Association of Mortality and Hospitalization with Achievement of Adult Hemoglobin Targets in Adolescents Maintained on Hemodialysis
Sandra Amaral*,
Wenke Hwang,
Barbara Fivush*,
Alicia Neu*,
Diane Frankenfield and
Susan Furth*,
* Department of Pediatrics, Welch Center for Prevention, Epidemiology and Clinical Research, Johns Hopkins Medical Institutions, and Centers for Medicare & Medicaid Services, Office of Clinical Standards and Quality, Baltimore, Maryland; and Wake Forest University School of Medicine, Winston-Salem, North Carolina
Address correspondence to: Dr. Sandra Amaral, Emory University School of Medicine, Division of Pediatric Nephrology, 2015 Uppergate Drive NE, Atlanta, GA 30322. Phone: 404-727-2450, Fax: 404-727-8213; E-mail: sandra_amaral{at}oz.ped.emory.edu
Received for publication November 22, 2005.
Accepted for publication July 10, 2006.
With the use of data from the Centers for Medicare & MedicaidServices’ ESRD Clinical Performance Measures Project (Octoberthrough December 1999 and 2000) linked with US Renal Data Systemhospitalization and mortality records, whether achieving adulttarget hemoglobin (Hb) levels in adolescents who are on hemodialysis(HD) was associated with decreased risk for death or hospitalizationwas assessed. Of 677 adolescents, 238 were hospitalized and54 died. In bivariate analysis, 11.7% with Hb <11 g/dl atstudy entry died versus 5% of those with initial Hb 11 g/dl(P = 0.001); 40.3% with baseline Hb <11 g/dl were hospitalizedversus 31.1% with initial Hb 11 g/dl (P = 0.013). In multivariateanalysis, Hb 11 g/dl was associated with decreased risk fordeath (hazard ratio [HR] 0.38; 95% confidence interval [CI]0.20 to 0.72) but did not show a statistically significant associationwith decreased risk for hospitalization (HR 0.87; 95% CI 0.66to 1.15). When Hb was recategorized as Hb <10, 10 and <11,11 and 12, and >12 g/dl, risk of mortality declined as Hblevel increased. At Hb 11 to 12 g/dl (versus Hb <10 g/dl),mortality risk decreased by 69% (HR 0.31; 95% CI 0.14 to 0.65).Risk for mortality was similar for Hb 11 to 12 and >12 g/dl.For hospitalization, no statistically significant differencein risk between Hb categories was found. This observationalstudy of adolescents who are on HD is consistent with adultliterature showing decreased mortality in patients who haveESRD and meet adult Hb targets. Further studies in the formof randomized, clinical trials are needed to assess optimalHb levels for adolescents who are on HD.
The National Kidney Foundation Dialysis Outcomes Quality Initiative(DOQI) practice guidelines first were established in 1997 tocreate standards for dialysis care, including anemia management.In 1999, some of these guidelines were adopted by the Centersfor Medicare & Medicaid Services as clinical performancemeasures (CPM), or therapeutic goals, for adult dialysis patients.The target range for hemoglobin (Hb) in adults is 11 to 12 g/dl.Studies in adults with ESRD have consistently demonstrated reducedrisk for death and hospitalization when Hb levels are 11 g/dl;a lowering of Hb by 1 g/dl has been associated with a 14% increasedrisk for mortality in an incident adult dialysis population(1–3). Higher Hb levels have been associated with improvedoxygen utilization, exercise capacity, cognitive ability, andcardiac function (4,5). Little evidence is available to assessthe relevance of adult target Hb levels in pediatric patientswith ESRD. This study is a preliminary analysis to examine therelationship between risk for death and hospitalization andthe achievement of adult targets for Hb in adolescent patientswho have ESRD and are on hemodialysis (HD). A secondary aimof this study was to assess differential risk for mortalityand hospitalization among smaller Hb subcategories: <10,10 to <11, 11 to 12, and >12 g/dl.
