Kidney International, Vol. 59 (2001), pp.

2361–2367

Children with hypoalbuminemia on continuous peritoneal

dialysis are at risk for technique failure
SANJEEV GULATI, DEREK STEPHENS, JUDITH A. BALFE, DONNA SECKER, ELIZABETH HARVEY,
and J. WILLIAMSON BALFE
Department of Pediatrics, Divisions of Nephrology, Population Health Sciences, and Clinical Nutrition, The Hospital for
Sick Children, University of Toronto, Toronto, Ontario, Canada

Children with hypoalbuminemia on continuous peritoneal dial- CPD, which is associated with an increased incidence of CPD
ysis are at risk for technique failure. failure.
Background. Few data are available on the clinical signifi-
cance of hypoalbuminemia [serum albumin (SA) ⬍35 g/L] in
children with end-stage renal disease (ESRD) on continuous
peritoneal dialysis (CPD). This study was conducted to analyze Low serum albumin (SA) is reported to be an adverse
the prevalence of hypoalbuminemia, its predictive factors, and prognostic factor in adults with end-stage renal disease
its clinical impact in these children. (ESRD) on continuous peritoneal dialysis (CPD) [1, 2].
Methods. A retrospective analysis was done of 180 patients However, analbuminemic rats are able to live a normal
on CPD over the last 22 years. Patients were excluded from
life span, suggesting that perhaps the underlying process
the study if they were on CPD for less than four months or had
nephrotic syndrome. Demographic, clinical, and biochemical rather than hypoalbuminemia itself causes the increased
variables were studied. Children continued on CPD until they mortality [3]. Few data are available on the prevalence or
received a transplant or were transferred to an adult unit or to prognostic implications of hypoalbuminemia in children
hemodialysis as a result of technique failure. The subjects were with ESRD on CPD. Most studies have investigated
divided into two groups based on SA levels at last follow-up.
Results. A total of 135 children was included. After a mean
adult patients, including diabetics, and have been cross-
duration of CPD of 573 ⫾ 437 (120 to 2960) days, 54 children sectional rather than longitudinal [1, 4–9]. These studies
(40%) were observed to have hypoalbuminemia. Four patients found considerable variability in the factors predictive
(2.9%) died, 7 (5.2%) continued on continuous cyclic perito- of a low SA level in patients with ESRD on CPD. This
neal dialysis, and 13 (9.6%) were transferred to an adult unit study was conducted to analyze the prevalence, clinical
for continuation of CPD. Ninety-five (70.3%) were transplanted,
and 16 (11.8%) were transferred to hemodialysis because of relevance, and prognostic value of hypoalbuminemia in
technique failure. Children in group I (N ⫽ 54, SA ⬍35 g/L), children with ESRD who were on CPD.
compared with group II (N ⫽ 81, SA ⱖ35 g/L), were younger
at initiation of PD, more likely to have hypoalbuminemia at
one month and six months after initiation of PD, and have METHODS
more episodes of peritonitis. No differences were seen between Clinical data
the groups in gender, modality of CPD, body surface area,
initial body mass index, and presence of hypertension or acido- A retrospective analysis was done of 180 patients on
sis. The only factors predictive of hypoalbuminemia on follow- CPD at The Hospital for Sick Children over the last 22
up were low SA at one month after PD and recurrent peritonitis years (1978 to 1999). These patients have been under the
using multiple logistic regression analysis. Evaluating the clini-
continuing care, treatment, and follow-up of the senior
cal impact of hypoalbuminemia, we observed a higher inci-
dence of failed PD in children who had hypoalbuminemia. author (J.W.B.). Patients were excluded if they were on
Conclusion. Low SA at one month after PD and recurrent CPD for less than four months or had nephrotic syn-
peritonitis are predictive of hypoalbuminemia in children on drome (focal segmental glomerulosclerosis and congeni-
tal nephrotic syndrome) as the underlying disease. If a
child returned to PD after a failed transplant, his or her
Key words: end-stage renal disease, peritonitis, serum albumin, prog-
nostic factor, continuous cyclic peritoneal dialysis, kidney failure in subsequent time on PD was not considered for the study;
children. the reason was to exclude the effect on SA of drugs and
Received for publication June 16, 2000
of changes in the patient’s nutritional and anabolic status
and in revised form November 27, 2000 caused by renal transplant. From 1979 to 1984, children
Accepted for publication January 18, 2001 were initiated on continuous ambulatory peritoneal dial-
 2001 by the International Society of Nephrology ysis (CAPD). Since 1985, most children were started on

