Gaps in Nephrology Referral Care Utilization in Patients at High-Risk of Progression to Kidney Failure

Abstract

Early nephrology specialty care slows progression of chronic kidney disease (CKD) to end-stage renal disease (ESRD). However, identifying which patients are expected to progress to end-stage disease has been historically challenging to predict. With a limited supply of nephrologists, optimizing nephrology referral is essential for improving patient outcomes. The Kidney Failure Risk Equation (KFRE) provides an accurate metric to identify patients who are at high risk of progression to kidney failure. In this study, we utilize the KFRE to perform a retrospective analysis in a local health network to identify rates of nephrology referral for CKD patients stratified by risk of kidney failure progression. We found a nephrology referral gap in CKD patients at higher risk of progression and an underutilization of albuminuria testing in CKD, suggesting opportunities to improve outcomes by 1) proactively targeting high-risk patients using EHR-based informatics strategies and 2) increasing albuminuria testing as a screening tool.

Introduction

The burden of chronic kidney disease (CKD) is substantial, with a global prevalence of 9.1% and 4.6% of global deaths being attributed to CKD in 2017.^1,2 CKD increases the risk of cardiovascular disease events and progression to end-stage renal disease (ESRD), ultimately leading to increased morbidity and mortality.^1,2 Early interventions guided by nephrology specialty care may slow the progression of disease in CKD patients, while late evaluation by a nephrologist is associated with shorter duration of survival.^3,4,5,6 Individuals with CKD have significant heterogeneity in the risk of progression to ESRD. With the known workforce shortage of nephrologists in the United States and worldwide, a key challenge is to optimize referrals to nephrology to assure specialty care access to those at highest risk of ESRD. ⁷

The Kidney Failure Risk Equation (KFRE) is a prediction model developed by Tangri et al that may enable early, appropriate patient care for patients with CKD and high risk of progression to kidney failure.⁸ The KFRE uses demographic and laboratory data to accurately predict the risk of kidney failure. The risk equations have been internationally validated across diverse patient cohorts, demonstrating excellent discrimination in predicting which CKD patients progress to kidney failure.^8,9 When applied in a risk-based triage strategy in Canada and the United Kingdom, the KFRE model has been shown to minimize potentially unnecessary nephrology care for those at low risk of CKD progression and prevent delays in specialty care access for those at highest risk of kidney failure.^10,11 In the U.K., the use of KFRE to guide referral strategies to nephrology specialty care compared to previously established guidelines would have resulted in changes in nephrology referral for 14.8% of individuals with chronic kidney disease, including 8.8% of patients at high risk of ESRD who would not have been referred based on prior guidelines. ¹¹

Given substantial differences in the way healthcare is delivered in Canada, the United Kingdom and the United States, it is not clear whether these results would apply in the United States. To date, very little has been published about the complexity of patients seen in U.S. nephrology clinics or their risk for CKD progression. ^12,13 With this study, we aimed to (1) evaluate rates of nephrology referral in a representative regional health network across kidney failure risk levels, and (2) evaluate albuminuria ordering behavior in patients with chronic kidney disease. Results could inform future strategies to optimize the delivery of scarce nephrology care and improve clinical outcomes for patients with high risk of ESRD.

Methods

We conducted a retrospective cohort study that examined kidney failure risk among individuals receiving ambulatory care within one regional network in California. The STAnford Research Repository (STARR) electronic health record database includes all clinical data at Stanford Health Care, including electronic health record data from its main hospitals as well as data from various clinical ancillary systems.

For our study, we limited our cohort to patients with demonstrated kidney injury that we defined as an estimated glomerular filtration rate (eGFR) value that would classify as Stage IIIa CKD or above (eGFR<60 ml/min/1.73m²) since this is the eGFR range used in the validation studies for the KFRE.^8,9 We identified all patient encounters between 2015 and 2020 in which an eGFR<60 ml/min/1.73m² and UACR were obtained, allowing for up to 90 days separation between the patient’s lab results because the labs may not have been obtained simultaneously. These were considered “calculable KFRE encounters”, which were encounters during which calculation of kidney failure risk could have been available based on recent laboratory data. Some patients had multiple calculable KFRE encounters during our time window based on our criteria. We decided to include each of these encounters in our analysis because we were interested in determining how often patients received specialty nephrology care within 1 year of a possible risk stratification timepoint, regardless of whether the time of possible risk stratification was a new diagnosis of high-risk kidney disease. This is because patients with CKD who are determined to be at high risk of progression to kidney failure can benefit from timely nephrologist care even if their kidney disease is not newly diagnosed. Additionally, we excluded patients with a history of ESRD (either receiving dialysis or having a history of kidney transplant), identified using International Classification of Diseases, 10th Revision (ICD-10) codes. Because our goal was to examine nephrology follow-up care within 1 year following potential risk stratification for patients who were not at terminal stages of disease, we also excluded patients who were not alive at least one year following their encounter date.

