Improving Albumin Levels Among Hemodialysis Patients: A Community-Based Randomized Controlled Trial

●

fvsifwedtociiotil©

I

Hs

CtisC

M

o

RMFN

sMa

2

Improving Albumin Levels Among Hemodialysis Patients: ACommunity-Based Randomized Controlled Trial

Janeen B. Leon, MS, RD, LD, Jeffrey M. Albert, PhD, Gina Gilchrist, RD, LD, Irving Kushner, MD,Edith Lerner, PhD, Suzanne Mach, MS, RD, LD, Angela Majerle, RD, LD, David Porter, BA,

Edmond Ricanati, MD, Laurine Sperry, RD, LD, Catherine Sullivan, BS,Jennifer Zimmerer, MS, RD, LD, and Ashwini R. Sehgal, MD

Background: Low albumin level is a strong predictor of mortality and morbidity among hemodialysis patients, yetew interventions are available to improve albumin levels. Moreover, the relative importance of nutritional barriersersus inflammation in contributing to hypoalbuminemia is unclear. We sought to determine whether targetingpecific nutritional barriers will improve albumin levels. Methods: We conducted a randomized controlled trialnvolving 180 patients with baseline albumin levels less than 3.7 g/dL (<37 g/L) at 44 long-term hemodialysisacilities. Study coordinators identified and intervened on specific barriers present among intervention patients,hereas control patients continued to receive the usual care. Barriers targeted included poor nutritional knowl-dge, poor appetite, help needed with shopping or cooking, low fluid intake, inadequate dialysis dose, depression,ifficulty chewing, difficulty swallowing, gastrointestinal symptoms, and acidosis. Results: At baseline, interven-

ion and control patients had similar albumin levels, dietary intakes, levels of inflammatory markers, and numbersf nutritional barriers. After 12 months, intervention patients had greater increases in albumin levels compared withontrol patients (�0.21 versus �0.06 g/dL [�2.1 versus �0.6 g/L]; P < 0.01), as well as greater increases in energyntake (�4.1 versus �0.6 Kcal/d/kg; P < 0.001) and protein intake (�0.13 versus �0.06 g/d/kg; P < 0.001). Thentervention appeared most effective for barriers related to poor nutritional knowledge, help needed with shoppingr cooking, and difficulty swallowing. About half the subjects had elevated levels of inflammatory markers, buthere was no relationship between change in levels of albumin and inflammatory markers. Conclusion: A nutritionntervention tailored to patient-specific barriers resulted in modest improvements in albumin levels regardless ofevels of inflammatory markers. Am J Kidney Dis 48:28-36.

2006 by the National Kidney Foundation, Inc.

NDEX WORDS: Hypoalbuminemia; hemodialysis (HD); dietary intake; inflammation.

lrhmtthpidiri

lpinkpdtbOd

Editorial, p. 171

YPOALBUMINEMIA frequently is presentin hemodialysis patients and correlates

trongly with mortality and morbidity.1,2 Estab-

From the Division of Nephrology, MetroHealth Medicalenter; Departments of Medicine, Epidemiology and Biosta-

istics, Nutrition, and Biomedical Ethics; Center for Reduc-ng Health Disparities; and Center for Health Care Re-earch and Policy, Case Western Reserve University,leveland, OH.Received December 30, 2005; accepted in revised formarch 22, 2006.Originally published online as doi:10.1053/j.ajkd.2006.03.046

n May 15, 2006.Support: Supported by grants DK51472 and GCRC M01

R00080 from the National Institutes of Health, Bethesda,D, and by the Leonard C. Rosenberg Renal Researchoundation, Cleveland, OH. Potential conflicts of interest:one.Address reprint requests to Ashwini R. Sehgal, MD, Divi-

ion of Nephrology, MetroHealth Medical Center, 2500etroHealth Dr, Cleveland, OH 44109. E-mail:

[email protected]© 2006 by the National Kidney Foundation, Inc.0272-6386/06/4801-0004$32.00/0

idoi:10.1053/j.ajkd.2006.03.046

American Journ8

ishing the causality of this relationship wouldequire showing that interventions that correctypoalbuminemia also improve mortality andorbidity.3 However, few good strategies exist

o improve albumin levels in hemodialysis pa-ients. Moreover, many hemodialysis patientsave high levels of C-reactive protein, an acute-hase reactant that is elevated in patients withnflammatory states.4-6 Because albumin levelsecrease in patients with inflammatory states, its unclear to what extent hypoalbuminemia is aesult of malnutrition or simply a reflection ofnflammatory states.