For this retrospective, cohort study, national data on all prevalentin-center HD patients who were aged 12 to 18 yr between Octoberand December 1999 and October and December 2000 were obtainedfrom the Centers for Medicare & Medicaid Services’ESRD CPM Project. This information then was linked with 3-yrhospitalization records (October 1999 through December 2002)and 4-yr mortality records (October 1999 through November 2003)of Medicare-eligible HD patients in the US Renal Data System(USRDS) by unique USRDS identification numbers.
The ESRD CPM Project collects demographic and clinical dataon measures that may have an impact on quality of dialysis care,including anemia management, serum albumin, vascular access,and dialysis adequacy (6). In the ESRD CPM Project, Hb was recordedmonthly during the periods of October through December 1999and 2000; up to three separate Hb measures were available foreach patient per year. The monthly levels were averaged andanalyzed as means in each year. Adolescents who were on HD werecategorized as failing to meet target when mean Hb was <11g/dl (versus Hb 11 g/dl). They then were subclassified as Hb<10, 10 and <11, 11 to 12, and >12 g/dl.
Clinical characteristics that were obtained from the CPM dataset were compared among the dichotomized Hb groups. Covariatesthat were examined included age as a continuous variable, race(white versus other), gender, ESRD cause (congenital/urologicversus other), time since most recent HD initiation (180 versus<180 d), access type (catheter versus graft/fistula), meanserum ferritin (800 versus <800 ng/ml), and mean serum albumin(3.5/3.2 versus <3.5/3.2 g/dl, as measured by the bromcresolgreen and bromcresol purple methods, respectively). The cutoffsfor dialysis vintage, serum ferritin, and serum albumin werechosen from previous evidence showing differential survivaland/or anemia correction between these levels (7–11).Exposure to iron (yes/no) and mean epoetin dosage (units/kgper wk) also were examined. Epoetin dosing was examined as acontinuous variable and by quartiles (150, 150 to 250, 251 to400, and >400 units/kg per wk). No data were available oniron dosing. In addition, presence of baseline hypertension(yes/no), estimated baseline GFR by Schwartz formula, and Medicaid/Medicareinsurance coverage data were obtained from the USRDS MedicalEvidence Form 2728 and were included in the analysis. Fewerthan 3% of patients had diabetes, and fewer than 1% were smokers;therefore, these traditional risk factors in the adult ESRDpopulation were not included.
Risk for death and hospitalization was analyzed by whether patientsachieved the minimum target Hb (mean Hb <11 versus11 g/dl)and then by Hb subcategories (<10, 10 and <11, 11 to 12,and >12 g/dl). Death and hospitalization data were obtainedfrom the USRDS patient and hospitalization files, respectively.2, Fisher exact test, ANOVA, and t test were used to comparedichotomized clinical characteristics and death or hospitalization(as yes/no) between baseline Hb <11 versus11 g/dl. NonparametricKaplan-Meier curves with log-rank tests and semiparametric Coxproportional hazards regression models were used to comparesurvival, adjusting for the aforementioned covariates. Hospitalizationrates were analyzed as total claims per time at risk using Poissonregression. The starting date for time at risk was either entrydate of the CPM cohort (October 1, 1999, or October 1, 2000)or entry date into USRDS if the patient entered the USRDS afterthe cohort entry date. The last reported death date and lasthospitalization claim were used as end points for time at riskfor mortality and hospitalization analyses to account for lagtime between hospital admission or death notification and entryinto the USRDS files. Propensity score matching also was implementedfor the mortality analysis. Two sensitivity analyses were performed.The first sensitivity analysis excluded any patients who receiveda transplant during the study period. The second sensitivityanalysis excluded any patients without reported Medicare orMedicaid coverage listed on Form 2728.
In both survival and longitudinal hospitalization analyses,patients were followed to the end of the follow-up period ordeath or censored at transplantation. For patients who had Hbdata in both ESRD CPM Project study years, clustering by USRDSidentification number was used to account for repeated measures.P < 0.05 was considered statistically significant. All datamanagement and analysis were conducted using Stata 8.0 software(StataCorp, College Station, TX).
Patient Characteristics
A total of 680 individual patients who were aged 12 to 18 yrwere identified as common to both the ESRD CPM and USRDS databases.Three patients were dropped from this merged data set secondaryto transplantation within the first 2 mo of data collection.Hb data were available for 424 patients in October through December1999 and 430 patients in October through December 2000. A totalof 177 patients remained on HD from October 1999 through December2000 and therefore had Hb information in both years of ESRDCPM data collection (Figure 1).