The independent variable was 5-year risk of progression to kidney failure calculated by the 4-variable Kidney Failure Risk Equation.⁸ Risk predictions were calculated for all patients using age, sex, estimated glomerular filtration rate, and urine albumin to creatinine ratio.^7,8

The outcome was receipt of nephrology care within 1 year following potential risk stratification. Patients were categorized as having received follow-up nephrology care in this interval if they had at least one nephrology office visit at Stanford Health Care within 1 year of their initial kidney failure risk calculation.

We stratified patient encounters into 5-year kidney failure risk levels from 0% to 100% in increments of 2% and identified the proportion of calculable KFRE patient encounters that received nephrology specialty care within +/- 1 year of KFRE calculation.

To evaluate albuminuria ordering behavior, we determined the total number of Stanford Health Care patients with eGFR values that would categorize them as CKD III or above (eGFR<60 ml/min/1.73m²). Among these patients, we then determined how many received albuminuria testing +/-90 days from their abnormal eGFR lab result.

The study was approved by the Stanford University Institutional Review Board. The analysis was conducted in R, version 4.0.1 (R Foundation).

Results:

Between 2015-2020, there were a total of 1,507,891 distinct patients in the hospital system. We identified 11,877 calculable ambulatory KFRE encounters representing a cohort of 6,444 patients. Mean age was 72.8 years old, and approximately 51% were male, 51% were White, 19% were Asian, 12% were Black, 8% were Hispanic/Latino, 1% were Pacific Islander, and 0.4% were Native American. 42% of patients were privately insured, while 55% were on Medicare and 3% on Medicaid (Table 1). The most common encounter diagnoses for patients in our cohort were Type II diabetes, hypertension, hyperlipidemia, chronic kidney disease, and routine exam.

Table 1.

Patient Cohort Characteristics and Demographic Information

Nephrology referral rate by kidney failure risk

5-year risk of kidney failure was calculated for all patient encounters that met inclusion criteria. For each kidney failure risk level, the proportion of corresponding patient encounters that received follow-up nephrology care within 1 year was determined and graphically recorded in Figure 1. Local estimated scatterplot smoothing (LOESS) was used for the regression line plotted within the scatterplot to allow for easy referencing of kidney failure risk level to estimated corresponding nephrology referral rate. A histogram outlining the proportion of total encounters in the cohort for each 5% kidney failure risk interval was also included in Figure 1.

Figure 1.

Nephrology Referral Rate for Calculable KFRE Encounters across Kidney Failure Risk

% of the patient encounters in the analysis (8,223 out of 11,877) had a 5-year risk of kidney failure of less than 5%. 10.2% of these patient encounters received nephrology care within 1 year. Receipt of nephrology care increased with increasing kidney failure risk but appeared to level off above a 40% 5-year kidney failure risk level. Only 54% of patients with a calculable KFRE encounter indicating greater than a 40% risk of ESRD in 5 years received a nephrology visit within one year (216 out of 403).

Evaluating Albuminuria Test Ordering Behavior

Among the 859,877 total patients at Stanford Health Care between 2015-2020, 56,910 had at least one eGFR lab result that would be classified at or above the Stage III CKD category of eGFR<60mL/min/1.73m². For these patients, 6,855 received an albuminuria test within 90 days of their abnormal eGFR result (Figure 2).

Figure 2.

Flowchart Summarizing Albuminuria (uACR) Ordering Behavior between 2015-2020 at SHC

Discussion

A risk of 3% over 5 years for kidney failure, as determined by the KFRE, has been used in the past as a threshold to triage nephrology care.¹⁰ In our study, while the majority of patients in our study population who receive both a urine albumin to creatinine ratio and eGFR result have a 5-year risk of progression to kidney failure of less than 3% (59% of cohort; 6,900 out of 11,689 patient encounters), a substantial fraction of the patients greater than a 3% risk of kidney failure in 5 years are not receiving timely nephrology specialty care – the patients from 64% of patient encounters at 10% kidney failure risk level (420 out of 652 patient encounters), 52% of patient encounters at 20% kidney failure risk level (110 out of 212 patient encounters), and 40% of patient encounters at 50% kidney failure risk level (20 out of 50 patient encounters) did not receive follow-up nephrology care within 1 year of a calculable KFRE encounter. Early intervention for these higher risk patients could help slow CKD progression and ultimately improve patient outcomes.