We reasoned that efforts to improve albuminevels should be based on an understanding ofatient-specific barriers. In previous work, wedentified a number of potentially modifiableutritional barriers, including poor nutritionalnowledge, poor appetite, help needed with shop-ing or cooking, low fluid intake, and inadequateialysis dose.7 We also pilot tested an interven-ion tailored to the presence or absence of thesearriers in individual hemodialysis patients.8

ther investigators have identified depression,ifficulty chewing, difficulty swallowing, gastro-

al of Kidney Diseases, Vol 48, No 1 (July), 2006: pp 28-36

trrstolp

S

actrTcspg3mlmntantppwhfiff$tIC

cnrffd

NI

ttocd

tfcvtamaw1ndsteaeadagsKsbl(

I

vpbw

pbcr

ws5Pfb

act

sPo

laad

IMPROVING ALBUMIN LEVELS 29

ially modifiable nutritional barriers.9-11 We noweport results of a full-scale community-basedandomized controlled trial targeting patient-pecific nutritional barriers. To better understandhe role of inflammatory states in modifying theutcome of this intervention, we also measuredevels of 2 inflammatory markers (C-reactiverotein and serum amyloid A).6

METHODS

ubjects and FacilitiesAll 47 long-term hemodialysis facilities in northeast Ohio

greed to participate. We pilot tested our methods at 1onveniently located facility, excluded 2 very small facili-ies, and used a random-number generator to assign theemaining 44 facilities to an intervention or control group.o determine eligibility, study coordinators abstracted medi-al records to identify patients for whom the most recenterum albumin level and mean serum albumin level for therevious 3 months were both less than 3.70 g/dL (�37.0/L) by means of the bromcresol green method or less than.40 g/dL (�34.0 g/L) by means of the bromcresol purpleethod.12 This ensured that only patients with persistently

ow albumin levels were included. We selected these albu-in cutoff values because about one third of dialysis patients

ationally have values less than these thresholds.1 Addi-ional patient eligibility criteria were age of 18 to 85 yearsnd receiving dialysis for at least 9 months. We excludedew patients because the first several months of dialysisreatment often is a time of changes in diet and nutritionalarameters.13 We also excluded subjects who could notarticipate (did not speak English, mentally impaired) orere likely to have unique nutritional issues (ie, nursingome residents, patients with cirrhosis, acquired immunode-ciency syndrome, active malignancy, terminal illness, tubeeedings, and total parenteral nutrition). We obtained in-ormed consent from eligible patients, and each was given25 at the beginning and again at the end of the trial to thankhem for their participation. This study was approved by thenstitutional Review Board of MetroHealth Medical Center,leveland, OH.Before entering the study, all subjects received nutritional

are from their facility’s registered dietitian. This includedutritional status assessment, monthly laboratory test resulteview, and education regarding the renal diet. On average,acility dietitians were responsible for 124 patients perull-time equivalent and had worked for 10 years as renalietitians.

utritional Parameters, Quality of Life, andnflammatory Markers

Study coordinators abstracted medical records of interven-ion and control subjects to obtain demographic characteris-ics (age, sex, race, ethnicity), medical characteristics (causef renal failure, time receiving dialysis, number of comorbidonditions), and nutritional parameters (albumin level, post-

ions was calculated based on the presence or absence of theollowing 10 disease categories: coronary artery disease,ongestive heart failure, peripheral vascular disease, cerebro-ascular disease, depression or psychosis, previous solidumor or hematologic malignancy, connective tissue disease,sthma or chronic obstructive pulmonary disease, diabetesellitus, and drug or alcohol abuse. Albumin levels gener-

lly were available on a monthly basis, whereas postdialysiseight was available after every dialysis treatment (typically3 treatments/mo). Dietary intake was assessed at the begin-ing and again at the end of the trial by using two 24-hourietary recalls. This involved 1 dialysis day and 1 nondialy-is day, generally within a 2-week period.14 Study coordina-ors performed a brief nutrition-focused history and physicalxamination (referred to as a subjective global assessment)t the beginning and end of the trial. Patients with inad-quate protein stores will be noted to have low muscle massnd/or edema on this focused examination.15,16 Study coor-inators also assessed patient quality of life at the beginningnd end of the trial by using several subscales (related toeneral health, physical functioning, emotional well-being,ocial function, pain, and dialysis-related symptoms) of theidney Disease Quality of Life questionnaire.17 A predialy-

is blood sample was obtained from each patient at theeginning and end of the study and sent to a single centralaboratory for measurement of 2 inflammatory markersC-reactive protein and serum amyloid A).

dentification of BarriersStudy coordinators abstracted medical records and inter-

iewed intervention and control subjects to determine theresence of 10 specific nutritional barriers. Categorization ofarriers was based on our pilot study, as well as previousork by other investigators.7,8,18,19

Poor nutritional knowledge. Patients identified high-rotein foods from a list of 10 high-protein foods (eg, eggs,eef, fish) and 10 minimal-protein or nonprotein foods (eg,hips, carrots, bread). Patients with 5 or more incorrectesponses were categorized as having this barrier.

Poor appetite. Patients rated their overall appetite, asell as their appetite for 12 common high-protein foods,

uch as eggs, beef, cheese, and peanut butter, by using a-point Likert scale ranging from very good to very poor.atients with fair/poor overall appetite or fair/poor appetiteor 4 or more specific foods were categorized as having thisarrier.Help needed with shopping or cooking. Patients were

sked whether they needed additional help with shopping orooking. Patients who said yes were categorized as havinghis barrier.

