Abstract
Background
Early detection and treatment of chronic kidney disease (CKD) is important for slowing progression to renal failure and preventing cardiovascular events, but CKD is often not recognized and patients are referred to nephrologists too late for timely management. Automated laboratory reporting of estimated glomerular filtration rate (eGFR) has been introduced in many health systems to improve CKD recognition, but its impact in large, US-based health systems remains unclear.
Research Design
Retrospective time-series study examined change in renal care services and CKD recognition across VA healthcare system facilities in 2000-2009. Hierarchical generalized linear models were used to estimate immediate and long-term impacts of eGFR reporting across facilities on monthly rates of outpatient CKD diagnoses, utilization of CKD diagnostic tests (urine microalbumin and kidney ultrasound), and outpatient nephrology visits.
Results
Rates of CKD recognition through diagnoses in patient medical records changed an average of 11.4 additional diagnosed patients per 10,000 in the general outpatient population per month, with sustained long-term increases in CKD diagnoses (P<.001). Diagnostic microalbumin and kidney ultrasound testing increased significantly, with long-term increases microalbumin testing (P<.001) and short-term increases in kidney ultrasound (P=.01-.04) rates across the VHA. There was no significant change in nephrology consultation rates.
Conclusions
Automated eGFR reporting was associated with moderate system-level improvements in documentation of CKD diagnoses and use of diagnostic tests, but had no impact on nephrology consultation. To effectively reduce the large burden of disease and its associated complications, further strategies are needed to identify and provide timely treatment to those with CKD.
Keywords: health system, IT implementation, chronic kidney disease, Veterans Affairs
Introduction
The early detection and treatment of chronic kidney disease (CKD) is important for slowing progression to end-stage renal disease and preventing cardiovascular events (1-3). However, opportunities for timely intervention are often missed when CKD is not recognized and patients are referred to nephrologists too late (1, 4). Automated laboratory reporting of estimated glomerular filtration rate (eGFR), a measure of kidney function based on an algorithm of patient's serum creatinine and demographic characteristics, has been introduced in many health systems to improve CKD recognition. Despite clinical guidelines promoting the use of eGFR in practice (5-7) and increased implementation of eGFR reporting (8), the impact of eGFR reporting on health systems’ CKD management remains unclear.
Evaluations of automated eGFR reporting have assessed the impact in large, population-based samples of national health care systems outside of the United States (US) and small clinical practices in the US (8). However, the impact of eGFR reporting in large US integrated health systems is unknown. A detailed examination of how automated eGFR reporting impacts recognition of CKD has important implications for health systems’ capacity to treat CKD patients and may better inform refinements to optimize the use GFR reporting for CKD prevention and disease management efforts.
The Veterans Health Administration (VHA) is the nation's largest integrated health care system, and the prevalence of CKD among veterans is 30% greater than among the general population (9). In 2004, the VHA National Pathology and Laboratory Service mandated that all Veterans Affairs medical centers (VAMCs) install software that allowed their laboratory reporting systems to provide eGFR values from serum creatinine measures. Implementation of the mandate was decentralized, such that each VAMC independently determined when to activate this reporting feature as part of their standardized laboratory reporting (10). We hypothesized that the implementation of systematic reporting of eGFR would increase the rate of documented diagnoses of CKD in the electronic medical record (defined as eGFR < 60 mL/min/1.73 m2 for CKD stages 1-5), diagnostic testing, and outpatient nephrology visits across the VHA. Results from this large national sample may inform future allocation of resources (i.e. nephrology units and staffing for consultation) and efforts to improve initial screening and evaluation of CKD in large health systems (11).
Methods
Study Design, Data Sources, Population
We used a time-series analysis to examine changes in renal care services and CKD recognition across VAMC facilities in 2000-2009. The study sample and system-level outcomes were drawn from health system and patient-level data in the VA Decision Support System's Laboratory Results National Data Extract, National Patient Care Database Patient Treatment File (PTF) for inpatient care, and Outpatient Care (OPC) files. The institutional review board of the Durham VAMC approved this study.
VA facility laboratories implemented eGFR reporting, which uses the Modification of Diet in Renal Disease equation (12), from 2.5 months to 4 years after the VHA mandate on July 2004 (10). We accounted for facility-specific initiation and eGFR reporting patterns in determining pre- and post-eGFR reporting periods, where time zero is defined as the end of the pre-reporting and beginning of the post-reporting periods of interest at each VAMC laboratory. Pre- and post-eGFR reporting periods varied across VAMCs, reflecting a range of reporting patterns (10).
