Clinical outcomes in pediatric hemodialysis patients in the USA: lessons from CMS’ ESRD CPM Project
- First Online:
- Cite this article as:
- Neu, A.M. & Frankenfield, D.L. Pediatr Nephrol (2009) 24: 1287. doi:10.1007/s00467-008-0831-0
- 270 Downloads
Although prospective randomized trials have provided important information and allowed the development of evidence-based guidelines in adult hemodialysis (HD) patients, with approximately 800 prevalent pediatric HD patients in the United States, such studies are difficult to perform in this population. Observational data obtained through the Center for Medicare & Medicaid Services’ (CMS’) End Stage Renal Disease (ESRD) Clinical Performance Measures (CPM) Project have allowed description of the clinical care provided to pediatric HD patients as well as identification of risk factors for failure to reach adult targets for clinical parameters such as hemoglobin, single-pool Kt/V (spKt/V) and serum albumin. In addition, studies linking data from the ESRD CPM Project and the United States Renal Data System have allowed evaluation of associations between achievement of those targets and the outcomes of hospitalization and death. The results of those studies, while unable to prove cause and effect, suggest that the adult ESRD CPM targets may assist in identifying pediatric HD patients at risk for poor outcomes.
KeywordsPediatricHemodialysisOutcomesClinical performance measuresCPMHospitalizationMortality
Although large-scale, prospective randomized studies in pediatric HD patients are lacking, there are several observational studies that have provided important information regarding the care provided to pediatric ESRD patients. An example is the Centers for Medicare & Medicaid Services’ (CMS’) ESRD Clinical Performance Measures (CPM) Project. Since 2000, the ESRD CPM Project has collected clinical data on pediatric HD patients and has provided important information about the care provided to pediatric HD patients in the United States [5–12]. In addition, several studies have linked the intermediate outcomes collected by the ESRD CPM Project with administrative data from the USRDS in an effort to demonstrate associations between the attainment of certain clinical targets (such as a hemoglobin ≥ 11 g/dl) and morbidity and mortality [13–15]. Although these studies have added significantly to the information available to those who care for children on HD, it is important to understand the limitations of these data—perhaps most importantly the inability of these studies to demonstrate causality. The purpose of this article is to review data available on pediatric HD patients from the ESRD CPM Project and studies linking data from the ESRD CPM Project with USRDS administrative data. To elucidate the strengths and limitations of these data, the history and structure of the ESRD CPM Project will be reviewed briefly. A detailed history of the project can be found on the CMS Web site (www.cms.hhs.gov/CPMProject).
History and background of CMS’ ESRD CPM project
All patients with ESRD in the United States, including children, have been eligible to receive Medicare coverage since 1972. As part of its oversight of the Medicare program, CMS contracts with 18 ESRD network organizations throughout the United States to monitor the quality of care for ESRD patients and to facilitate improvements in that care. Since 1994, this quality assurance effort has included collection of clinical parameters in the areas of anemia management, dialysis adequacy, and serum albumin [5, 16]. In 1997, CMS funded the development of Clinical Performance Measures (CPMs) to be based on the National Kidney Foundation (NKF) Dialysis Outcomes Quality Initiative (DOQI) Guidelines [17–20]. Sixteen ESRD CPMs were developed that have subsequently been reduced to 13: three for HD adequacy, three for peritoneal dialysis (PD) adequacy, three for vascular access, and four for anemia management . A full description of the CPMs is presented on the CMS Web site (www.cms.hhs.gov/CPMProject) . Similar to the KDOQI guidelines on which they are based, the ESRD CPMs establish targets for the frequency of monitoring the various clinical parameters, as well as thresholds or targets for the parameters themselves. That is, within the anemia-management CPMs, the target level for hemoglobin is specified, as is the frequency of monitoring serum transferrin saturation (TSAT) and serum ferritin and the target values for serum TSAT and serum ferritin . In addition to collecting information necessary to calculate the ESRD CPMs, the project also collects information on other aspects of dialysis care. In particular, although measurement of or obtaining a certain level of serum albumin is not a CPM, these data are felt to be an important indicator of care, so they are collected and reported by the ESRD CPM Project . It should be noted that the ESRD CPMs are for adult dialysis patients, and at present, there are no pediatric ESRD CPMs .