Figure 1. Study population subsets by time at risk.
Among the 677 included patients, the mean age was 16 yr (SD1.7); 52% were male, 49% were white, and 41% were black; 41%had congenital or urologic disease; 45% were on HD for <180d. At time of first entry, 44% of patients had Hb <11 g/dl,47% had catheters, 19% had serum albumin <3.5/3.2 g/dl (bromcresolgreen/bromcresol purple), 6% had serum ferritin 800 ng/ml, and34% had hypertension. Mean estimated GFR by Schwartz formulawas 9.4 ml/min per 1.73 m2 (SD 6.5). A total of 94% of patientswere prescribed epoetin, and 75% were prescribed iron. A totalof 50% of patients had Medicare/Medicaid coverage reported ontheir 2728 form at study entry; 45% of patients had an updated2728 completed within 3 mo of the study onset; 88% of the CPM2000cohort and 94% of the CPM2001 cohort had Medicare insurancefor the entire span of data collection as based on payer historyfiles. A total of 216 (32%) patients received a transplant duringthe study period.
Table 1 shows characteristics compared by Hb at study entry:<11 versus11 g/dl. The group with Hb <11 g/dl was morelikely to be on HD <180 d, have a catheter as vascular access,have lower serum albumin, and have baseline hypertension. Patientswith Hb <11 g/dl also received higher mean epoetin doses.
Table 1. Clinical characteristics of cohort at study entry, Hb <11 versus11 g/dl, N = 677a
In comparison with adults on HD in the CPM2000 and CPM2001 cohortsas described in the annual reports, adolescents who were onHD were more likely to be on HD for <6 mo and more likelyto have catheters as vascular access; these adolescents alsowere more likely to have Hb <11 g/dl (44 versus 32% adults).The adult HD CPM patients were more likely to have diabetes(40%) and had higher mean serum ferritin (489 to 529 ng/ml)(12,13).
During the designated follow-up period for hospitalization,238 (35%) of the 677 included patients were hospitalized. Hospitalizedpatients comprised 165 patients who had >3 yr of hospitalizationtime at risk (October 1999 through December 2002) and 73 patientswho were exclusively in CPM2001 and had >2 yr of hospitalizationtime at risk (October 2000 through December 2002; Figure 1).Mean follow-up time for hospitalization analysis was 1.7 yr(±0.9). Only 3.9% of hospitalizations had infection reportedas one of the top three diagnoses.
There were 54 deaths during a 4-yr mortality follow-up period.Of those deaths, 26 occurred in patients with Hb data exclusivelyin CPM2000, 18 deaths occurred in patients with data exclusivelyin CPM2001, and 10 deaths occurred in patients with Hb datafrom both periods (Figure 1). Causes of death on the ESRD deathnotification form included 14 cardiac events, six vascular events,10 infectious causes, eight other, and 16 unknown or missing.Mean follow-up time for mortality analysis was 2.1 yr (±1.3).
Bivariate Analysis Survival.
In bivariate analysis, 11.7% of patients with Hb <11 g/dlat study entry died versus 5% of those with initial Hb 11 g/dl(P = 0.001). Patients with Hb <11g/dl had five versus 1.9deaths per 100 patient-years in the Hb 11 g/dl group (P = 0.0009;Table 2). Of the covariates examined, dialysis vintage was theonly covariate that was associated with differential mortalityrates. Thirty-seven (10%) patients who were initiated on hemodialysis180 d before study entry died versus 17 (5.5%) patients whowere on HD <180 d (P = 0.033).
Table 2. Bivariate analysis: Risk for death and hospitalization, Hb <11 versus 11 g/dl, N = 677
Kaplan-Meier curves and log rank testing also were used to assesssurvival risk. Log rank tests demonstrated a statistically significantimproved survival in patients with mean Hb 11 g/dl (P = 0.0003).The only other covariate that demonstrated a statistically significantassociation with improved survival by log rank testing was serumalbumin 3.5/3.2 g/dl (P = 0.0247).
Hospitalization.