The study sample consisted of VHA community-based outpatient clinics (CBOCs) or Veteran Affairs medical centers (VAMCs) with available laboratory data. VHA facilities that did not have their own clinical laboratories (44 facilities) and independent VHA facilities (e.g. skilled nursing facilities, rehabilitation centers, and domiciliaries) were excluded. In addition, facilities were excluded from analysis if there was a gap in time-series data >2 months in duration or if they did not have complete laboratory data at least 24 months in the pre-eGFR period or at least 12 months of post-eGFR reporting period laboratory data.
Measurements
We created monthly time-series data for each included VA facility. Consistent with previous studies (13, 14), the primary outcomes of interest were monthly rates of 1) outpatient nephrology visits (identified through VHA encounter codes) ; 2) outpatient diagnoses of CKD stages 1-5 in the patient medical record (ICD-9-CM diagnosis codes 250.4, 250.40, 504.1, 250.42, 504.3, 582.x, 583.x, 586.x, 587.x, 403.x, 404.x, and 585.x); and 3) utilization of relevant CKD diagnostic procedures (urine microalbumin and kidney ultrasound). In VA facilities, diagnostic codes for outpatient encounters are entered by physicians associated with the visit rather than by coders or billing specialists (15). Rates were defined as counts per 10,000 in the general outpatient population per month. We present both overall rates and first (or initial) rates for each outcome. Diagnosis and utilization outcomes were defined as “first” if obtained for the first time ever or the first time in 4 years. These outcomes were chosen because they serve as proxies for disease identification (e.g. diagnosis) or represent an active response (e.g. testing to evaluate CKD or nephrology consultation), and therefore greater recognition of CKD in response to eGFR reporting. All outcomes were identified from procedural and diagnosis codes found in administrative records from OPC files.
Analysis
For each VAMC facility, each time series was interrupted at some point by the implementation of automated eGFR reporting by their affiliated VAMC laboratory. We analyzed these interrupted time-series data using a segmented regression analysis (16), to assess the effect of eGFR reporting and account for secular time trend and underlying trends on our facility-level outcomes. All time series started 24 to 48 months before and continued up to 36 months after implementation of automated eGFR reporting, but were subject to censoring (i.e. study dates of January 2000 and December 2009). Time was measured in months and because implementation dates varied by facility, the last month of each VAMC facility laboratory's pre-eGFR reporting period was considered time zero.
With each facility's monthly time series as the dependent variable, we applied the general model in separate analyses:
where Yt is the time series value at time t. Explanatory variables included a variable for time, an indicator for the post-eGFR reporting period, and an interaction term between time and the post-eGFR reporting period. The post- eGFR reporting indicator allowed the estimated trend lines to be discontinuous at the time of eGFR reporting implementation. The interaction term allowed the slopes of the trend lines to vary before and after eGFR reporting implementation. Since eGFR reporting in this natural experiment is an exogenous treatment (i.e., available and exposed to all VA physician or patients), it is not likely to be confounded by provider or patient factors and thus, no further adjustment was made for confounding.
We estimated the regression models using hierarchical generalized linear model methods to enable simultaneous analysis of time-series data across multiple facilities within the VA healthcare system. Models specified an identity link with normally distributed errors to model effects on a linear, or absolute, scale. We allowed for site-specific variability around each of the parameters in the model (β0, β1, β2, β3) using random effects. This approach had the effect of accounting for autocorrelation between time-series values within each facility over time and resulted in conditional (within-facility) estimates of the effect of eGFR reporting. We report the parameter estimates for the average facility by assuming no random variation.
The impact of eGFR reporting was evaluated by assessing changes in the level and the slope of the outcome rates before and after eGFR reporting. A significant change in the level of the trend after eGFR reporting (β2) indicates an immediate increase (or decrease) in the outcome rate following implementation. A significant change in the slope of the trend after eGFR reporting (β3) indicates an increase (or decrease) in the rate of change of the outcome rate over time after automated eGFR reporting.
To evaluate the robustness of our findings, we also conducted additional analyses using Poisson regression as well as a weighted regression model, and the results were largely similar.