Low fluid intake. Study coordinators examined flowheets for the first 6 treatments of the enrollment month.atients with a mean interdialytic weight gain less than 2.5%f their dry weight were categorized as having this barrier.Inadequate dialysis dose. Patients with a mean Kt/V

ess than 1.2 during the previous 3 months were categorizeds having this barrier. Because missed treatments do notffect Kt/V, patients with more than 2 missed treatmentsuring the previous 3 months also were categorized as

Ish

crsi

dLar

hc

whswth

I

ttttctoat

cisp

iptc

tefmbcl

wpap

m

pcn

Daa

ta

tpc

mc

wgba

mmacfts

C

ntqat

F

SddDcd

O

oembimvf

LEON ET AL30

Depression. Patients completed the Beck Depressionnventory, a standard screening questionnaire for depres-ion. Patients with scores of 15 or higher were categorized asaving this barrier.19

Difficulty chewing. Patients were asked if they had diffi-ulty chewing and answered by using a 4-point Likert scaleanging from never to always. Patients who always orometimes had difficulty chewing were categorized as hav-ng this barrier.

Difficulty swallowing. Patients were asked if they hadifficulty swallowing and answered by using a 4-pointikert scale ranging from never to always. Patients wholways or sometimes had difficulty swallowing were catego-ized as having this barrier.

Gastrointestinal symptoms. Patients were asked if theyad heartburn or nausea. Patients who answered yes wereategorized as having this barrier.

Acidosis. Patients whose predialysis bicarbonate levelas less than 22 mEq/L (�22 mmol/L) were categorized asaving this barrier.18 Blood samples were collected in serumeparator tubes, allowed to clot for 30 minutes, and thenere centrifuged and refrigerated. Study staff transported

he separated samples on ice to a central laboratory within 24ours for analysis.

ntervention GroupStudy coordinators educated all intervention patients about

he meaning and importance of good nutritional status. Theyhen provided feedback and recommendations to interven-ion patients. The information was provided during a dialysisreatment and tailored to the specific barriers present. Studyoordinators also communicated information about barrierso facility dietitians and modified recommendations basedn feedback from these dietitians. Facility dietitians weresked to reinforce study coordinator recommendations whenhey met with their study patients.

Poor nutritional knowledge. Study coordinators edu-ated patients about high-protein foods by using a variety ofnteractive activities (eg, puzzles, nutrition label reading),elf-teaching activities (eg, word searches, crossworduzzles), and educational handouts.Poor appetite. Study coordinators recommended increas-

ng the intake of specific foods for which patients hadreserved appetite. Study coordinators also provided pa-ients with limited amounts of supplements, such as commer-ially available enteral nutrition drinks and cookies.

Help needed with shopping or cooking. Study coordina-ors, in collaboration with facility dietitians and social work-rs, explored the possibility of obtaining help from family,riends, and social support agencies. Subjects most oftenentioned a need for a shopper’s aide to assist with carrying

ags or providing transportation to stores and a need forooking assistance because of fatigue and functional statusimitations.

Low fluid intake. Study coordinators, in collaborationith facility dietitians, recommended that patients add arotein-containing beverage to their diet (eg, a commerciallyvailable enteral nutrition supplement). Study coordinatorsrovided patients with limited amounts of supplements.Inadequate dialysis dose. Study coordinators deter-

rescription, noncompliance with prescribed treatment time,atheter use) and shared this information with patients’ephrologists.20

Depression. Study coordinators shared elevated Beckepression Inventory scores with patients’ social workers

nd nephrologists and recommended that patients be evalu-ted and, if appropriate, treated for depression.

Difficulty chewing. Study coordinators recommendedhat patients see their nephrologist, primary care physician,nd/or dentist for evaluation.

Difficulty swallowing. Study coordinators recommendedhat patients see their nephrologist and/or primary carehysician for evaluation (eg, barium swallow). In someases, difficulty swallowing was linked to difficulty chewing.

Gastrointestinal symptoms. Study coordinators recom-ended that patients see their nephrologist and/or primary

are physician for evaluation.Acidosis. Study coordinators shared bicarbonate levels

ith patients’ nephrologists and recommended that nephrolo-ists address reversible causes of acidosis, increase dialysateicarbonate concentrations, and/or prescribe oral bicarbon-te supplements.