Of the 677 patients, 40.3% with Hb <11 g/dl at study entrywere hospitalized versus 31.1% of patients with Hb 11 g/dl (P= 0.013). Patients with Hb <11 g/dl averaged 0.45 hospitaladmissions per year at risk versus 0.31 visits per year at riskfor those with Hb 11 g/dl (P = 0.013; Table 2). A total of 39%of patients who were on HD for 180 d were hospitalized versus31% of those who were on HD <180 d (P = 0.024). A total of39% of patients with baseline Medicare/Medicaid coverage werehospitalized versus 28% of those without this coverage (P =0.009). Patients who received a transplant were less likelyto be hospitalized (20% versus 42% of patients who did not receivea transplant; P = 0.000). Patients with baseline hypertensionalso had decreased risk for hospitalization (28 versus 37% ofthose without hypertension; P = 0.004).
Multivariate Analysis Survival.
Cox proportional hazard regression models were constructed withadjustment for the aforementioned covariates (Table 3). In crudemodels, Hb 11 g/dl was associated with a 63% decreased riskfor mortality (hazard ratio [HR] 0.37; 95% confidence interval[CI] 0.21 to 0.65), and serum albumin 3.5/3.2 g/dl was associatedwith a 50% decreased risk for mortality (HR 0.50; 95% CI 0.27to 0.93). All epoetin quartiles showed a more than double increasedrisk for mortality in crude analysis; however, this associationseemed statistically significant only for epoetin 151 to 250units/kg per wk. The full adjusted model demonstrated decreasedrisk for death (HR 0.38; 95% CI 0.20 to 0.72) for patients withHb 11 g/dl After covariate adjustment, the association betweenalbumin and mortality no longer was statistically significant.All epoetin quartiles continued to show increased risk for mortality;however, only epoetin 151 to 250 units/kg per wk was statisticallysignificant. In both crude and adjusted models, epoetin as acontinuous variable showed no statistically significant associationwith mortality (HR 1.00; 95% CI 0.99 to 1.00). Dialysis vintagewas statistically significant in the full model: Patients whowere on HD 180 d were at greater risk for mortality (HR 1.91;95% CI 1.03 to 3.52). Because there were multiple covariatesand few events, propensity score matching also was performed.Adjusting for all covariates, propensity score matching yieldedan HR of 0.36 for decreased risk for mortality in patients withHb 11 g/dl (95% CI 0.19 to 0.67), consistent with the unadjustedand full models (Table 3).
Hospitalization.
Hospitalization was examined by hospital admissions per timeat risk using Poisson regression. In the crude model, Hb 11g/dl was associated with decreased risk for hospitalization(incidence rate ratio [IRR] 0.74; 95% CI 0.56 to 0.96). In unadjustedanalyses, mean serum albumin 3.5/3.2 g/dl was associated withdecreased risk for hospitalization (IRR 0.65; 95% CI 0.46 to0.92), and serum ferritin 800 ng/ml was associated with increasedrisk (IRR 1.21; 95% CI 1.00 to 1.47). All epoetin quartileswere associated with increased risk for hospitalization; however,the associations at quartiles 151 to 250 and >400 units/kgper wk (versus <150 units/kg per wk) were statistically significant(Table 4).
In multivariate analysis, although the point estimate suggesteddecreased risk for hospital admissions per time at risk in patientswith Hb 11 g/dl, this association was no longer statisticallysignificant (IRR 0.87; 95% CI 0.66 to 1.15). After adjustmentfor the aforementioned covariates, only epoetin 151 to 250 and>400 units/kg per wk (versus < 150 units/kg per wk) werestatistically associated with increased risk for hospitalization.