Results
We identified 104 VHA VAMCs with laboratories that activated automated eGFR reporting between July 2004 and December 2008. Of these facilities, 97 (93%) had at least 24 months of data during the pre-eGFR reporting period and at least 12 months of data during the post-EGFR reporting period and were included in the main analysis. After implementation of automated eGFR reporting (Table 1), mean rates of overall CKD diagnoses increased 74.2% (from 46.1 to 80.3 per 10,000 patients in the general VHA outpatient population), urine microalbumin testing increased 47.8% (109.8 to 162.3), overall kidney ultrasound procedures increased 21.2% (20 to 24.3), and overall nephrology visits increased 29.8% (29.2 to 37.9). Incident outcomes rates also increased: 24.7% for first CKD diagnoses (from 8.5 to 10.6), 57.5% for first urine microalbumin testing (29.9 to 47.1), 16.7% for first kidney ultrasound procedures (13.8 to 16.1), and 20% for first nephrology visits (4.5 to 5.4).
Table 1.
Mean (SD) Monthly Time-Series Values for all VA Facilities, Sample Months
| Month Relative to eGFR reporting implementation | CKD Diagnoses | CKD Diagnostic Testing: Microalbumin | CKD Diagnostic Testing: Kidney Ultrasound | Nephrology Visits | ||||
|---|---|---|---|---|---|---|---|---|
| All | First | All | First | All | First | All | First | |
| −48 | 27.6 (17.8) | 6.8 (3.1) | 84.2 (52.4) | 34.4 (20.9) | 18.1 (12.0) | 12.4 (7.5) | 23.6 (13.4) | 4.0 (2.0) |
| −42 | 25.7 (16.7) | 6.1 (2.8) | 85.3 (51.4) | 38.4 (33.4) | 18.4 (11.1) | 12.8 (6.6) | 23.9 (13.8) | 4.0 (2.2) |
| −36 | 28.5 (22.2) | 6.0 (3.5) | 87.6 (61.7) | 34.9 (20.9) | 16.9 (10.3) | 12.2 (6.5) | 23.8 (13.7) | 3.8 (2.3) |
| −30 | 31.3 (22.8) | 6.6 (3.3) | 96.4 (59.5) | 36.7 (20.9) | 17.2 (10.0) | 12.3 (6.2) | 24.6 (13.3) | 4.0 (1.7) |
| −24 | 32.5 (26.0) | 6.8 (3.4) | 97.1 (57.4) | 34.4 (21.0) | 17.9 (10.6) | 12.5 (6.4) | 25.8 (14.9) | 4.2 (1.9) |
| −18 | 35.7 (24.6) | 7.5 (3.7) | 99.1 (56.8) | 34.9 (21.4) | 18.4 (10.9) | 12.7 (6.9) | 26.7 (15.8) | 4.2 (2.1) |
| −12 | 39.8 (27.7) | 7.8 (3.7) | 107.4 (63.3) | 30.9 (14.2) | 19.0 (11.7) | 12.9 (6.5) | 27.3 (15.7) | 4.3 (1.8) |
| −6 | 42.4 (29.9) | 8.4 (4.6) | 109.6 (61.5) | 30.5 (13.8) | 18.4 (12.1) | 12.5 (7.2) | 26.8 (14.9) | 4.0 (1.9) |
| 0 | 46.1 (35.7) | 8.5 (4.2) | 109.8 (72.0) | 29.9 (14.5) | 20.0 (13.5) | 13.8 (9.3) | 29.2 (17.9) | 4.5 (2.2) |
| 6 | 60.5 (38.2) | 11.0 (3.7) | 131.3 (74.2) | 36.6 (16.6) | 21.2 (11.4) | 14.7 (7.4) | 31.7 (16.3) | 5.1 (2.3) |
| 12 | 65.2 (37.3) | 10.7 (3.8) | 134.8 (80.9) | 36.6 (16.7) | 21.9 (16.1) | 15.4 (12.5) | 30.7 (16.5) | 4.8 (3.2) |
| 18 | 70.6 (40.9) | 11.0 (4.6) | 152.1 (77.6) | 41.9 (24.6) | 22.6 (14.9) | 16.2 (10.3) | 31.4 (15.9) | 4.7 (2.1) |
| 24 | 69.3 (31.9) | 10.7 (4.6) | 130.7 (64.3) | 38.6 (22.7) | 21.2 (12.1) | 14.8 (8.0) | 34.4 (18.5) | 5.0 (2.6) |
| 30 | 74.7 (35.1) | 10.5 (3.4) | 150.5 (61.3) | 41.3 (16.5) | 24.5 (14.5) | 16.5 (9.0) | 34.2 (19.2) | 4.9 (2.1) |
| 36 | 80.3 (42.2) | 10.6 (3.4) | 162.3 (58.8) | 47.1 (21.4) | 24.3 (14.1) | 16.1 (9.1) | 37.9 (24.5) | 5.4 (2.1) |
Notes: Reported as mean of all facility rates, as counts per 10,000 in the general outpatient population per month. Standard deviations noted in parens.
