Analysis of Specialty Nephrology Care Among Patients With Chronic Kidney Disease and High Risk of Disease Progression

Key Points Question Are patients with chronic kidney disease who are at high risk of kidney failure receiving nephrology care within 1 year of established risk? Findings In this cohort study of 156 733 adult patients with chronic kidney disease, 58% of patients at high risk of progressing to kidney failure had a nephrology visit within 1 year of established risk. Meaning These findings suggest that many patients with chronic kidney disease at high risk of kidney failure do not receive nephrology care within 1 year of established risk; better strategies are needed for identifying and referring high-risk patients.

This supplementary material has been provided by the authors to give readers additional information about their work.

Sensitivity Analyses
Excluding patients with a nephrology visit in the last three years: It is possible that excluding patients with a nephrology visit within the past year may not be a long enough timeframe to accurately measure the rates of incident nephrology care. Some patients may have had appropriate reasons not to see a nephrologist within one year of a previous visit. We thus extended the timeframe of exclusion from one year to three years and calculated the visit rates for this smaller patient population. Prevalent nephrology care: In our primary analysis, we excluded patients with a nephrology visit within one year prior to index time and calculated visit rates using visits to nephrology after index time. This approach captured the rates of incident nephrology care, i.e. care administered to patients who had not been previously referred or seen. To evaluate the rates of prevalent nephrology care, we kept patients with visits to nephrology within one year prior to index time and calculated a "two-sided" nephrology visit rate as the fraction of patients with a visit to nephrology within one year prior or one year after index time.
Kaplan-Meier analysis: Because the outcome of our primary was defined by having a nephrology visit within one year of the index date, we only included patients who had at least one year of follow-up time. However, this may bias the results if excluded patients (those who disenrolled or died within one year) differ substantially from retained patients. To examine this possible bias, we reintroduced the patients that had been excluded due to death or disenrollment within one year and used Kaplan-Meier survival models to determine the one-year cumulative incidence of nephrology visits.
Urine PCR to urine ACR conversion: Many patients were excluded from the primary analysis due to not having an available ACR measurement. To address this limitation and construct a larger, more representative cohort, we utilized a validated conversion equation to convert urine PCR measurements to urine ACR values. 1 Visit rates in patients without an available ACR: If patients without an available ACR measurement differed systematically from patients with available an ACR measurement, then excluding the former could bias our estimates of visit rates. To address this concern, we calculated visit rates using the date of eGFR result as index time for (1) patients with eGFR < 60 mL/min/1.73 m 2 and (2) patients with eGFR < 60 mL/min/1.73 m 2 and missing ACR. With (1), we also addressed the concern that requiring a calculable KFRE may skew index time: that is, patients may have been referred for nephrology care using an eGFR value alone, and ACR measurements only became available after the nephrology visit.
2009 CKD-EPI: KFRE was originally developed to use eGFR calculated from the 2009 version of the CKD-EPI equation, rather than the 2021 version. 2 However, the 2009 version is now discouraged due to its use of race as an input variable. 3 To determine whether the version of the CKD-EPI equation affected the results, we calculated nephrology visit rates across KFRE scores that used eGFR values computed with the 2009 CKD-EPI equation.

eFigure 1. KFRE Calibration
We assessed the calibration of the KFRE risk score by comparing the true rate of 5-year kidney failure to the KFRE estimated risk of kidney failure. Within each risk group, we computed the probability of kidney failure occurring within 5 years as the number of individuals in the group who underwent dialysis or transplant within 5 years of index time divided by the total number of individuals in the group. A well-calibrated score should fall close to the y=x line (dashed line in plots). We observed that the KFRE risk score was reasonably well-calibrated, both overall and within subgroups. The "2018-2019" subgroup was omitted due to unobserved 5-year follow up, as data were only available through 2021. Similarly, the "80 or above" subgroup had high variability due to small patient numbers. For our sensitivity analysis, we computed the nephrology visit rates in populations derived with criteria that were modified from those used to construct the primary analysis population. See Supplemental Methods for details on the alterations made to the criteria. We note that only the prevalent nephrology care sensitivity analysis yielded different visit rates, which we comment on in Discussion. The 95% confidence interval was obtained from 100 bootstrapping iterations.