During the next 12 months, study coordinators metonthly with patients to reinforce these recommendations,onitor progress, and answer questions. Study coordinators

lso updated patients’ dietitians monthly. Because the studyoordinators carried out the intervention, it was not possibleor them to be blinded to patients’ assignments to interven-ion versus control groups. There were no adverse events oride effects associated with the intervention.

ontrol GroupControl patients continued to receive usual care from their

ephrologists, dietitians, and social workers. Study coordina-ors met monthly with control patients and administereduestionnaires related to dietary intake, nutritional barriers,nd/or quality of life. However, neither control patients norheir providers received feedback from study coordinators.

ollow-Up ProceduresAll patients were recruited between February 2002 and

eptember 2003 and followed up for 12 months or until theyied, moved, received a transplant, or were hospitalized (andid not return to outpatient dialysis by the end of the study).uring this interval, medical records of intervention and

ontrol patients were abstracted on a monthly basis to obtainata for nutritional parameters and barriers.

utcomesA primary outcome is change in serum albumin level. To

btain more precise estimates, the final albumin level ofach patient is defined as the mean of albumin measure-ents in the final 3 months of study participation, whereas

aseline albumin level is the mean of albumin measurementsn the 3 months before subject enrollment. Change in albu-in level was calculated as final minus baseline albumin

alue. Because we expected it would take about 3 monthsor our intervention to have an effect, patients who died,

fi(

wbeasgd

ngtscbbwt

S

offt

atcootcoWrmfpoda�po

S

w(ptbhijtdrtds

bcmabswdwct

C

3

IMPROVING ALBUMIN LEVELS 31

rst 3 months were not included in the main analysisFig 1).8

Because improved albumin levels may decrease mortality,e defined another primary composite outcome that com-ined change in albumin level and survival. Specifically, wexamined the proportion of patients who had an increase inlbumin level of 0.20 g/dL or greater (�2.0 g/L) andurvived to the end of the study. We selected 0.20 g/dL (2.0/L) because it corresponds to a 25% decrease in the odds ofeath in observational studies.2

Secondary outcomes were changes in values for otherutritional parameters (weight, dietary intake, subjectivelobal assessment), whether patients overcame specific nutri-ional barriers, and changes in quality of life. Patients with apecific barrier at baseline were considered to have over-ome the barrier if they no longer met the definition of thearrier at the end of the trial. For example, subjects withaseline Beck Depression Inventory scores of 15 or higherere categorized as overcoming the depression barrier if

heir final scores were less than 15.

tatistical AnalysisBecause facilities comprised the unit of randomization,

ur main analyses account for clustering of patients byacility. Specifically, we compared change in albumin levelor intervention versus control patients by using an adjusted

Fig 1. Flow of participants through the trial.

-test that reflects this clustering. Similarly, we used an p

djusted chi-square test that reflects clustering to comparehe proportion of patients in each group who achieved theomposite outcome of increase in albumin level of 0.20 g/dLr greater (�2.0 g/L) and survival. Baseline characteristicsf intervention versus control patients and changes in nutri-ional parameters, specific barriers, and quality of life wereompared by using chi-square test for categorical variablesr Mann-Whitney rank-sum test for continuous variables.e used Spearman correlation coefficient to determine the

elationship between change in levels of albumin and inflam-atory markers. JMP and SAS statistical software were used

or all analyses (both from SAS Institute Inc, Cary, NC). Weerformed a post hoc power calculation and estimated thatur trial of 180 patients had greater than 90% power toetect a difference in albumin levels between interventionnd control patients of 0.065 g/dL (0.65 g/L) with a 2-tailedof 0.05. We also estimated that we had greater than 80%

ower to detect a difference of 15% in the compositeutcome between intervention and control patients.21

RESULTS

ubject and Facility Characteristics

Of 44 long-term hemodialysis facilities, 41 (93%)ere free standing (versus hospital based) and 35

80%) were for profit. Figure 1 shows the flow ofarticipants through the trial. Of all ineligible pa-ients, about 80% in both groups were excludedecause their albumin levels were too high. Oneundred eighty patients completed the trial, includ-ng 86 intervention subjects and 94 control sub-ects. One hundred forty eligible subjects declinedo participate, and another 27 subjects died, with-rew, moved, or became mentally impaired beforeeaching the evaluation phase. These 167 nonpar-icipants did not differ from the 180 participants inemographic characteristics, time receiving dialy-is, or baseline albumin level.

Intervention and control patients had similaraseline demographic characteristics, medicalharacteristics, nutritional parameters, and inflam-atory marker levels (Table 1). The 2 groups

lso had a similar total number of nutritionalarriers, although they differed somewhat onpecific nutritional barriers. Intervention patientsere more likely to have low fluid intake andifficulty swallowing, whereas control patientsere more likely to have poor appetite. The most

ommon barriers in both groups were poor nutri-ional knowledge and poor appetite.

hanges in Nutritional Parameters

As listed in Table 2, intervention patients had.5-fold greater increases in albumin levels com-

[0iacc

actbd

MFR

HC

MMMUT

MAMMMMS

CCSSTS

LEON ET AL32

SE] versus 0.06 � 0.03 g/dL [2.1 � 0.4 versus.6 � 0.3 g/L]; P � 0.01). Forty-two percent ofntervention patients and 29% of control patientschieved the composite albumin/survival out-ome, but this difference was of marginal statisti-