Figure 1 illustrates the short- and long-term impacts of EGFR reporting in VA facilities, where the dotted line represents the expected trajectory of pre-eGFR trends, had eGFR reporting not been implemented. The impact of automated eGFR reporting by VAMC laboratories varied across outcomes (Table 2). Specifically, rates of CKD recognition through diagnoses in patient medical records increased significantly: for overall CKD diagnoses, there was an average increase of 11.4 of diagnosed patients per 10,000 in the general outpatient population per month (95% confidence interval [CI], 5.3, 17.4, P<.001) and for first CKD diagnoses, we found an increase of 3.1 patients (95% CI, 2.4, 3.9, P<.001) in the post-eGFR reporting period. The increase in CKD diagnosis rates was largely driven by increased diagnoses among older patients (Table 3). In our analyses of long-term trends, there was a sustained increase in CKD diagnoses up to 3 years after implementation of automated eGFR laboratory reporting. There was a small but significant increase in the slope of overall CKD diagnoses (Figure 1), indicating that the magnitude of growth in diagnosis rate in the medical record increased over time following automated eGFR reporting (P<.001). However, the decrease in the slope of first CKD diagnoses indicates diminishing growth of first CKD diagnoses in the medical record over time (P<.05).
Figure 1.
Time series of monthly rates of outcomes (per 10,000 population in Veterans Health Administration facilities)
Table 2.
Results of Time-Series Regression Models Estimating the Overall Effect of Automated eGFR Laboratory Reporting on Facility CKD Outcomes, per Month1
| Time Series | Change in Level After eGFR Reporting | Slope After eGFR Reporting | |||
|---|---|---|---|---|---|
| Change Level2 (95% CI) | P | % change | Change Trend (95% CI) | P | |
| CKD Diagnoses | |||||
| All | 11.37 (5.31, 17.43) | <.001 | 24.66% | 0.46 (0.26, 0.66) | <.001 |
| First | 3.11 (2.35, 3.88) | <.001 | 36.59% | −0.05 (−0.09, −0.01) | .02 |
| CKD Diagnostic Testing: Microalbumin | |||||
| All | 3.21 (−6.33, 12.75) | .50 | 2.92% | 1.39 (0.62, 2.16) | <.001 |
| First | 2.75 (−2.39, 7.90) | .29 | 9.20% | 1.07 (0.67, 1.47) | <.001 |
| CKD Diagnostic Testing: Ultrasound | |||||
| All | 2.01 (0.07, 3.94) | .04 | 10.05% | 0.04 (−0.05, 0.12) | .39 |
| First | 1.98 (0.59, 3.37) | .006 | 14.35% | 0.03 (−0.03, 0.09) | .35 |
| Nephrology Visits | |||||
| All | 0.46 (−1.44, 2.36) | .63 | 1.58% | 0.07 (−0.07, 0.22) | .32 |
| First | 0.32 (−0.05, 0.69) | .09 | 7.11% | 0.01 (−0.01, 0.02) | .24 |
Notes:
A significant change in the level of the trend after eGFR reporting indicates an immediate increase (or decrease) in the outcome rate following implementation. A significant change in the slope of the trend after eGFR reporting indicates an increase (or decrease) in the rate of change of the outcome rate over time following implementation.
Change level reported in facility rates, as counts per 10,000 in the general outpatient population per month.
3. Percent change reported from overall change level and baseline rates (time 0).
Table 3.