Additional Analyses
To assess the differential risk of mortality and hospitalizationamong smaller Hb subcategories, we classified Hb as <10,10 and <11, 11 to 12, and >12 g/dl and repeated the analyses(Tables 5 and 6). Using these classifications, in adjusted analyses,there was a trend toward both decreased hospitalization (HR0.95; 95% CI 0.85 to 1.07) and decreased mortality (HR 0.58;95% CI 0.42 to 0.81) as Hb level increased. Hb 11 to 12 g/dl(versus Hb <10 g/dl) was associated with a 69% decreasedrisk for mortality (HR 0.31; 95% CI 0.14 to 0.65). At Hb >12g/dl (versus Hb <10 g/dl), mortality risk was decreased by72% (HR 0.28; 95% CI 0.13 to 0.61). Kaplan-Meier curves forthe varying Hb groups showed a significant difference in survivalbetween Hb <10 g/dl and the other Hb categories (P = 0.0007,log rank; Figure 2). For hospitalization, differences amongHb groups were not statistically significant.
To address possible ascertainment bias from unreported claimsof patients without Medicare or Medicaid, we performed a sensitivityanalysis excluding patients who did not have Medicare/Medicaidat study entry. For the remaining 340 patients with baselineMedicare or Medicaid coverage, Hb 11 g/dl demonstrated a 70%decreased risk for mortality (IRR 0.30; 95% CI 0.13 to 0.71).Hb 11 to 12 versus Hb <10 g/dl showed an 85% decreased riskfor mortality (IRR 0.15; 95% CI 0.04 to 0.53). The hospitalizationanalysis continued to show no statistically significant differenceamong Hb groups.
Initial analyses censored patients at time of transplantation.Because patients who received a transplant may have representeda healthier subset of the adolescent ESRD population, a sensitivityanalysis was performed excluding those 216 patients. For bothsurvival and hospitalization analyses, the exclusion of patientswho received a transplant did not significantly change the pointestimates.
We also considered survival bias in the patients with repeatedmeasures. To examine this possibility, we repeated the analysisexcluding the 177 patients who had Hb data in both CPM cohorts.The results of mortality and hospitalization analyses were notsignificantly altered by this exclusion. In addition, we performed2 analysis to examine whether there was difference in the probabilityof survival or hospitalization in patients with repeated measureson the basis of whether their Hb increased, decreased, or stayedthe same over time. We were unable to detect a statisticallysignificant difference on the basis of direction of Hb changefor either mortality (P = 0.372) or hospitalization (P = 0.762).
To examine the contribution of inflammation toward the relationshipbetween anemia and mortality or hospitalization, we composeda model that was limited to the surrogates of inflammation thatwere available in our data set—mean serum albumin andferritin—and we adjusted for iron use and epoetin quartiles.Using this model, neither the association between Hb level andmortality nor the association between Hb level and hospitalizationwas significantly changed.
The National Kidney Foundation DOQI practice guidelines foranemia management were based primarily on adult studies. Thereis limited evidence to support use of these adult targets inpediatric patients who are on HD. Warady and Ho (14) demonstratedan association between baseline hematocrit <33% at 30 d afterinitiation of dialysis and increased risk for prolonged hospitalizationand death in incident ESRD patients who were younger than 18yr and from the North American Pediatric Renal Transplant CooperativeStudy (NAPRTCS) registry. We sought to assess whether meetingthe specific adult Hb target of 11 to 12 g/dl is associatedwith decreased hospitalization and mortality risk in prevalentadolescents who are on HD. Our study included the mean of Hbmeasures over 3 mo in each year and incorporated all prevalentin-center HD patients who were aged 12 to 18 yr in the ESRDCPM Project and USRDS, including patients who were cared forin nonpediatric centers and would not have been included inNAPRTCS. We looked at Hb as a dichotomized measure (<11 versus11 g/dl) and as a categorical variable (<10, >10 and <11,11 to 12, and >12 g/dl).
Our analysis demonstrated a consistent and strong positive associationbetween achievement of adult target Hb 11 g/dl and improvedsurvival in up to 4 yr of follow-up for adolescent HD patients.Hb 11 g/dl was associated with a 60 to 70% decreased risk formortality across multiple models and with propensity score matching.Poisson regression demonstrated a 12 to 28% decreased risk forhospitalization across varying models; however, this associationwas not statistically significant.