Data collection for ESRD CPM Project
The data collection for the ESRD CPM Project occurs in the spring of each study year, when a five-page data collection form is distributed to all facilities that have been identified by the 18 ESRD networks as having a pediatric (age < 18 years) patient receiving in-center HD on December 31 of the preceding year. The dialysis facilities provide clinical information abstracted from patients’ medical records for the months of October, November, and December of the preceding year. That is, for the 2006 CPM study year, data are abstracted from October through December 2005 on any pediatric patient alive and on in-center HD as of 31 December 2005. Demographic information provided by the facility includes gender, race, ethnicity, age, primary cause of ESRD, and date of first dialysis. Information used to assess the dialysis care provided to these patients includes patient height, pre- and postdialysis weight, pre- and postdialysis blood urea nitrogen (BUN) values, dialysis session length, dialyzer KUf values, reported urea reduction ratios and reported Kt/V values, type of vascular access [catheter, arteriovenous fistula (AVF), arteriovenous graft (AVG)], blood-pump flow rates, monitoring of the access site for stenosis, monthly hemoglobin, TSAT, serum ferritin, and serum albumin values [with the associated laboratory method—bromcresol green (BCG) or bromcresol purple (BCP) , as these two methods have been shown to yield systematically different results], erythropoietin dose and route of administration, and iron prescription information. Although reported Kt/V values are collected on the data collection form, single-pool Kt/V (spKt/V) values presented in the annual report and used for subsequent analysis are calculated from the raw data points using the Daugirdis II formula . Height, weight, and body mass index (BMI) standard deviation scores (SDS) are calculated using the appropriate gender-/age-specific mean, SD and adjustment parameters for the national population derived from the Third National Health and Nutrition Examination Survey (NHANES III) (2000) of the National Center for Health Statistics . Although data are requested on all prevalent pediatric HD patients, to remain in the analysis, a patient must have the minimum data elements submitted, which include a completed data collection form with at least one monthly hemoglobin value, at least one paired pre- and postdialysis BUN value, and at least one serum albumin value over the 3-month study period .
General limitations and strengths of the ESRD CPM Project
Now that the process of the ESRD CPM Project has been described, before presenting the data from that project, it is important to reiterate some of the key points of the data collection process that highlight the strengths and weaknesses of the data and their applicability to pediatric HD patients in general. First and foremost, data collection occurs on a prevalent, not incident, population. That is, patients are not enrolled at the time of dialysis initiation and followed over time; rather, the ESRD CPM Project provides a snapshot of the care provided over a 3-month period to patients on dialysis at the time of data collection. It cannot be assumed that these 3 months accurately reflect the long-term status of the patient. The second general limitation of the data collection is that many factors that may impact the ability to attain the various threshold values, such as adherence to prescribed oral therapies, adherence to dialysis treatments, markers of inflammation beyond serum albumin and serum ferritin, and nutritional assessment beyond height, weight and serum albumin, are not collected by the ESRD CPM Project and therefore any influence cannot be evaluated or factored. Finally, the ESRD CPM Project focuses on intermediate outcomes of anemia management, vascular access, dialysis adequacy, and serum albumin and does not collect data on the more firm clinical outcomes of neurocognitive development, hospitalization, or mortality. Although data on linear growth, another clinical outcome in pediatric patients, are collected, data on factors that may affect linear growth, such as Tanner stage, calcium and phosphorous balance, and prescription of recombinant human growth hormone, have not been routinely collected . Finally, measures of quality of life are not obtained, although a special analysis to evaluate quality of life in pediatric dialysis patients in several ESRD networks is underway .
Although data on hospitalization and mortality are not available in the ESRD CPM Project, they are available in the United States Renal Data System (USRDS) . Thus, by linking the intermediate outcomes in the ESRD CPM Project with hospitalization and mortality data in USRDS, associations between achieving threshold values for the adult ESRD CPMs and hospitalization and mortality can be evaluated. A full description of the USRDS, including details of how to obtain data from the USRDS and ESRD CPM Project for research, is available in a published annual report and on the Internet (www.usrds.org) . It should be noted that although all ESRD patients in the United should have a file in USRDS, information in the USRDS on hospitalization are obtained from Medicare billing claims. Therefore, only patients for whom a bill was submitted to Medicare will have that hospitalization data captured .
The strengths of the ESRD CPM Project data set lie in the fact that it captures information on all prevalent pediatric HD patients, unlike data collection in adult dialysis patients, which occurs on a representative sample. Not only does the ESRD CPM Project allow evaluation of the universe of pediatric HD patients in each study year, but patients who remain on dialysis will have data submitted in sequential study years, allowing longitudinal assessment of clinical parameters in a subgroup of patients. Finally, data for the ESRD CPM Project are collected for monitoring clinical care, and outlying values are verified . This is in contrast to data obtained from the USRDS, for example, where data are abstracted from forms completed for administrative purposes and are not validated or verified .
