subgroup were smaller. Among these individuals, unrecognized patients had 13% higher total costs and 7% higher CKD-related costs compared with patients with prior recognition after controlling for the other covariates. For patients with ESKD at index, total costs were also higher among individuals without prior recognition, but CKD-related costs were slightly lower in the follow-up period. Table 2 presents the predicted follow-up costs for the 2 subgroups.
Leveraging a unique deidentified claims database and a large sample of commercial and Medicare enrollees, we compared health care costs for patients with unrecognized progression to late-stage CKD or ESKD with costs for other individuals with late-stage CKD or ESKD who had prior disease recognition. Total all-cause costs, total medical costs, and CKD-related costs in the year following late-stage diagnosis were higher for patients without prior recognition. One factor contributing to higher overall costs is increased risk of inpatient health care utilization. In a separate analysis, we observed higher odds of incurring inpatient care costs during the follow-up period for our sample without prior recognition compared with individuals with prior recognition (odds ratio [OR], 1.52; 95% CI, 1.50-1.55), and we observed this for patients with ESKD at index (OR, 1.62; 95% CI, 1.54-1.71) and for those whose initial late-stage diagnosis was stage G4 or G5 (OR, 1.35; 95% CI, 1.33-1.37). Our findings for the subset of patients with ESKD reinforce past studies that compared costs for patients receiving dialysis with late referral to nephrologist care or unplanned dialysis starts with those for patients with early referral and optimal dialysis initiation.9-11 Similar to these studies, our study found that patients with ESKD without prior CKD recognition had higher overall costs in the first year following ESKD diagnosis compared with those with prior recognition, although CKD-related costs were slightly lower for those without prior recognition.
Past studies evaluating the costs of late nephrology referral and suboptimal dialysis starts have focused on patients receiving dialysis and did not evaluate the costs associated with unrecognized disease progression for other patients with late-stage CKD. Our analysis builds on prior research, showing higher overall costs among not only patients with ESKD without prior recognition, but also patients with late-stage kidney disease (stages G4 and G5) who lacked prior diagnosis.
Our study has certain limitations. First, claims data are created for billing purposes and lack detailed clinical information. Our definitions of CKD and unrecognized progression were based on diagnosis codes used in claims processing and therefore may have misclassified individuals. Furthermore, to the extent that the coding of CKD staging was based on estimated glomerular filtration rate results, patients’ stage may have been poorly estimated due to the inclusion of race within estimated glomerular filtration rate calculations.15 To offset these concerns about diagnosis codes, we used a nuanced CKD definition, accounting for care setting (inpatient vs outpatient), claim type (eg, excluding diagnostic procedures), the number of diagnoses, and time between codes. Prior research on claims-based CKD algorithms has shown high specificity and reasonable predictive value but mixed levels of sensitivity.16-18 Our measurement of unrecognized progression was limited to the 12 months before late-stage diagnosis. With a longer look back, some unrecognized individuals may have instead been classified as recognized, although it is noteworthy that these individuals had no or little health care for CKD immediately before late-stage diagnosis. In addition, our definition of CKD-related costs, which focused on claims with diagnosis codes for CKD and claims for dialysis and transplantation, likely excluded other claims relevant to CKD (eg, medications to delay or slow progression) and may have underestimated the costs associated with CKD. Finally, our findings may not extend to individuals with other health insurance coverage (eg, Medicaid) or uninsured individuals.
We find that the higher costs associated with undiagnosed CKD not only pertain to patients with ESKD (similar to prior research on patients who crash into dialysis) but also extend to patients upstream who do not yet require renal replacement therapy. Our findings highlight potential savings from earlier disease detection and management, providing additional support for current American Diabetes Association19 and Kidney Disease: Improving Global Outcomes20 recommendations to screen and monitor patients at high risk for CKD, as well as for health system and payer approaches to increase awareness among patients and providers about CKD and its progression, engage patients earlier in disease management programs, and improve care coordination across primary and specialty care.
The authors wish to acknowledge David Mosley, PhD, for his input on the analysis; Lillian Hang, MBA, MPH, and Pamela Morin, MBA, for their programming support; and Molly Diethelm, PMP, for her project management support in this work.
Author Affiliations: Optum Labs (DS, SD, JM, CH), Minnetonka, MN; National Kidney Foundation (JSCR), New York, NY.
Source of Funding: This research was funded by the National Kidney Foundation and Optum (Optum Kidney Services).
Author Disclosures: At the time this research was conducted, Drs Spencer and Hane, Mr Dunning, and Mr McPheeters were employed by Optum Labs (under Optum/UnitedHealth Group), and Drs Spencer and Hane own UnitedHealth Group stock through an employee purchase program. This research was funded in part by Optum Kidney Services, which is part of UnitedHealth Group. While Optum Kidney Services and UnitedHealth Group may, like all providers and payers, benefit indirectly from the findings, the staff of Optum Kidney Services and the larger enterprise had no role in the collection, analysis, or interpretation of the data; the writing of the article; or the decision to submit the article for publication. Any views conveyed in the publication represent the opinions of the authors. Dr St. Clair Russell reports no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
Authorship Information: Concept and design (DS, SD, JM, JSCR, CH); acquisition of data (SD, JM); analysis and interpretation of data (DS, SD, JM, JSCR, CH); drafting of the manuscript (DS, JSCR, CH); critical revision of the manuscript for important intellectual content (DS, SD, JSCR, CH); statistical analysis (JM); obtaining funding (SD); and administrative, technical, or logistic support (DS, JSCR).