We also tried to assess differential risk for mortality andhospitalization among smaller Hb subcategories. In adult studies,the optimal target Hb remains controversial. Higher Hb levelshave been associated with increased access thrombosis, hypertension,and cardiovascular events. In 2004, the Cochrane Renal Groupperformed a meta-analysis of randomized, clinical trials toevaluate the harms and benefits of various Hb targets in patientswith chronic kidney disease; they found lower risk for all-causemortality with Hb <12 versus >13 g/dl (relative risk 0.84;95% CI 0.71 to 1.00) (15). In 1998, Besarab et al. (16) halteda randomized, clinical trial in adults who had cardiac disease,were on HD, and were receiving epoetin to achieve a hematocritof 42 versus 30%. The group with higher hematocrit experienceddecreased event-free survival. A longitudinal study of HD patientsover 6 mo found no increased risk for mortality or hospitalizationassociated with Hb 12 to <13 versus11 and <12 g/dl butdid demonstrated increased risk for mortality and hospitalizationwith Hb <9 versus11 and <12 g/dl (2). More recently,Volkova and Arab (17) performed an evidence-based literaturereview of the relationship between Hb and/or hematocrit andmortality in dialysis patients. They concluded that "most observationalstudies supported the increased mortality associated with Hblevels less than the reference range [of 11 to 12 g/dl] ....Evidence of risks or benefits of Hb levels greater than 11 to12 g/dl is variable."
We hypothesized that because cardiovascular damage in adolescentstends to reflect uremic arteriopathy and metabolic dysregulationrather than traditional atherogenic risk factors, achievinghigher Hb in adolescents with ESRD may not pose the same riskfor mortality and cardiovascular events as in the adult ESRDpopulation (18). In addition, pediatric norms for Hb vary byage with peaks in infancy and adolescence; mean Hb values formale and female adolescents who are aged 12 to 18 yr are 14.5(±0.75) and 14.0 g/dl (±1), respectively (19).We considered that adolescents who had ESRD and were on HD actuallymay require higher Hb levels than the current adult target tomeet increased metabolic needs.
When we recategorized Hb as <10, 10 and <11, 11 to 12,and >12 g/dl, the greatest risk for mortality was at Hb <10g/dl, and risk for mortality was similar between Hb 11 to 12and >12 g/dl (versus Hb <10 g/dl; Table 5). The mean Hbin the subcategory of Hb >12 g/dl was 12.8, which may explainour inability to detect a difference between Hb 11 to 12 and>12 g/dl. Sixty-five patients had Hb >13 g/dl, and amongthose, there was only one death.
The model that excluded those without Medicare/Medicaid coverageat study entry was the only model that differed from the othermodels for mortality risk. This model showed overall lower pointestimates at all Hb levels (i.e., showed a greater degree ofdecreased risk for mortality at all Hb categories above Hb <10g/dl). This model was composed of patients who were more likelyto be on HD longer because there is a waiting period to becomeMedicare eligible. These results suggest that patients who haveHb <10 g/dl and have been on HD >6 mo are at greatestrisk for mortality.
Because of the observational nature of this study, we cannotinfer causality. Patients with more severe anemia and longerHD duration may have more severe hyperparathyroidism or a greaterdegree of long-standing uremia or may be hyporesponsive to epoetin.In our study population, we were unable to detect a statisticallysignificant dose-dependent relationship between epoetin dosingand mortality. We did not see a relationship between being prescribediron and mortality; however, we did not have available dataon iron dosing. It also is possible that the patients in ourdata set who had Hb <10 g/dl and were on HD 180 d representedpatients who were most medically nonadherent and were not presentingfor HD and therefore not receiving their prescribed epoetinor iron. We are limited by the available covariates to explainwhether this observed association between Hb <10 g/dl andincreased mortality is causal or whether higher hemoglobin isa marker for healthier patients with better compliance and/orless inflammation, for example. Clinically, however, we canapply this evidence by recognizing the risk that is associatedwith low Hb and looking more closely at these patients withmore severe anemia. Increasing epoetin dosing alone may notbe effective because other factors that limit epoetin responsemay be at work. Each patient’s comorbid conditions andrisk factors must be assessed using their level of Hb as a potentialmarker for poorer outcome.
For hospitalization analysis, we did not detect a statisticallysignificant association between Hb level and risk for hospitalization.None of the covariates was statistically significant in thefull adjusted model, with the exception of two of the four epoetinquartiles. The relationship between epoetin and risk for hospitalizationdid not seem dose dependent and therefore is difficult to interpret.