Intermediate outcomes in pediatric HD patients
Clinical parameters in 743 pediatric (< 18 years) hemodialysis patients from the 2006 End-Stage Renal Disease Clinical Performance Measures (ESRD CPM) Project
Mean ± Standard Deviation
Percent equal to or greater than Threshold
1.58 ± 0.33
88% ≥ 1.2
11.5 ± 1.6 g/dl
68% ≥ 11 g/dl
29 ± 14%
74% ≥ 20%
471 ± 471 ng/ml
83% ≥ 100 ng/ml
3.9 ± 0.5/3.5 ± 0.5 g/dl (BCG/BCP)
80% ≥ 3.5/3.2 g/dl (BCG/BCP)
Factors remaining predictive of achieving threshold values for single-pool Kt/V (spKt/V), mean hemoglobin, and mean serum albumin in 433 adolescent hemodialysis (HD) patients after multivariate logistic regression analyses
95% Confidence Interval
Body surface area (highest quartile = referent)
Mean serum albumin ≥ 3.5/3.2 g/dl (BCG/BCP)
Increasing mean dialysis session length (min)
Increasing mean blood-pump flow rate (ml/min)
≥ 3.5/3.2 g/dl (BCG/BCP)
Duration of dialysis (2 + years = referent)
< 0.5 years
Increasing mean hemoglobin (g/dl)
≥ 11 g/dl
Lower mean erythropoietin dose (U/kg/dose)
Mean serum albumin ≥ 3.5/3.2 g/dl (BCG/BCP)
Together, these two studies identify factors that are associated with failure to reach target values for urea clearance, hemoglobin, and serum albumin. However, it must reiterated that data are obtained only from a 3-month period and may not, therefore, be reflective of the typical status of the patient. In fact, an analysis of longitudinal data revealed that among patients who had data submitted to two consecutive study years, a significant percentage of patients who failed to meet threshold values for spKt/V, serum albumin, and hemoglobin in the first study year exceeded those values in the following year . It is also important to reiterate that these data evaluate the likelihood that pediatric HD patients will achieve a threshold value and do not evaluate whether or not achievement of that threshold is, in fact, associated with an improvement in patient outcome.
Association between achievement of adult ESRD CPMs and clinical outcomes
Height and linear growth
Adjusted odds ratios of final logistic multivariate model predicting short stature in pediatric hemodialysis patients 10 to < 15 years old
Odds Ratio (95% Confidence Interval)
Female (vs male)
Black (vs white)
Hispanic (vs non-Hispanic)
Congenital/urologic cause of ESRD (vs acquired/other)
Years on dialysis (per 1 year)
Increase in nPCR (0.1 g/kg/day)
Adjusted odds ratios of final logistic multivariate model predicting short stature in pediatric hemodialysis patients 15 to < 18 years old
Odds Ratio (95% Confidence Interval)
Male (vs female)
Congenital/urologic cause of ESRD (vs acquired/other)
Years on dialysis (per 1 year)
Increase in spKt/V (0.1 increase)
Decrease in mean hemoglobin (1 g/dl)
In a more recent analysis, Gorman and colleagues evaluated linear growth, that is, change in height SDS, in 407 patients who had data submitted to two or more consecutive ESRD CPM study years . In multivariate analysis, younger age, longer duration on dialysis, and increasing normalized protein catabolic rate (nPCR) were associated with a worsening in height SDS over subsequent ESRD CPM study years [–0.20 height SDS/year (95% CI –0.3, –0.1) for < 13 years old vs ≥ 13 years old; –0.03 height SDS/year (95% CI -0.14, -0.08) for duration of dialysis ≥ 6 months vs < 6 months; –0.02 height SDS/year (95% CI –0.04, –0.001) for each 0.1 increase in nPCR] . Conversely, female patients and patients who had lower height SDS in their first year of data entry into the ESRD CPM Project had improvement in their height SDS over time [+0.10 height SDS/year (95% CI 0.02, 0.19) for girls vs boys; +0.03 height SDS/year (95% CI 0.01,0.05) for each 0.5 decrease in height SDS] . Etiology of ESRD, race, ethnicity, and achieving target values for hemoglobin, spKt/V, or serum albumin were not significantly associated with either an improvement or worsening in height SDS . These studies suggest that younger age and longer duration of dialysis are associated both with short stature and poor linear growth, even after controlling for multiple possible confounders [10, 12]. It should be reiterated, however, that many factors that might effect linear growth and which are necessary to interpret relationships between intermediate outcomes and linear growth, such as protein intake, use of recombinant growth hormone, Tanner stage, and bone age, are not collected by the ESRD CPM Project.