Table 1. Baseline Characteristic

ean age (y)emale (%)ace (%)BlackWhiteOtherispanic (%)ause of renal failure (%)DiabetesHypertensionGlomerulonephritisOtherean no. of comorbid conditionsean years receiving dialysisean duration of treatment (min)sing high-flux dialyzer (%)ype of vascular access (%)FistulaGraftCatheterean Kt/Vlbumin (g/dL)ean postdialysis weight (kg)ean body mass index (kg/m2)ean energy intake (Kcal/d/kg)ean protein intake (g/d/kg)ubjective global assessment (%)Well nourishedMildly to moderately malnourishedSeverely malnourished-Reactive protein (mg/L)-Reactive protein � 10 mg/L (%)erum amyloid A (mg/L)erum amyloid A � 10 mg/L (%)otal no. of barrierspecific barriers (%)Poor nutritional knowledgePoor appetiteHelp needed with shopping or cookingLow fluid intakeInadequate dialysis doseDepressionDifficulty chewingDifficulty swallowingGastrointestinal symptomsAcidosis

NOTE. To convert albumin in g/dL to g/L, multiply by 10.

al significance (P � 0.06). Intervention patients (

lso had increased energy and protein intakeompared with control patients. On average, in-ervention patients increased their energy intakey 333 � 70 Kcal/d, whereas control patientsecreased their energy intake by 47 � 66 Kcal/d

tervention and Control Subjects

ntion86)

Control(n � 94) P

60 0.3553 0.50

5938 0.2031 0.95

4924 0.731413

.9 2.1 0.38

.8 3.1 0.47228 0.9376 0.39

2649 0.2426

.46 1.40 0.17

.40 3.40 0.90

.3 78.0 0.30

.0 27.9 0.40

.3 19.9 0.73

.83 0.83 0.90

4055 0.604

19 0.8146 0.1532 0.3554 0.45

.2 3.1 0.74

44 0.6974 0.0423 0.1212 0.0219 0.0631 0.9433 0.6713 �0.0133 0.2027 0.74

s of In

Interve(n �

6258

455131

4524201012

22770

35372813

8129200

48493

185746603

416034269

3136312424

P � 0.001). Most of this increase (�70%)

ot1c�s

C

kw

b2tecd0cb2

ACPBEPS

P

P

H

L

I

D

D

D

G

A

IMPROVING ALBUMIN LEVELS 33

ccurred on nondialysis days. Similarly, interven-ion patients increased their protein intake by0.7 � 3.3 g/d, whereas control patients de-reased their protein intake by 4.7 � 3.2 g/d (P

0.001). There were no differences in postdialy-is weight or subjective global assessment.

hanges in Nutritional Barriers

Of the 76 patients with poor nutritionalnowledge at baseline, intervention patients

Table 2. Changes in Nutritiona

lbumin (g/dL)hange in albumin � 0.20 g/dL and survived (%)ostdialysis weight (kg)ody mass index (kg/m2)nergy intake (Kcal/d/kg)rotein intake (g/d/kg)ubjective global assessment (%)ImprovedNo changeWorsened

oor nutritional knowledgeSubjects with barrierOvercame barrier (%)

oor appetiteSubjects with barrierOvercame barrier (%)elp needed with shopping or cookingSubjects with barrierOvercame barrier (%)

ow fluid intakeSubjects with barrierOvercame barrier (%)

nadequate dialysis doseSubjects with barrierOvercame barrier (%)epressionSubjects with barrierOvercame barrier (%)ifficulty chewingSubjects with barrierOvercame barrier (%)ifficulty swallowingSubjects with barrierOvercame barrier (%)astrointestinal symptomsSubjects with barrierOvercame barrier (%)

cidosisSubjects with barrierOvercame barrier (%)

NOTE. To convert albumin in g/dL to g/L, multiply by 10.

ere 4 times more likely to overcome this 6

arrier by the end of the study (89% versus2%; P � 0.001; Table 2). Intervention pa-ients also were more likely to overcome barri-rs related to help needed with shopping orooking (86% versus 55%; P � 0.01) andifficulty swallowing (74% versus 42%; P �.05). A greater percentage of interventionompared with control patients also overcamearriers related to poor appetite (33% versus0%), inadequate dialysis dose (88% versus

eters and Nutritional Barriers

Intervention Control P

�0.21 �0.06 �0.0142 29 0.06�0.06 �0.50 0.52�0.06 �0.18 0.62�4.1 �0.6 �0.001�0.13 �0.06 �0.001

16 1677 76 0.937 9

35 4189 22 �0.001

52 7033 20 0.11

29 2286 55 0.01

22 1155 64 0.62

8 1888 67 0.25

27 2941 34 0.63

31 3158 55 0.80

27 1274 42 0.05

21 3162 45 0.23

21 2576 72 0.75

l Param

7%), and gastrointestinal symptoms (62% ver-

st

tmawcw

Q

tstp

I

lmcTalTil

tip

lAcemTrshvi

dgiaalctjAdagpo

leatamtdmd

psnrepmasur

cv

LEON ET AL34

us 45%), but these differences were not statis-ically significant (all P � 0.11).