In crude analyses, serum albumin <3.5/3.2 g/dl and mean serumferritin 800 ng/ml both were associated with increased riskfor hospitalization. We used serum albumin and ferritin as surrogatesfor inflammation. In our database, we did not have data on specificacute or chronic inflammatory markers. A chronic inflammatorystate was demonstrated previously in pediatric patients whowere on HD (20). An inflammatory state may modify the relationshipbetween anemia and hospitalization; however, we were unableto detect this association.
Patients with Hb 11 g/dl consistently demonstrated a 60 to 70%decreased risk for mortality across multiple models and withpropensity score matching. Within Hb subcategories, Hb 11 to12 versus <10 g/dl also was strongly associated with decreasedrisk for mortality (HR 0.30; 95% CI 0.19 to 0.74). Risk formortality was similar between Hb 11 to 12 g/dl and >12 versus<10 g/dl. Therefore, in our population of adolescents whohad ESRD and were on HD, those who failed to achieve the KDOQItarget of Hb 11 to 12 g/dl showed increased risk for mortality.Because of the observational nature of our study and limitednumbers of patients in each subcategory, we cannot concludethat Hb above the current adult KDOQI target is beneficial foradolescents who have ESRD and are on HD. We also are unableto infer any causal relationship between anemia and mortalityor hospitalization. Further study in the form of randomized,clinical trials is necessary to investigate causality and toelucidate optimal Hb levels in this particular subset of patientswith ESRD.
Acknowledgments
At the time of research, S.A. was supported by a National Instituteof Diabetes and Digestive and Kidney Diseases Renal DiseaseEpidemiology training grant (5 T32 DK07732), Johns Hopkins University,Bloomberg School of Public Health. S.F. receives support forthis research through an R21 grant (R21 DK 064313-01).
This work was presented in part in abstract form at the AmericanSociety of Pediatric Nephrology conferences (May 1 to 4, 2004,and April 29 to May 2, 2006; San Francisco, CA) and the annualmeeting of the American Society of Nephrology (October 29 throughNovember 1, 2004; St. Louis, MO).
Footnotes
Published online ahead of print. Publication date availableat www.jasn.org.
The views expressed in this article are those of the authorsand do not necessarily represent the official policy of theCenters for Medicare & Medicaid Services or the United Statesgovernment.
Foley RN, Parfrey PS, Harnett JD, Kent GM, Murray DC, Barre PE: The impact of anemia on cardiomyopathy, morbidity and mortality in end-stage renal disease.
Am J Kidney Dis 28
: 53
–61, 1996[Medline]
Ofsthun N, Labrecque J, Lacson E, Keen M, Lazarus JM: The effects of higher hemoglobin levels on mortality and hospitalization in hemodialysis patients.
Kidney Int 63
: 1908
–1914, 2003[CrossRef][Medline]
Xia H, Ebben J, Ma JZ, Collins AJ: Hematocrit levels and hospitalization risks in hemodialysis patients.
J Am Soc Nephrol 10
: 1309
–1316, 1999[Abstract/Free Full Text]
Collins AJ, Ma JZ, Ebben J: Impact of hematocrit on morbidity and mortality.
Semin Nephrol 20
: 345
–349, 2000[Medline]
Eknoyan G: The importance of early treatment of the anaemia of chronic kidney disease.
Nephrol Dial Transplant 16[Suppl 5]
: 45
–49, 2001
Sugarman JR, Frederick PR, Frankenfield DL, Owen WF Jr, McClellan WM: Developing clinical performance measures based on the Dialysis Outcomes Quality Initiative Clinical Practice Guidelines: Process, outcomes, and implications.
Am J Kidney Dis 42
: 806
–812, 2003[CrossRef][Medline]
Owen WF Jr, Lew NL, Liu Y, Lowrie EG, Lazarus JM: The urea reduction ratio and serum albumin concentration as predictors of mortality in patients undergoing hemodialysis.
N Engl J Med 329
: 1001
–1006, 1993[Abstract/Free Full Text]
Goldwasser P, Mittman N, Antignani A, Burrell D, Michel MA, Collier J, Avram MM: Predictors of mortality in hemodialysis patients.