Address Correspondence to: Donna Spencer, PhD, RTI International, 3040 E Cornwallis Rd, Research Triangle Park, NC 27709. Email: email@example.com.
1. Chronic kidney disease in the United States, 2021. CDC. March 4, 2021. Updated July 12, 2022. Accessed October 22, 2021. https://www.cdc.gov/kidneydisease/publications-resources/ckd-national-facts.html
2. United States Renal Data System. US Renal Data System 2019 Annual Data Report: Epidemiology of Kidney Disease in the United States. National Institute of Diabetes and Digestive and Kidney Diseases; 2019. Accessed October 22, 2021. https://www.niddk.nih.gov/-/media/Files/USRDS/2019/2019-executive-summary.pdf
3. Szczech LA, Stewart RC, Su HL, et al. Primary care detection of chronic kidney disease in adults with type-2 diabetes: the ADD-CKD study (awareness, detection and drug therapy in type 2 diabetes and chronic kidney disease). PLoS One. 2014;9(11):e110535. doi:10.1371/journal.pone.0110535
4. Smart NA, Dieberg G, Ladhani M, Titus T. Early referral to specialist nephrology services for preventing the progression to end-stage kidney disease. Cochrane Database Syst Rev. 2014;(6):CD007333. doi:10.1002/14651858.CD007333.pub2
5. Arulkumaran N, Navaratnarajhah A, Pillay C, et al. Causes and risk factors for acute dialysis initiation among patients with end-stage kidney disease—a large retrospective observational cohort study. Clin Kidney J. 2019;12(4):550-558. doi:10.1093/ckj/sfy118
6. Chiu K, Alam A, Iqbal S. Predictors of suboptimal and crash initiation of dialysis at two tertiary care centers. Hemodial Int. 2012;16(suppl 1):S39-S46. doi:10.1111/j.1542-4758.2012.00744.x
7. Hassan R, Akbari A, Brown PA, Hiremath S, Brimble KS, Molnar AO. Risk factors for unplanned dialysis initiation: a systematic review of the literature. Can J Kidney Health Dis. 2019;6:2054358119831684. doi:10.1177/2054358119831684
8. Górriz JL, Sancho A, Pallardó LM, et al. Significado pronóstico de la diálisis programada en pacientes que inician tratamiento sustitutivo renal. un estudio multicéntrico español. Abstract in English. Nefrologia. 2002;22(1):49-59.
9. McLaughlin K, Manns B, Culleton B, Donaldson C, Taub K. An economic evaluation of early versus late referral of patients with progressive renal deficiency. Am J Kidney Dis. 2001;38(5):1122-1128. doi:10.1053/ajkd.2001.28619
10. Sprangers B, Evenepoel P, Vanrenterghem Y. Late referral of patients with chronic kidney disease: no time to waste. Mayo Clin Proc. 2006;81(11):1487-1494. doi:10.4065/81.11.1487
11. Shimizu Y, Nakata J, Yanagisawa N, et al. Emergent initiation of dialysis is related to an increase in both mortality and medical costs. Sci Rep. 2020;10(1):19638. doi:10.1038/s41598-020-76765-0
12. Wallace PJ, Shah ND, Dennen T, Bleicher PA, Crown WH. Optum Labs: building a novel node in the learning health care system. Health Aff (Millwood). 2014;33(7):1187-1194. doi:10.1377/hlthaff.2014.0038
13. Clinical Classifications Software (CCS) for ICD-9-CM. Agency for Healthcare Research and Quality. March 2017. Accessed October 22, 2021. https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp
14. Graubard BI, Korn EL. Predictive margins with survey data. Biometrics. 1999;55(2):652-659. doi:10.1111/j.0006-341x.1999.00652.x
15. Delgado C, Baweja M, Crews DC, et al. A unifying approach for GFR estimation: recommendations of the NKF-ASN Task Force on reassessing the inclusion of race in diagnosing kidney disease. Am J Kidney Dis.2021;79(2):268-288.e1. doi:10.1053/j.ajkd.2021.08.003
16. Grams ME, Plantinga LC, Hedgeman E, et al; CDC CKD Surveillance Team. Validation of CKD and related conditions in existing data sets: a systematic review. Am J Kidney Dis. 2011;57(1):44-54. doi:10.1053/j.ajkd.2010.05.013
17. Grabner M, Chen Y, Palli SR, Quimbo RA. PUK34 validation of ICD-9 codes for the identification of patients with stage 3-5 chronic kidney disease in administrative claims data. Value Health. 2012;15(4):PA158. doi:10.1016/j.jval.2012.03.854
18. Muntner P, Gutiérrez OM, Zhao H, et al. Validation study of Medicare claims to identify older US adults with CKD using the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study. Am J Kidney Dis. 2015;65(2):249-258. doi:10.1053/j.ajkd.2014.07.012
19. American Diabetes Association. 11. Microvascular complications and footcare: Standards of Medical Care in Diabetes–2019. Diabetes Care. 2019;42(suppl 1):S124-S138. doi:10.2337/dc19-S011
20. CKD evaluation and management. Kidney Disease: Improving Global Outcomes. Accessed May 4, 2021. https://kdigo.org/guidelines/ckd-evaluation-and-management/