Results of Time-Series Regression Models Estimating the Effect of Automated eGFR Laboraton Reporting on Facility CKD Outcome First Events, per Month by Patient Characteristics1
| Time Series | Level After eGFR Reporting | Slope After eGFR Reporting | |||
|---|---|---|---|---|---|
| Change Level2 | P | % change | Change Trend | P | |
| First CKD Diagnoses | |||||
| Age | |||||
| <40 | 0.21 (0.06, 0.36) | .006 | 30.00% | −0.00 (−0.01, 0.00) | .41 |
| 40-49 | 0.60 (0.34, 0.85) | <.001 | 31.58% | 0.01 (−0.01, 0.02) | .41 |
| 50-59 | 1.19 (0.81, 1.57) | <.001 | 27.67% | 0.01 (−0.02, 0.03) | .53 |
| 60-69 | 2.54 (1.89, 3.19) | <.001 | 33.42% | −0.02 (−0.06, 0.02) | .25 |
| 70-79 | 5.28 (4.03, 6.54) | <.001 | 40.62% | −0.06 (−0.12, 0.01) | .08 |
| ≥80 | 6.65 (5.05, 8.26) | <.001 | 38.00% | −0.13 (−0.21, −0.05) | .002 |
| Male | 3.41 (2.58, 4.23) | <.001 | 37.47% | −0.05 (−0.09, −0.01) | .02 |
| Female | 0.91 (0.52, 1.29) | <.001 | 39.57% | 0.00 (−0.02, 0.02) | .97 |
| First CKD Diagnostic Testing: Microalbumin | |||||
| Age | |||||
| <40 | 1.25 (−3.31, 5.80) | .58 | 11.90% | 0.35 (0.10, 0.59) | .006 |
| 40-49 | 1.39 (−3.61, 6.39) | .58 | 5.82% | 0.74 (0.41, 1.08) | <.001 |
| 50-59 | 0.25 (−5.32, 5.82) | .93 | 0.69% | 1.27 (0.79, 1.74) | <.001 |
| 60-69 | 4.49 (−2.10, 11.08) | .18 | 11.06% | 1.32 (0.86, 1.78) | <.001 |
| 70-79 | 3.88 (−2.07, 9.83) | .20 | 13.02% | 1.28 (0.82, 1.73) | <.001 |
| ≥80 | 3.20 (−1.19, 7.58) | .15 | 13.91% | 0.98 (0.62, 1.35) | <.001 |
| Male | 2.86 (−2.64, 8.37) | .30 | 9.05% | 1.15 (0.73, 1.58) | <.001 |
| Female | 2.89 (−1.40, 7.18) | .18 | 14.82% | 0.55 (0.28, 0.83) | <.001 |
| First CKD Diagnostic Testing: Ultrasound | |||||
| Age | |||||
| <40 | −0.11 (−0.97, 0.75) | .80 | −1.57% | −0.01 (−0.05, 0.03) | .68 |
| 40-49 | −0.00 (−1.26, 1.26) | >.99 | 0% | 0.06 (0.01, 0.12) | .03 |
| 50-59 | 0.54 (−0.94, 2.02) | .47 | 3.51% | 0.08 (0.01, 0.16) | .03 |
| 60-69 | 3.37 (1.37, 5.38) | .001 | 20.06% | 0.03 (−0.06, 0.11) | .51 |
| 70-79 | 4.60 (1.78, 7.42) | .002 | 28.05% | −0.00 (−0.14, 0.14) | .97 |
| ≥80 | 0.84 (−0.10, 1.79) | .08 | 6.83% | −0.01 (−0.05, 0.03) | .55 |
| Male | 2.16 (0.69, 3.63) | .004 | 14.90% | 0.03 (−0.03, 0.10) | .31 |
| Female | 0.57 (−0.47, 1.61) | .28 | 5.70% | 0.02 (−0.02, 0.06) | .37 |
| First Nephrology Visits | |||||
| Age | |||||
| <40 | −0.14 (−0.38, 0.10) | .24 | −14.00% | 0.00 (−0.01, 0.01) | .36 |
| 40-49 | 0.18 (−0.09, 0.45) | .18 | 9.00% | 0.01 (−0.00, 0.02) | .08 |
| 50-59 | −0.06 (−0.40, 0.28) | .72 | −1.40% | 0.02 (0.01, 0.03) | .001 |
| 60-69 | 0.55 (0.06, 1.04) | .03 | 9.82% | 0.01 (−0.01, 0.03) | .27 |
| 70-79 | 0.80 (0.18, 1.41) | .01 | 13.56% | 0.02 (−0.01, 0.04) | .13 |
| ≥80 | 0.80 (0.11, 1.49) | .02 | 14.29% | 0.00 (−0.02, 0.02) | .99 |
| Male | 0.35 (−0.04, 0.75) | .08 | 7.14% | 0.01 (−0.01, 0.03) | .18 |
| Female | 0.40 (0.09, 0.70) | .01 | 25.00% | 0.01 (−0.01, 0.02) | .32 |
Notes:
A significant change in the level of the trend after eGFR reporting indicates an immediate increase (or decrease) in the outcome rate following implementation. A significant change in the slope of the trend after eGFR reporting indicates an increase (or decrease) in the rate of change of the outcome rate over time following implementation.
Change level reported in facility rates, as counts per 10,000 in the general outpatient population per month.