J Am Soc Nephrol 3
: 1613
–1622, 1993[Abstract]
Soucie JM, McClellan WM: Early death in dialysis patients: Risk factors and impact on incidence and mortality rates.
J Am Soc Nephrol 7
: 2169
–2175, 1996[Abstract]
Pisoni RL, Bragg-Grasham JL, Young EW, Akizawa T, Asano Y, Locatelli F, Bommer J, Cruz MJ, Kerr PG, Mendelssohn DC, Held PJ, Port FK: Anemia management and outcomes from 12 countries in the Dialysis Outcomes and Practice Patterns Study (DOPPS).
Am J Kidney Dis 44
: 94
–111, 2004[CrossRef][Medline]
National Kidney Foundation-Dialysis Outcomes Quality Initiative: NKF-DOQI clinical practice guidelines for the treatment of anemia of chronic renal failure.
Am J Kidney Dis 30
: S192
–S240, 1997[Medline]
Centers for Medicare and Medicaid Services: 2001 Annual Report: End stage renal disease clinical performance measures project.
Am J Kidney Dis 39[Suppl 2]
: S1
–S98, 2002
Health Care Financing Administration: 2000 Annual Report: ESRD Clinical Performance Measures Project.
Am J Kidney Dis 37[Suppl 3]
: S1
–S74, 2001
Warady BA, Ho M: Morbidity and mortality in children with anemia at initiation of dialysis.
Pediatr Nephrol 18
: 1055
–1062, 2003[CrossRef][Medline]
Strippoli GF, Craig JC, Manno C, Schena FP: Hemoglobin targets for the anemia of chronic kidney disease: A meta-analysis of randomized, controlled trials.
J Am Soc Nephrol 15
: 3154
–3165, 2004[Abstract/Free Full Text]
Besarab A, Bolton WK, Browne JK, Egrie JC, Nissenson AR, Okamoto DM, Schwab SJ, Goodkin DA: The effects of normal as compared with low hematocrit values in patients with cardiac disease who are receiving hemodialysis and epoetin.
N Engl J Med 339
: 584
–590, 1998[Abstract/Free Full Text]
Volkova N, Arab L: Evidence-based systematic literature review of hemoglobin/hematocrit and all-cause mortality in dialysis patients.
Am J Kidney Dis 47
: 24
–36, 2006[CrossRef][Medline]
Oh J, Wunsch R, Turzer M, Bahner M, Raggi P, Querfeld U, Mehls O, Schaefer F: Advanced coronary and carotid arteriopathy in young adults with childhood-onset chronic renal failure.
Circulation 106
: 100
–105, 2002[Abstract/Free Full Text]
Dallman, PR: Blood-forming tissues. In:
Pediatrics, 16th Ed., edited by Rudolph AM, New York, Appleton-Century-Crofts, 1977
, p 1111
S. Amaral, W. Hwang, B. Fivush, A. Neu, D. Frankenfield, and S. Furth Serum Albumin Level and Risk for Mortality and Hospitalization in Adolescents on Hemodialysis
Clin. J. Am. Soc. Nephrol.,
May 1, 2008;
3(3):
759 - 767.
[Abstract][Full Text][PDF]
Top
Abstract
Introduction
11 g/dl (P = 0.001); 40.3% with baseline Hb <11 g/dl were hospitalized versus 31.1% with initial Hb 
12, and >12 g/dl, risk of mortality declined as Hb level increased. At Hb 11 to 12 g/dl (versus Hb <10 g/dl), mortality risk decreased by 69% (HR 0.31; 95% CI 0.14 to 0.65). Risk for mortality was similar for Hb 11 to 12 and >12 g/dl. For hospitalization, no statistically significant difference in risk between Hb categories was found. This observational study of adolescents who are on HD is consistent with adult literature showing decreased mortality in patients who have ESRD and meet adult Hb targets. Further studies in the form of randomized, clinical trials are needed to assess optimal Hb levels for adolescents who are on HD. 
2, Fisher exact test, ANOVA, and t test were used to compare dichotomized clinical characteristics and death or hospitalization (as yes/no) between baseline Hb <11 versus 