Hospitalization and mortality
To evaluate possible associations of the intermediate outcome data obtained by the ESRD CPM Project with the outcomes of hospitalization and mortality, several studies have linked data from the ESRD CPM Project and the USRDS [13–15]. In one study, data from the 2000 and 2001 ESRD CPM study years were linked with hospitalization data from USRDS through 29 January 2001 and mortality data through 15 October 2001 to evaluate associations between attainment of a spKt/V ≥ 1.2 and risk for hospitalization and mortality . Recall that these ESRD CPM study years included only adolescent (age 12 to < 18 years) HD patients . In this study, spKt/V values were available from the ESRD CPM Project on 613 adolescent HD patients, with repeated measures (i.e. values submitted to sequential study years) in 160 (26%) of those patients . Median follow-up for hospitalizations was 7.4 ± 3.8 months and 10.8 ± 2.9 months for mortality .
Unadjusted analysis suggested that a hemoglobin ≥ 11 g/dl was associated with a decreased risk for hospitalization (0.31 ± 0.76 hospitalizations per year at risk vs 0.45 ± 0.87 hospitalizations per year at risk for patients with hemoglobin < 11 g/dl; p = 0.0128). However, in multivariate analysis, this was no longer statistically significant (incidence-rate ratio 0.87; 95% CI 0.66, 1.15) . Evaluation of risk for hospitalization by hemoglobin subcategories also did not reveal significant differences .
Although these studies demonstrate a strong association between spKt/V and risk for hospitalization and hemoglobin and risk for mortality, respectively, again, these studies determine only associations and not cause and effect. Because the ESRD CPM data were collected after published KDOQI recommendations, it is assumed that values for spKt/V and hemoglobin below the recommended levels were unintentional . Recall, also, that the study of spKt/V evaluated delivered, and not prescribed, dose of dialysis . It is possible that factors that led to reduced delivered dialysis dose, such as shortened runs secondary to noncompliance or medical instability, may have also led to an increased risk for hospitalization . As for the increased risk for hospitalization seen with spKt/V between 1.4 and 1.6, it is also possible that factors that led to the higher spKt/V, either prescribed because of some clinical abnormality such as hyperkalemia or the result of patient factors such as lower volume of distribution of urea (due to lower body weight), also contributed to a higher risk for hospitalization . Similarly, factors that contribute to lower hemoglobin including inflammation, infection, malnutrition, and blood loss might also contribute to the higher mortality rates seen in these patients .
Although mortality and hospitalization rates remain unacceptably high in pediatric HD patients compared with the general pediatric population, death and hospitalization are infrequent events in pediatric HD patients compared with the experience in adult HD patients [4, 25]. With fewer than 800 prevalent United States pediatric HD patients, it is unlikely that meaningful prospective randomized studies evaluating the impact of intermediate outcomes, such as dialysis dose and hemoglobin, on the firmer outcomes of linear growth, hospitalization, and death will ever be conducted in pediatric patients. Therefore, observational studies such as the ESRD CPM Project are important to try to define risk factors for poor outcomes. These studies suggest that the majority of prevalent pediatric HD patients in the United States achieve the threshold values established by the adult ESRD CPMs. However, data from the ESRD CPM Project repeatedly demonstrate that despite controlling for multiple confounders, black race is a significant risk factor for failing to meet the threshold spKt/V of 1.2, a finding that has also been demonstrated in a cohort of pediatric HD patients enrolled in the North American Pediatric Renal Trials and Collaborative Studies . Further study is needed to identify factors responsible for this phenomenon. Patients who fail to meet the threshold value for mean albumin are at risk for failing to reach the threshold value for mean hemoglobin and visa versa, suggesting that these thresholds can be used as markers for patients who are doing poorly. This is emphasized by studies demonstrating that failure to achieve an spKt/V of 1.2 is associated with an increased risk for hospitalization and that failure to achieve a hemoglobin of 11 g/dl is associated with an increased risk of death [13, 14]. Taken together, these data suggest that failure to reach the adult ESRD CPM targets may be a marker of poor outcomes and alert the physician to investigate causes for the failure to achieve them.