Five: Costs of chronic kidney disease
The economic impact of chronic kidney disease on the healthcare system poses challenges on several levels. First, as noted earlier, the case definition is dependent on the reported data. A biochemical definition would be the most quantitative, but health plan datasets rarely contain this information on a large scale. Defining the CKD cohort from diagnosis codes, however, may represent only the more advanced — and therefore most expensive — cases. In addition, CKD is a highly interactive disease, associated with cardiovascular disease, stroke, and infectious complications. Given these limitations, the USRDS has developed a method based on diagnosis codes, identifying point prevalent CKD patients in the prior year, and adding new CKD cases — defined from the same coding profile — in the current year. This is similar to the period prevalent cohort used to assess costs of the ESRD population in Volume Two.
Trends in CKD costs over time are vulnerable to changes in recognition of the disease. According to the random sample estimates from NHANES (described in Chapter One), the Medicare population with CKD of any stage is expected to be nearly 40 percent of those age 65 and older. Over the last decade, CKD prevalence, as shown in the NHANES data, has grown approximately 20–25 percent. Trend data on the next spread show that the recognized CKD population within Medicare increased from 3.1 percent of the population in 1996 to 8.6 percent in 2006 — nearly a three-fold increase. This observation alone shows that increased recognition of the disease is playing a role in associated costs. Expenditures associated with CKD have more than doubled, from 11.3 percent of the total in 1996 to 24.5 percent in 2006.
These assessments, however, likely underestimate the true number and cost of those with CKD. In contrast to the ESRD population, the CKD population lacks a patient registration system to accurately define the number of cases. And the new CKD diagnosis codes, introduced in 2006, have led to a 20 percent increase in recognized CKD since 2005. We define CKD from both inpatient/outpatient services and physician/supplier codes, requiring outpatient codes to appear at least two times to ensure the presence of the disease; this is in contrast to the single code from any source used in the Medicare hierarchical condition category (HCC) model. Data from both the HCC and standard actuarial models are reported at the end of the chapter.
Also shown on the next page, the Employer Group Health Plan (EGHP) data — from a younger population — use a similar definition. The recognized CKD population here is low, at 1.3 percent, and associated costs are equal to those associated with recognized heart failure. This population is again considerably smaller than that of the population-level estimates from NHANES, suggesting that data under-represent the true impact of CKD in the private health plans.
Overall per person per month costs for CKD are $2,311 for dually-enrolled patients, compared to $1,902 for Medicare overall and $2,609 for the younger EGHP patients in the Medstat dataset. The higher EGHP payments may reflect cost shifting from the Medicare program, as well as the lower ability of private payors to set fees compared to Medicare.
As illustrated earlier, CKD is highly interactive with other chronic diseases. Expenditures in patients with CKD, diabetes, and congestive heart failure are more than twice those of patients with CKD alone, at $2,973 compared to $1,241. And in the dually-enrolled and EGHP populations, costs for those with each of the three diseases rise to $3,281 and $5,480, respectively. These data illustrate the “multiplier” impact of CKD on healthcare services and expenditures. From this perspective, and noting the data on event rates and the monitoring of risk factors in prior chapters, it appears that identification of CKD by simple biochemical tests would provide an opportunity for providers to target at-risk populations for active interventions to control salt intake, weight, blood pressure, lipid disorders, and the progression of kidney disease, particularly in Stages 4–5.
Expenditures during the transition from CKD to ESRD are considerable, with the most striking occurring in the month of dialysis initiation — nearly $14,500 for Medicare patients, $15,100 for those with dual enrollment, and $29,000 in the EGHP population. PPPM expenditures in the first six months on dialysis are twice as high for EGHP patients than for Medicare patients, a finding that also may reflect pricing issues. Medicare sets prices for dialysis procedures and ancillary services, while private health plans, with few providers and accounting for only 15 percent of patients under treatment, have little leverage to negotiate pricing. Patients with commercial coverage can thus generate considerably more margin for dialysis providers, offsetting costs of the Medicare population or increasing profits for the large commercial dialysis chains.
Health plans traditionally use an as-treated actuarial expenditure method to determine the service distribution of covered populations, and to set expectations for annual budgets. Medical DRG hospital services dominate expenditures for Medicare patients, while surgical DRGs are most common for the younger EGHP population. The nature of these differences will be further explored in the 2009 ADR. Not unexpectedly, prescription drugs costs are noted in the outpatient setting in the EGHP population; Medicare Part D prescription costs, however, are not yet available for comparison. And since costs for skilled nursing and home health services are considerably higher in the older Medicare populations, it can be inferred that primary differences in CKD costs between Medicare and EGHP center on prescription medications and hospitalizations.