The number of barriers remaining at the end ofhe trial correlated strongly with change in albu-in level for all patients. Mean increase in

lbumin level was 0.20 g/dL (2.0 g/L) in patientsith no barriers remaining, whereas mean in-

rease was only 0.02 g/dL (0.2 g/L) in patientsith 4 or more barriers remaining (Fig 2).

uality of Life

There were no differences between interven-ion and control patients in quality-of-life sub-cales, including general health, physical func-ioning, emotional well-being, social function,ain, and dialysis-related symptoms.

nflammatory Markers

There were no significant changes in meanevels of C-reactive protein (mean change, �0.3g/L; P � 0.21) or serum amyloid A (mean

hange, �5 mg/L; P � 0.15) during the trial.here was no relationship between change inlbumin level and baseline C-reactive proteinevel (correlation coefficient, 0.03; P � 0.69).here also was no relationship between change

n albumin level and baseline serum amyloid Aevel (correlation coefficient, �0.003; P � 0.97).

DISCUSSION

We found that a nutrition intervention tailoredo patient-specific barriers resulted in modestmprovements in albumin levels in hemodialysis

Fig 2. Final number of barriers remaining versushange in albumin levels among all patients. To con-ert albumin in g/dL to g/L, multiply by 10.

atients. Although many patients had elevated a

evels of C-reactive protein and serum amyloid, these inflammatory markers did not predict

hanges in albumin levels. This suggests thatven patients with high levels of inflammatoryarkers can respond to nutritional interventions.he graded relationship between number of bar-

iers and change in albumin level (Fig 2) furtherupports the role of these nutritional barriers inemodialysis patients. Our approach has the ad-antages of being simple, low cost, and easy tomplement.

By engaging the participation of virtually allialysis facilities and renal dietitians in a largeeographic area, we enhanced the generalizabil-ty of our findings. With the exception of racend ethnicity, patient and facility characteristicsre approximately similar to national data.22 Thearge number of black subjects reflects the inner-ity location of many of the participating facili-ies, whereas the small number of Hispanic sub-ects reflects the population of northeast Ohio.nother strength of this study is its randomizedesign, which resulted in similar interventionnd control groups in terms of patient demo-raphic and medical characteristics, nutritionalarameters, inflammatory markers, and numberf nutritional barriers (Table 1).

The magnitude of improvements in albuminevels is consistent with what we found in anarlier pilot study.8 Three recent clinical trialslso found improvements of approximately 0.2o 0.3 g/dL (2 to 3 g/L) in serum albumin levelsfter administration of oral nutritional supple-ents or appetite stimulants.23-25 Although these

rials had small sample sizes and were not ran-omized, they further support the conclusion thatodest improvements in albumin levels in hemo-

ialysis patients are possible.Our results have important implications for

atients, providers, and policy makers. Patientshould be actively involved in improving theirutritional knowledge, reporting nutritional bar-iers to providers, and trying to overcome barri-rs and increase dietary intake. The level ofatient involvement required to improve albu-in levels is much greater than that needed to

ddress other dialysis-related quality parameters,uch as hemoglobin levels, which are largelynder the control of providers. Providers shouldoutinely monitor not only albumin levels, but

iposssefiaderacOsdc

umfiepambsaraapsladn

pttfedtlnap

ipmisatp

pmacwbietpm�ipitphsisei

mAamitset

ACRBMD

IMPROVING ALBUMIN LEVELS 35

n this clinical trial. In particular, we found thatroviders were unaware of the high prevalencef depression, difficulty chewing, and difficultywallowing among their patients. Policy makershould ensure that sufficient personnel and re-ources are available to address nutritional barri-rs. Addressing the nutritional barriers we identi-ed requires a moderate amount of time, as wells collaboration among various providers (eg,ietitians, social workers, nephrologists). How-ver, dietitians and social workers typically areesponsible for more than 100 patients apiece,nd patient to provider ratios are likely to in-rease further if reimbursement is reduced.26

ther key resources needed to address suchpecific barriers as nutritional supplements andental care currently are not covered by Medi-are.

Our results also are relevant to the continuedtility of albumin level as a clinical performanceeasure by the Medicare program.1 Clinical per-

ormance measures are indicators that deal withmportant conditions for which quality of care isither variable or substandard and can be im-roved by providers.27 Because our trial, as wells 3 recent trials, showed improvements in albu-in levels, it may be argued that providers should

e able to improve albumin levels in usual careettings, as well.23-25 However, improvements inlbumin levels in these trials were modest andequired additional personnel and resources notvailable under current Medicare coverage. Iflbumin level continues to be used as a clinicalerformance measure, it should be with an under-tanding that improvements in albumin levels areikely to be modest, that multiple providers have

role in improving albumin levels (not justietitians), and that additional resources may beecessary (such as supplements and dental care).Compared with control patients, a greater pro-

ortion of intervention patients overcame 9 ofhe 10 nutritional barriers (Table 2). However,his difference was statistically significant onlyor barriers related to poor nutritional knowl-dge, help needed with shopping or cooking, andifficulty swallowing. The inability to find statis-ically significant differences for other barriersikely is caused by a combination of the smallumbers of patients who had specific barriersnd a limited impact of our intervention com-