3. Percent change reported from overall change level and baseline rates (time 0).
Diagnostic testing to further characterize CKD with microalbumin and kidney ultrasound increased significantly (Figure 1). Although eGFR reporting did not have an effect on rates of microalbumin testing immediately after implementation of eGFR reporting (P=.29-.50), we found long-term increases in total and first microalbumin testing rates across the VHA (P<.001). In contrast, overall and first kidney ultrasound facility rates increased approximately 2 patients per 10,000 general population per month (P=.01-.04) and appeared to be driven by first ultrasound testing among patients between 60 and 79 years old (mean increase= 3.37-4.60, P=.001-.002). However, the rate of increase in ultrasound testing did not change following eGFR reporting in the longer-term (P=.35-.39).
Lastly, there was no association between estimated eGFR reporting and changes in general rates of overall and first outpatient visits with nephrologists (Figure 1). However in subanalyses, rates of first nephrology visits among patients 60 years and older (p<.05) and women (p= .01) increased (Table 3).
Discussion
This is the first study to examine the system-level impacts of automated eGFR laboratory reporting on changes in CKD recognition in the largest integrated health care system in the US that serves a patient population particularly vulnerable to CKD. While most prior studies examined eGFR reporting impacts on nephrology referrals and visits (8), we capitalized on the VHA's national electronic medical record to examine an expanded list of modalities of CKD recognition.
Our results suggest that eGFR reporting facilitated awareness and recognition of the presence of CKD in patients. Consistent with Wyatt's analysis of eGFR reporting in a single VA outpatient clinic (17), we found automated eGFR reporting was associated with immediate system-level increases in overall and first CKD diagnostic coding. It is possible that CKD recognition varied by different levels of kidney function, but we were unable validate and examine the appropriateness of CKD staging in documented diagnoses because we individual-level eGFR data could not be linked to patient records. However, there is no reason to believe that eGFR reporting would improve the accuracy of VA physician coding practices. And although the sensitivity of CKD diagnoses are generally poor, its specificity is usually high (18, 19).
Few prior studies evaluated the impact of eGFR reporting on the use of diagnostic testing, we found growth in short-term utilization rates of overall and first kidney ultrasound testing and longer-term increases in overall and first microalbumin testing among VAMC outpatients. Our system-level findings may demonstrate appropriate provision of care in the VA system. The short-term “bump” and long-term “decline” in first CKD diagnoses coupled with the “bump” and “incline” in first microalbumin testing (Figure 1) may reflect awareness of eGFR reported values and possible CKD, leading to confirmatory testing of CKD via microalbumin. Consequently, microalbumin results may indicate no further need for follow-up, which would then explain the modest impacts we found in rates of kidney ultrasound procedures and nephrology consultation. Further analyses conducted at the patient-level, tracking this potential confluence of outcomes in patients before and after implementation of automated eGFR reporting, are needed to confirm this explanation.
Considering the significant increase in rates of CKD identification through diagnostic coding and testing, the modest overall impact of eGFR reporting on VAMC nephrology visits was surprising. Our findings stand in contrast with several notable increases in nephrology referrals that occurred after eGFR reporting in other vertically integrated healthcare settings including, Canada (13, 20, 21), the United Kingdom (22), and Australia (23). Several factors may explain this difference. First, the VHA is a national system that provides varying amounts of care to a small segment of the US population, whereas publicly funded healthcare systems in other countries typically provide the majority of care to large segments of their populations. Accordingly, our study may not have captured all of the care episodes for VA patients who may have received disease management by generalist physicians (e.g., VA primary care physicians) or healthcare services outside of the VHA. Second, the VHA operates in a fixed resource environment that may leave local VAMCs with limited capacity to schedule follow-up nephrology consultations. Although resource constraints have been noted for some aspects of VA care (24), its contribution to effective evaluation of CKD is not known.
However, our subgroup analyses found important differences in eGFR reporting impacts on rates of the VA system's overall and first CKD outcomes that deserve greater attention. While many of the “all” nephrology consultations may reflect monitoring of patients with CKD who have already been referred to a nephrologist, it is important to highlight eGFR reporting effects on the more serious and important “first” visits. In particular, the greater eGFR impacts on increased rates of first nephrology visits among older veterans (aged 60 and older) – are an important patient subpopulation that may have been previously under-diagnosed and untreated for CKD.
