Abstract: In order to identify drivers of chronic kidney disease (CKD) stage progression and health care costs associated with progression, we conducted a retrospective, longitudinal observational study querying the Humedica electronic medical records database for adult patients in the United States with new, sustained, or progressive CKD (stage 2, 3a, 3b, or 4/5) with ≥ 1 year pre-index and ≥ 3 years post-index data. Data was analyzed for 212,920 CKD stages among 189,799 patients (52,280 age 65 years; 137,519 age ≥ 65 years). On Poisson regression, diabetes, heart failure, cardiovascular disease, and pre-index hyperkalemia predicted CKD progression. CKD progression rates were highest among patients with stage 3a or 3b disease (38%-43% at 3 years). Progressing patients averaged 12 to 16 months in their index stage. Incremental cost increased with each successive stage (all P .001, except stage 3a vs 2 in Medicare patients) and was higher in Commercial patients vs Medicare. Hyperkalemia and the well-known comorbidities were associated with CKD progression. Incremental costs with CKD progression are substantial, especially in younger patients.
Acknowledgments: Writing and editorial support were provided by Impact Communication Partners Inc., and funded by Relypsa, Inc., a Vifor Pharma Group Company.
Chronic kidney disease (CKD) affects 13.6% of the US adult population, with a projected increase to 16.7% by 2030.1,2 Risk factors for progression of CKD include diabetes, albuminuria, and various cardiovascular comorbidities, which provide the targets for current CKD management.3-6 With CKD progression adverse cardiovascular events and mortality increase.7-11 Importantly, the associations between decreased estimated glomerular filtration rate (eGFR) and risk of major cardiovascular events and death appear independent of concomitant chronic diseases (eg, cardiovascular disease or heart failure).7
Prevalence estimates of CKD, based on cross-sectional analyses,12,13 provide a snapshot with little information about CKD progression over time. Most information on effects of CKD progression on health care costs was also derived from cross-sectional studies. Notwithstanding these limitations, a direct association between progression and cost was demonstrated. In Australia, annual health care costs increased from $1829 for individuals without CKD to $14,545 for those with stage 4/5 CKD.14 In a Japanese cohort, medical expenditures increased with CKD stage.15,16 In our previous study of CKD patients with a prescription history of renin-angiotensin-aldosterone system inhibitor (RAASi) therapy, all-cause costs per patient were exponentially higher at each successive CKD stage.17 We hypothesized that other factors, such as hyperkalemia, may also contribute to cost independently through increased provider-driven hospitalization.
Longitudinal analyses of real-world clinical practice are needed to evaluate the rate of CKD stage progression, explore risk factors driving progression, and assess how CKD progression affects health care costs. Importantly, identifying cost drivers of CKD progression may help uncover cost-reducing strategies. We report a retrospective longitudinal study of a large electronic US medical database evaluating the rates and risk factors for progression by CKD stage over 3 years and impact on health care costs.
Study Population and Cohorts
We queried Humedica (Boston, MA) database electronic medical records (EMR) covering approximately 7 million patients during 2007-2012. Included patients were indexed at first evidence of new, sustained, or progressive CKD (stage 2, 3a, 3b, or 4–5) identified by eGFR or diagnosis code (Supplementary Table 1
), if they had EMR data for ≥ 1 year before and ≥ 3 years after the index event that included ≥ 2 eGFR readings separated by ≥ 90 days. Patients were re-included at each subsequent CKD stage for which an outcome period extended ≥ 3 years from the stage start/index date. Patients with end-stage renal disease (ESRD) at index—ineligible for analysis of primary outcomes—were included for comparisons of patient characteristics and ongoing costs. Exclusions were applied for missing data (Supplementary Figure 1