ntervention and usual care had little effect onoor appetite and depression. Further refine-ents of our approach may be needed to increase

ts potency for specific barriers. Moreover, theomewhat larger number of subjects with poorppetite in the control group may have disadvan-aged the control group because this barrier wasarticularly difficult to overcome.Several limitations must be considered in inter-

reting our results. First, improvements in albu-in levels were modest. Second, we focused onsingle geographic area. Third, it is possible thatontrol patients improved in part because theyere being observed (the Hawthorne effect) orecause control dietitians were influenced byntervention dietitians (contamination). How-ver, both these effects would tend to decreasehe difference between control and interventionatients; therefore, the measured effect size (albu-in change, 0.21 � 0.06 � 0.15 g/dL [2.1 � 0.61.5 g/L]) may underestimate the value of our

ntervention. Fourth, we did not have sufficientower to rigorously evaluate the impact of ourntervention on specific subgroups. Fifth, al-hough we did not specifically try to excludeatients with chronic inflammatory diseases origh C-reactive protein levels, it is possible thatome of our exclusion criteria (eg, being a nurs-ng home resident) may have had this effect inome cases. Thus, our intervention may not gen-ralize to all patients with elevated levels ofnflammatory markers.

In conclusion, we recommend that providersonitor and address specific nutritional barriers.lthough our sample size and follow-up period

re insufficient to show an impact on patientortality and morbidity, larger observational stud-

es showed a link between albumin levels andhese outcomes.2,28,29 Thus, overcoming patient-pecific nutritional barriers has the potential tonhance survival and decrease both hospitaliza-ions and inpatient expenditures.

ACKNOWLEDGMENT

The authors thank Arianna M. Aoun, MS, RD, CSR;nika Avery-Grant, MS, RD, LD; Earlyn Bentfeld, RD, LD;armen Blakely-Adams, MEd, RD, LD; Donna M. Bodnar,D, CSR, LD; Renée W. Boehnlein, RD, LD; Angela Oritirainard, DTR; Christina Buccino, MHHS, RD, LD; Cindy. Carrell, RD, LD; Lisa Cary, RD, LD; Julie A. Charif, BS,TR; Iris Chears, DTR; Sherilyn Churchia, RD, LD; Rose

pPLCDPLRLLRLPRDSSlMR

APS

atd

fpm1

s4

Ait

oK

p8

Vps

K

i

maa1

mtv

In

O

sE

s1

W(

tK

sr

Bat1

s

RD2

It

afi

Cmp

RRf

m

paf

W

LEON ET AL36

isch, RD, LD; Susan Dombrowski, MEd, RD, LD; Evaetkac Donnelly, RD, LD; Patricia W. Ellis, MS, RD, CSR,D; Laura Eusanio, RD, LD; Mary Gamberale, RD, LD,DE; Suzanne Gregory, RD, LD; L. Gail Groves, RD, LD;eborah A. Hutsler, MS, RD, LD; Linda Janson, RD, LD;amela S. Kent, MS, RD, CSR, LD; Jennifer Kernc, RD,D; Linda Lackney, MS, RD, LD; Margaret L. Lander, MS,D; Marla Lipman, MS, RD, LD; Gina M. Mendiola, RD,D; Lisa A. Miller, MS, RD, LD; Lois A Morris, RD, CSR,D; Eileen Moore, CNSD, RD, LD; Christine M. Muñoz,D, LD; Donna K. Neroni, RD, LD; Heather Ohlrich, RD,D; Sally Oneacre, RD, LD; Kristin E. Paccione, MS;amela Pochatila, MS, RD, LD; Chatura Ravishankar, MS,D, LD; Kristin Roach, RD, LD; JoAnn Ruggeri, RD, LD;iane Rupp, RD, LD; Laura Schoeffler, RD, LD; Janet B.chueller, MS, RD, LD; Kathy Seese, MS, RD, LD; Maxinemith, RD, LD; Jeanette Soinski, RD, LD; Hollie Sunder-

and, RD, LD; Camille Switzer, RD, LD; Virginia M. Viselli,Ed, RD, LD; Alice Watkins, RD, LD; Melissa A. Wilson,D, LD; and Wendy Youmans, MS, RD, LD, for their help.