We conclude the chapter by comparing actual PPPM expenditures with the Average Annual Per Capita Cost (AAPCC) payments, which minimally adjust for age, gender, and disability status, and the new Medicare HCC disease group payments for health plans, used with the Medicare Advantage HMO population. We show the increasing interaction of CKD with diabetes and CHF to determine how closely the new HCC payment system reflects actual expenditures. The AAPCC payment system generally underpaid for sicker populations and overpaid for the less complex groups, a finding that has been dramatically reduced by the HCC system. These improvements can be seen not only in comparisons of monthly payments, but in the predictive ratio, addressing the relative matching of actual versus modeled payments. The improved matching of the payment model based on detailed disease case definitions from claim diagnosis codes provides support for the clinical linking of CKD, diabetes, and CHF.
Overall, the impact of CKD on the healthcare system appears to be large, and highly interactive with major diseases associated with kidney failure, such as diabetes and cardiovascular disease. In future ADRs we will further characterize the complex clinical and economic impact of CKD across many organ system domains.
figure 5.1 Medicare: period prevalent patients, age 65 & older, without ESRD. Medstat & Ingenix i3: period prevalent patients, age 50–64, enrolled in a fee-for-service plan, & without ESRD. CHF, diabetes, & CKD determined from claims, & costs are for calendar year 2006.
figure 5.2 In 2006, costs for Medicare patients with CKD exceeded $49 billion — nearly five times greater than costs in 1993. Overall Medicare expenditures, in contrast, have grown only 91 percent in the same period. Expenditures for patients with CKD now account for nearly one-quarter of Medicare spending. With the 2005 introduction of the new ICD-9-CM diagnosis codes for CKD, the recent sharp rise in CKD’s percentage of the Medicare program may in part reflect greater numbers of patients with recognized CKD. Total Medstat expenditures in 2006 reached $15.6 billion, a seven-fold increase since 2000. Medstat CKD costs are now nearly eleven times greater than in 2000, at $1.2 billion. And CKD’s percentage of the overall Medstat costs has grown from 5.1 to 7.7.
figure 5.3 Overall PPPM costs in the CKD population are greatest for those in the Medstat program, at $2,609 in 2006 — an increase of 31 percent since 2000. Costs for Medicare and dually-enrolled (Medicare/Medicaid) patients have grown 46 and 63 percent, respectively, to reach $1,902 and $2,311.
figure 5.4 Data on PPPM expenditures by diagnosis show the multiplier effect of diabetes and congestive heart failure (CHF) in patients with CKD. In the Medicare population, for example, PPPM costs in 2006 were $1,326 for patients with CKD and diabetes, but rose to $2,973 in patients with CHF as well. Costs are slightly higher in the dually-enrolled (Medicare/Medicaid) population, at $1,613 for those with CKD and diabetes, and $3,281 for those who also have CHF. In the EGHP population, however, costs for patients with all three diagnoses rise to $5,480, illustrating in part the smaller leverage health plans hold to negotiate pricing.
figures 5.5, 5.6, & 5.7 In the month following ESRD initiation, per member per month (PPPM) expenditures are similar for Medicare and dually-enrolled (Medicare/Medicaid) patients age 67 and older, at $14,461 and $15,112. Costs for younger patients in the Medstat database, in contrast, are nearly twice as high, at $28,718. After a sharp decrease in the next month, and more gradual changes in the following months, costs at six months reach $6,300–$6,900 for Medicare and dually-enrolled patients, and $12,561 for Medstat patients, still nearly two times greater. A similar pattern is seen for inpatient costs during the transition to ESRD, with Medstat PPPM costs reaching $20,792 in month of initiation, compared to $9,588 and $9,924 for Medicare and dually-enrolled patients, respectively. These differences, however, lessen by month six, with Medstat PPPM costs now 1.1–1.2 times greater than those for patients covered by Medicare. PPPM cardiovascular hospitalization costs during the transition also spike in the first month, reaching $6,885 for patients in the Medstat database, and approximately $3,200–$3,500 for those who are dually-enrolled or under Medicare coverage. By month six, these costs are lowest for Medstat patients, at $269, compared to $617 and $559, respectively.
table 5.a The net cost for medical DRGs is highest for dually-enrolled patients, while EGHP patients have the highest net cost for surgical DRGs. Total inpatient PPPM net costs are similar for the three populations. For EGHP patients, outpatient net costs total $1,038 PPPM — significantly higher than for Medicare and dually-enrolled (Medicare/Medicaid) patients. Within the EGHP population, dialysis and EPO account for about 15 percent of net outpatient costs, while pharmacy accounts for around 40 percent. Skilled nursing facility, home health, and hospice net costs are much higher for the Medicare and dually-enrolled populations than for the EGHP population, probably because of the age of each population and the fact that EGHP patients are able to work, and are therefore unlikely to require this type of care. Total physician/supplier net costs are similar in the three groups. In each case, prescription drugs account for a significant portion of the cost, totaling as much as 15 percent for EGHP patients. Inpatient and outpatient surgery also account for a large percentage of total net physician/supplier costs, at approximately 15 percent for Medicare patients, 11 percent for those with dual Medicare/Medicaid coverage, and 24 percent for the EGHP population.