To advance CKD care beyond greater CKD recognition alone, appropriate follow-up evaluations and consultation are often necessary to confirm the presence and severity of CKD, and to develop appropriate strategies for disease management (25). Altogether, our results identify modest system-level responsiveness after automated eGFR reporting that may reflect the relative ease of CKD documentation in patient medical records and follow-up diagnostic evaluation. However, we did not identify a strong and consistent system-level response with increases in nephrology consultation, suggesting that the impact on disease management may have been limited, and signaling opportunities for improvement in management of CKD across the VHA. The extent to which such lessons apply to other healthcare systems remains unknown, but plausible. Without greater improvements in the recognition and specialty management of CKD, healthcare systems risk hastening patient progression to life-threatening and costly outcomes including kidney failure and cardiovascular events.
Despite the strengths of examining the system-level impact of eGFR reporting in over 100 centers across the US, our study has some limitations. First, the study was limited to outpatient care financed by the VA. To the extent that patients had limited access to VA services, received early-stage disease management from VA primary care providers, or received diagnostic tests and nephrology consultation from non-VA providers, our results may underestimate the impact of eGFR reporting on CKD detection. Second, when evaluating facility-level nephrology visits, we were unable to distinguish between appropriate and inappropriate referrals (26) or evaluate wait times to assess whether VAMC resource constraints limited access to nephrologists. Finally, it is unclear whether identification of CKD by diagnostic code results in improved outcomes such as timely referral to specialty care or appropriate medication dosing and use. Future research may examine eGFR's impacts on the disease management and treatment at the patient-level.
In summary, implementation of eGFR laboratory reporting in the largest integrated healthcare system in the US represents a promising strategy to increase CKD detection. However, additional steps will be necessary for such improvements in disease recognition to translate to improvements in CKD prevention and disease management. Future research should examine the extent to which healthcare providers prioritize and act upon laboratory-related clinical information in their treatment decisions. Given the high burden of CKD among veterans, greater efforts to identify and provide timely treatment to those with CKD will be necessary to effectively reduce the large burden of disease and its associated complications.
Acknowledgments
Funding: K23DK075929-04W1
References
- 1.Huisman RM. The deadly risk of late referral. Nephrol Dial Transplant. 2004;19:2175–2180. doi: 10.1093/ndt/gfh409. [DOI] [PubMed] [Google Scholar]
- 2.Weiner DE, Tighiouart H, Amin MG, et al. Chronic kidney disease as a risk factor for cardiovascular disease and all-cause mortality: a pooled analysis of community-based studies. J Am Soc Nephrol. 2004;15:1307–1315. doi: 10.1097/01.asn.0000123691.46138.e2. [DOI] [PubMed] [Google Scholar]
- 3.Wright JT, Jr., Bakris G, Greene T, et al. Effect of blood pressure lowering and antihypertensive drug class on progression of hypertensive kidney disease: results from the AASK trial. Jama. 2002;288:2421–2431. doi: 10.1001/jama.288.19.2421. [DOI] [PubMed] [Google Scholar]
- 4.Drey N, Roderick P, Mullee M, et al. A population-based study of the incidence and outcomes of diagnosed chronic kidney disease. Am J Kidney Dis. 2003;42:677–684. doi: 10.1016/s0272-6386(03)00916-8. [DOI] [PubMed] [Google Scholar]
- 5.Hostetter TH, Lising M. National kidney disease education program. J Am Soc Nephrol. 2003;14:S114–116. doi: 10.1097/01.asn.0000070156.78824.c7. [DOI] [PubMed] [Google Scholar]
- 6.Levey AS, Eckardt KU, Tsukamoto Y, et al. Definition and classification of chronic kidney disease: a position statement from Kidney Disease: Improving Global Outcomes (KDIGO). Kidney international. 2005;67:2089–2100. doi: 10.1111/j.1523-1755.2005.00365.x. [DOI] [PubMed] [Google Scholar]
- 7.National Kidney F. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis. 2002;39:S1–266. [PubMed] [Google Scholar]
- 8.Kagoma YK, Weir MA, Iansavichus AV, et al. Impact of estimated GFR reporting on patients, clinicians, and health-care systems: a systematic review. Am J Kidney Dis. 2011;57:592–601. doi: 10.1053/j.ajkd.2010.08.029. [DOI] [PubMed] [Google Scholar]