Prior studies have shown that early referral to nephrology specialty care in high-risk patients is associated with better outcomes. ^{3,4,6,14,15,16} Despite the demonstrated improvement in clinical outcomes from early specialty care referral, many high-risk patients often do not receive timely nephrology specialty care.^{17, 18} In an effort to avoid late referral to specialty care services, The Kidney Disease Improving Global Outcomes (KDIGO) organization has outlined specific guidelines for nephrology referral in CKD patients in an effort to capture the heterogenous risk factors (ranging from eGFR cutoffs to history of recurrent nephrolithiasis) that lead to an increased likelihood of progression to kidney failure.⁴ However in a study aimed at analyzing the projected impact of implementing the KDIGO referral recommendations at a well-resourced tertiary care center, the projected increase in nephrology referrals based on KDIGO guidelines far outnumbered the institution’s supply of nephrology specialty care.¹⁹ Such a supply-demand mismatch in a well-resourced medical center suggests that strict adherence to KDIGO guidelines may not be feasible in other, lower-resourced institutions as well.¹⁹ While recommending guidelines to standardize early specialty care referral for high risk patients is a step in the right direction, alternative strategies to prevent late referral for high risk CKD patients must be considered. ¹⁹

As shown in our study, there appears to be an underutilization of nephrology specialty care for patients at highest risk of kidney failure, likely secondary to the heterogeneity in disease burden across CKD patients which makes clinical prognostication challenging for primary care providers. The KFRE is a widely validated metric for evaluating kidney failure risk that can be utilized with an institution’s EHR data to target a nephrology referral gap in patients who are at high risk of progression to kidney failure. Informatics strategies such as developing digital tools embedded in an EHR to automate triage of patients using KFRE-based risk stratification can help streamline the referral process for these high-risk CKD patients. Once an automated risk stratification is made, population management strategies such as e-referrals and outreach programs can be deployed to proactively target the 10-15% of patients who have a 5-year risk of kidney failure greater than a certain threshold, such as 3%. Additionally, a data-driven triage system also serves to prevent unnecessary referrals for patients at very low risk of progression to kidney failure, ultimately enabling more nephrologists to be available for those high-risk patients who need timely specialty care the most.

However, the limiting factor in such a population risk stratification strategy is the albuminuria testing. As shown in Figure 2, there is a marked underutilization of albuminuria testing for patients with signs of kidney injury. In our study population, only 12% of patients with an eGFR <60ml/min/1.73m² received an albuminuria test result within +/-90 days of their abnormal eGFR result (6,855 out of 56,910 patients). For the 88% of patients who had a decreased eGFR with no follow-up albuminuria test, the prognostic value of KFRE cannot be appreciated and therefore these patients may be excluded from any KFRE-based risk stratification strategy.

To effectively narrow the nephrology referral gap using risk prediction tools like the KFRE, the first step is to test for albuminuria more broadly in patients with demonstrated kidney disease. Doing so increases the total number of calculable KFRE predictions and enables more effective population risk stratification and better nephrology outreach based on individualized patient risk. Prevention of CKD complications requires a concerted effort on multiple fronts. Increased albuminuria testing to enable KFRE-based triage combined with additional interventions such as promoting self-efficacy of primary care providers as well as specialty e-consults for risk-based outreach may provide a means to optimize efficient use of nephrology specialty care and ultimately improve outcomes for CKD patients.

Limitations:

The analysis in this cohort is not within a closed system – patients could have received nephrology care outside of the Stanford health care network and therefore their follow-up nephrology care would not be captured, leading to an overestimation of the referral gap.

As risk of kidney failure increases, the number of patients in a given kidney failure risk level drops significantly, with roughly 70% of the entire study population having a 5-year risk of kidney failure less than 5%. This decrease in sample size decreases the statistical power of the nephrology referral gap evaluation in higher kidney failure risk levels.

The Stanford Health Care patient cohort is limited to one regional health network, therefore is not necessarily representative of referral gaps on a larger scale. However, the findings are consistent with prior studies indicating inefficient use of nephrology specialty referral.^{17, 18} The methodology used here to quantitatively evaluate the potential nephrology referral gap using EHR data can be applied broadly to target quality improvement projects and population risk stratification strategies aimed at improving nephrology care for CKD patients in other regional health networks.

Conclusion

In our study, when urine albumin screening was used at a local level in the United States, despite being predictive of high risk for renal failure in CKD patients using the Kidney Failure Risk Equation, there was often a missed opportunity in care delivery. We found modest nephrology referral rates in patients at higher risk of progression to kidney failure who would benefit from early nephrology specialty consultation and management, indicating an inefficient use of healthcare resources and a need for a more targeted approach to proactively provide care for higher risk patients. The KFRE provides an excellent tool to identify such patients and can possibly be utilized as an HER- embedded triage tool to optimize risk-stratification for patients with CKD and guide appropriate, efficacious specialty referral care. However, we found that urine albumin screening was markedly underutilized in patients with demonstrated kidney injury. To effectively risk-stratify patient populations based on risk of kidney failure, increased evaluation of albuminuria is needed in patients with chronic kidney disease to enable optimized nephrology specialty care and better patient outcomes for those at higher risk.