figures 5.8, & 5.9 Net PPPM costs for CKD patients have increased slowly since 1993 — not a surprising finding, since costs rise as the comorbidity burden increases. Within the EGHP population, the presence of CHF produces a significant increase in net PPPM costs, much more so than for Medicare and dually enrolled patients. This increase for EGHP patients is seen in both the inpatient/outpatient and physician/ supplier net costs, and does not seem to differ based on diabetic status.
tables 5.b, 5.c & figures 5.10, 5.11, & 5.12 Data in Table 5.b show actual and predicted Medicare costs per member per month (PPPM) for 2006 Medicare non-CKD and CKD patients. The predicted cost based on Medicare AAPCC shows larger differences from actual costs, especially for non-CKD patients and for CKD patients with diabetes and CHF. As shown in Table 5.c, the ratios of AAPCC costs to actual costs range from 1.37 (age 85 and older) to 2.39 (age 65–69) for non-CKD patients; for CKD patients with diabetes and congestive heart failure, they range from 0.24 (younger than 65) to 0.47 (age 70–74). Predicting costs with the CMS-HCC model significantly reduces differences from the actual costs. As shown in Table 5.c, the ratios of costs from the CMS-HCC model to actual costs range from 1.00 (age 85 and older) to 1.18 (younger than 65) for non-CKD; for CKD patients with diabetes and CHF, they range from 1.00 (younger than 65) to 0.84 (age 80–84). Figures 5.10–11 illustrate, by age, annualized actual costs and predicted costs based on the CMS-HCC model for 2006 Medicare non-CKD and CKD patients. The lowest actual Medicare cost occurs in non-CKD patients younger than 65, at $7,118 annually, while the highest, of $46,513, is found in those age 80–84 and with CKD, diabetes, and CHF. Compared to non-CKD patients younger than 65, the 2006 actual Medicare costs for CKD patients with comorbidities increased 146–447 percent, and 73–349 percent for those age 80–84. Figure 5.12 shows that, for CKD patients, predicted costs based on the 2007 CMS-HCC model are slightly lower than those based on the 2006 CMS-HCC model.
figures 5.2–4 Medicare: period prevalent general Medicare patients age 65 & older, with Medicare as primary payor, & not enrolled in Medicare Advantage. Medstat: period prevalent patients age 50–64, enrolled in a fee-for-service plan. CHF, diabetes, & CKD determined from claims. Dually-enrolled: a subset of the Medicare population; includes patients dually-enrolled for the entire cost-year follow-up time. figures 5.5–7 Medicare: patients age 67 & older, initiating ESRD therapy in 2005, & with Medicare as primary payor (not enrolled in Medicare Advantage). Dually-enrolled: a subset of the Medicare population; includes patients dually-enrolled for the entire transition period. Medstat: patients younger than 65, initiating ESRD therapy in 2005, & enrolled for the entire transition period. In Figure 5.7, hospitalization cause determined using principal diagnosis code for all groups.
table 5.a Medicare: period prevalent general Medicare patients age 65 & older, with Medicare as primary payor, & not enrolled in Medicare Advantage. Dually-enrolled: a subset of the Medicare population; includes patients dually-enrolled for the entire cost-year follow-up time. Medstat: period prevalent patients age 50–64, enrolled in a fee-for-service plan. CHF, diabetes, & CKD determined from claims, & PPPM costs are for calendar year 2006, with actuarial categories determined by place of service; CPT, revenue, & HCPCS codes; & provider specialty. Oral prescription drugs (not present for Medicare) are included in the outpatient pharmacy category.
figures 5.8–9 populations: same as in Table 5.a. CKD, diabetes, & CKD determined from claims. Figure 5.8: costs are PPPM inpatient plus outpatient costs, & do not include skilled nursing facility, home health, or hospice. Figure 5.9: costs are PPPM physician/supplier.
tables 5.b–c & figures 5.10–12 prevalent general Medicare patients, not enrolled in an HMO, without ESRD, & surviving in 2005. Medicare payments in 2006 include inpatient, skilled nursing, home health, outpatient, physician/supplier, & durable medical equipment. Diagnosis groups based on M+C risk adjustment model (see www.cms.gov). Actual PPPM cost not weighted by follow up time. Definitions for CKD, diabetes, & CHF are based on the CMS-HCC model, i.e., defined if there is any diagnosis code. HCC: hierarchical condition category.