- 9.Arora P, Rajagopalan S, Patel N, et al. The MDRD equation underestimates the prevalence of CKD among blacks and overestimates the prevalence of CKD among whites compared to the CKD-EPI equation: a retrospective cohort study. BMC nephrology. 2012;13:4. doi: 10.1186/1471-2369-13-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Hall RK, Wang V, Jackson GL, et al. Implementation of automated reporting of estimated glomerular filtration rate among veterans affairs laboratories: a retrospective study. BMC Med Inform Decision Making. 2012;12:69. doi: 10.1186/1472-6947-12-69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Patel TG, Pogach LM, Barth RH. CKD screening and management in the Veterans Health Administration: the impact of system organization and an innovative electronic record. Am J Kidney Dis. 2009;53:S78–85. doi: 10.1053/j.ajkd.2008.07.051. [DOI] [PubMed] [Google Scholar]
- 12.Levey AS, Coresh J, Greene T, et al. Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Annals of internal medicine. 2006;145:247–254. doi: 10.7326/0003-4819-145-4-200608150-00004. [DOI] [PubMed] [Google Scholar]
- 13.Hemmelgarn BR, Zhang J, Manns BJ, et al. Nephrology visits and health care resource use before and after reporting estimated glomerular filtration rate. JAMA. 2010;303:1151–1158. doi: 10.1001/jama.2010.303. [DOI] [PubMed] [Google Scholar]
- 14.Jain AK, Cuerden MS, McLeod I, et al. Reporting of the estimated glomerular filtration rate was associated with increased use of angiotensin-converting enzyme inhibitors and angiotensin-II receptor blockers in CKD. Kidney international. 2012;81:1248–1253. doi: 10.1038/ki.2012.18. [DOI] [PubMed] [Google Scholar]
- 15.VA Information Resource Center (VIReC) VHA Medical SAS® Outpatient Datasets and Inpatient Encounters Dataset FY2009: VIReC Research User Guide. U.S. Department of Veterans Affairs, Health Services Research and Development Service; Hines, IL: 2011. [Google Scholar]
- 16.Wagner AK, Soumerai SB, Zhang F, et al. Segmented regression analysis of interrupted time series studies in medication use research. Journal of clinical pharmacy and therapeutics. 2002;27:299–309. doi: 10.1046/j.1365-2710.2002.00430.x. [DOI] [PubMed] [Google Scholar]
- 17.Wyatt C, Konduri V, Eng J, et al. Reporting of estimated GFR in the primary care clinic. Am J Kidney Dis. 2007;49:634–641. doi: 10.1053/j.ajkd.2007.02.258. [DOI] [PubMed] [Google Scholar]
- 18.Kern EF, Maney M, Miller DR, et al. Failure of ICD-9-CM codes to identify patients with comorbid chronic kidney disease in diabetes. Health services research. 2006;41:564–580. doi: 10.1111/j.1475-6773.2005.00482.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Vlasschaert ME, Bejaimal SA, Hackam DG, et al. Validity of administrative database coding for kidney disease: a systematic review. Am J Kidney Dis. 2011;57:29–43. doi: 10.1053/j.ajkd.2010.08.031. [DOI] [PubMed] [Google Scholar]
- 20.Akbari A, Grimshaw J, Stacey D, et al. Change in appropriate referrals to nephrologists after the introduction of automatic reporting of the estimated glomerular filtration rate. CMAJ. 2012;184:E269–276. doi: 10.1503/cmaj.110678. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Jain AK, McLeod I, Huo C, et al. When laboratories report estimated glomerular filtration rates in addition to serum creatinines, nephrology consults increase. Kidney international. 2009;76:318–323. doi: 10.1038/ki.2009.158. [DOI] [PubMed] [Google Scholar]
- 22.Richards N, Harris K, Whitfield M, et al. The impact of population-based identification of chronic kidney disease using estimated glomerular filtration rate (eGFR) reporting. Nephrol Dial Transplant. 2008;23:556–561. doi: 10.1093/ndt/gfm839. [DOI] [PubMed] [Google Scholar]
- 23.Noble E, Johnson DW, Gray N, et al. The impact of automated eGFR reporting and education on nephrology service referrals. Nephrol Dial Transplant. 2008;23:3845–3850. doi: 10.1093/ndt/gfn385. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Krein SL, Bernstein SJ, Fletcher CE, et al. Improving eye care for veterans with diabetes: an example of using the QUERI steps to move from evidence to implementation: QUERI Series. Implement Sci. 2008;3:18. doi: 10.1186/1748-5908-3-18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Levin A, Stevens PE. Early detection of CKD: the benefits, limitations and effects on prognosis. Nature reviews. 2011;7:446–457. doi: 10.1038/nrneph.2011.86. [DOI] [PubMed] [Google Scholar]
- 26.Abdel-Kader K, Fischer GS, Johnston JR, et al. Characterizing pre-dialysis care in the era of eGFR reporting: a cohort study. BMC nephrology. 2011;12:12. doi: 10.1186/1471-2369-12-12. [DOI] [PMC free article] [PubMed] [Google Scholar]