REFERENCES1. Centers for Medicare and Medicaid Services: 2004

nnual Report, ESRD Clinical Measures Performanceroject. Baltimore, MD, Department of Health and Humanervices, Centers for Medicare and Medicaid Services, 20042. Owen WF, Lew NL, Liu Y, Lowrie EG, Lazarus JM, et

l: The urea reduction ratio and serum albumin concentra-ion as predictors of mortality in patients undergoing hemo-ialysis. N Engl J Med 329:1001-1006, 19933. Chertow GM, Lazarus JM. Malnutrition as a risk

actor for morbidity and mortality in maintenance dialysisatients, in Kopple JD, Massry SG, eds: Nutritional Manage-ent of Renal Disease. Baltimore, MD, Williams & Wilkins;

997, pp 257-2764. Gabay C, Kushner I: Acute-phase proteins and other

ystemic responses to inflammation. N Engl J Med 340:448-54, 19995. Ikizler TA, Wingard RL, Harvell J, Shyr Y, Hakim RM:

ssociation of morbidity with markers of nutrition andnflammation in chronic hemodialysis patients: A prospec-ive study. Kidney Int 55:1945-1951, 1999

6. Kaysen GA, Stevenson FT, Depner TA: Determinantsf albumin concentration in hemodialysis patients. Am Jidney Dis 29:658-668, 19977. Sehgal AR, Leon J, Soinski JA: Barriers to adequate

rotein nutrition among hemodialysis patients. J Renal Nutr:179-187, 19988. Leon JB, Majerle AD, Soinski JA, Kushner I, Ohri-

achaspati P, Sehgal AR: Can a nutrition intervention im-rove albumin levels among hemodialysis patients? A pilottudy. J Ren Nutr 11:9-15, 2001

9. Kimmel PL: Psychosocial factors in dialysis patients.idney Int 59:1599-1613, 200110. Mitch WE, Maroni BJ: Factors causing malnutrition

n chronic uremia. Am J Kidney Dis 33:176-179, 199911. Movilli E, Zani R, Carli O, et al: Correction ofetabolic acidosis increases serum albumin concentrations

nd decreases kinetically evaluated protein intake in hemodi-lysis patients: A prospective study. Nephrol Dial Transplant

12. Blagg CR, Liedtke RJ, Batjer JD, et al: Serum albu-in concentration-related Health Care Financing Adminis-

ration quality assurance criterion is method-dependent: Re-ision is necessary. Am J Kidney Dis 21:138-144, 199313. Mehrotra R, Berman N, Alistwani A, Kopple JD:

mprovement of nutritional status after initiation of mainte-ance hemodialysis. Am J Kidney Dis 40:133-142, 200214. Willett W: Nutritional Epidemiology. New York, NY,

xford, 1998.15. Detsky AS, McLaughlin JR, Baker JP, et al: What is

ubjective global assessment of nutritional status? J Parenternteral Nutr 11:8-13, 198716. Ottery FD: Definition of standardized nutritional as-

essment and interventional pathways in oncology. Nutrition2:S15-S19, 1996 (suppl 1).17. Hays RD, Kallich JD, Mapes DL, Coons SJ, CarterB: Development of the Kidney Disease Quality of Life

KDQOL) instrument. Qual Life Res 3:329-338, 199418. National Kidney Foundation: K/DOQI Clinical Prac-

ice Guidelines for Nutrition in Chronic Renal Failure. Am Jidney Dis 35:S1-S140, 2000 (suppl 2)19. Craven JL, Rodin GM, Littlefield C: The Beck Depres-

ion Inventory as a screening device for major depression inenal dialysis patients. Int J Psychiatry Med 18:365-374, 1988

20. Sehgal AR, Leon JB, Siminoff LA, Singer ME,unosky LM, Cebul RD: Improving the quality of hemodi-lysis treatment: A community-based randomized controlledrial to overcome patient-specific barriers. JAMA 287:1961-967, 200221. Hulley SB, Cummings SR: Designing Clinical Re-

earch. Baltimore, MD, Williams & Wilkins, 198822. US Renal Data System: USRDS 2004 Annual Data

eport. The National Institutes of Health, National Institute ofiabetes and Digestive and Kidney Diseases, Bethesda, MD,004

23. Caglar K, Fedje L, Dimmitt R, Hakim RM, Shyr Y,kizler TA: Therapeutic effects of oral nutritional supplemen-ation during hemodialysis. Kidney Int 62:1054-1059, 2002

24. Kalantar-Zadeh K, Braglia A, Chow J, et al: Annti-inflammatory and antioxidant nutritional supplementor hypoalbuminemic hemodialysis patients: A pilot/feasibil-ty study. J Ren Nutr 15:318-331, 2005

25. Rammohan M, Kalantar-Zadeh K, Liang A, Ghossein: Megestrol acetate in a moderate dose for the treatment ofalnutrition-inflammation complex in maintenance dialysis

atients. J Ren Nutr 15:345-355, 200526. Moore H, Reams SM, Wiesen K, Nolph KD, Khanna

, Laothong C: National Kidney Foundation Council onenal Nutrition survey: Past-present clinical practices and

uture strategic planning. J Ren Nutr 13:233-240, 200327. McGlynn EA, Asch SM: Developing a clinical perfor-ance measure. Am J Prev Med 14:S14-S21, 1998 (suppl 3)28. Lowrie EG, Lew NL: Death risk in hemodialysis

atients: The predictive value of commonly measured vari-bles and an evaluation of death rate differences betweenacilities. Am J Kidney Dis 15:458-482, 1990

29. Rocco MV, Soucie JM, Reboussin DM, McClellanM: Risk factors for hospital utilization in chronic dialysis

atients. J Am Soc Nephrol 7:889-896, 1996