2361
2362 Gulati et al: Children with hypoalbuminemia on CPD

continuous cyclic peritoneal dialysis (CCPD) and some Children continued on CPD until they received a trans-
on CAPD switched to CCPD. At each clinic visit, the plant, were transferred to an adult unit upon reaching
children were evaluated for blood pressure, exit site in- 18 years of age, or were transferred to hemodialysis as
fections, and general physical condition. For infants and a result of patient choice or technique failure. Technique
younger children on CAPD, an exchange volume of 40 failure was defined as transfer to hemodialysis because
to 45 mL/kg was prescribed, while older children were of access failure, repeated peritonitis, medical reasons,
maintained on volumes of 1000 to 1100 mL/m2 with at or patient/family burnout.
least four daily exchanges per day. Those on CCPD had
a total therapy time of 10 hours at night with a fill volume Laboratory investigations
of 1100 mL/m2 and a daytime dwell of 50 to 100% of the All children had a baseline evaluation of their SA
nighttime exchange. Although detailed dialysis adequacy (pre-PD SA), total protein, sodium, potassium, creati-
studies were not available for all children, an attempt nine, urea, pH, bicarbonate, calcium, phosphorus, alka-
was made to maximize the dialysis volume prescription line phosphatase, cholesterol, and complete blood count
as clinically tolerated (maximum of 1400 mL/m2). before insertion of the dialysis catheter. These measure-
A baseline evaluation of patients’ height and weight ments were repeated within one month and every three
was done, and the measurements were repeated at each months thereafter. SA was measured by the bromcresol
clinic visit every three months. Height was measured green method on the Kodak-Ektachem Series 700 ana-
with a stadiometer by the same nurse and calculated as an lyzer: Methodologies-1986 (Eastman Kodak, Rochester,
average of three measurements. Weight was measured NY, USA) [12]. Hypoalbuminemia was defined as a SA
using an electronic weighing scale, Scaletronix 6002 level of ⬍35 g/L at last follow-up. This cut-off point was
(Scale-Tronix Inc., Wheaton, IL, USA). Body surface arbitrarily chosen because CPD patients with SA above
area was calculated by the Du Bois method based on this level survived longer than those with SA below this
height and weight [10]. Initial and follow-up body mass level in the longitudinal study of Teehan [13].
index (BMI) was calculated on the basis of correspond- The 135 children were divided into two groups for
ing heights and weights as the ratio of weight and height analysis: group I (N ⫽ 54, SA ⬍35 g/L) and group II
squared. Growth was monitored by calculating the (N ⫽ 81, SA ⫽ 35 g/L). The following demographic,
change in height standard deviation scores. clinical, and biochemical variables were studied to evalu-
ate their predictive relationship with SA on last follow-
Nutrition up: SA pre-PD (immediately before initiating dialysis),
At each clinic visit the child had a dietary consultation. SA (1 month PD, SA 1 month after initiating CPD), SA
Although detailed dietary records were not available for (6 months PD, SA after 6 months of CPD), age, gender,
all patients, the dietary prescription in terms of protein time on PD, modality of PD (CAPD vs. CCPD vs. both),
and calories was maximized in each child as per the BMI, body surface area, serum bicarbonate levels, blood
recommended guidelines. Children between the 3rd and pressure, number of episodes of peritonitis (2 episodes
97th percentiles for height were targeted to receive 100% vs. ⬎2 episodes), and basic renal disease. Peritonitis was
of the recommended nutrient intake for calories based diagnosed in the presence of at least two of the following
on chronological age and actual weight (if weight was 90 three criteria: (1) symptoms (fever, abdominal pain, cloudy
to 120% of ideal body weight) or according to ideal body dialysate), (2) elevated white blood cells ⬎100/␮L,
weight if actual weight was ⬎120%. Children less than ⬎50% polymorphonuclear cells, and (3) positive gram
the 3rd percentile were targeted to receive a caloric intake stain or culture from PD fluid.
calculated according to height age and ideal body weight. The clinical impact of hypoalbuminemia was analyzed
These children were given protein supplements as per in terms of (1) serum cholesterol levels, (2) growth as
standard recommendations in the following amounts: defined by change in height standard deviation scores,
infants, 2.5 to 3.5 g/kg/day; 2 to 5 years, 2.5 g/kg/day; 5 (3) number of hospitalizations, (4) number of infections
to 12 years, 2 g/kg/day; and ⬎12 years, 1.5 g/kg/day [11]. (exit site, urinary tract infection, sepsis), and (5) clinical
Children whose nutrition was suspected of being inade- status at the end of the study (transplant, continuing
quate were assessed in detail by a renal dietitian using CPD, transition to an adult unit, or technique failure
three-day food records. Supplementary nutrition was with transfer to hemodialysis).
provided to some children by gastrostomy or gastrojeju-
nostomy tube feeds. In addition to monthly clinic visits, Statistical analysis
patients were also followed up by the dialysis nurse and All data are given as mean ⫾ SD. Initially, all the
the dietitian by telephone. For purposes of analysis, the factors were analyzed for the predictive significance of
underlying renal disease was classified as (1) chronic low SA on follow-up using univariate logistic regression
glomerulonephritis, (2) obstructive/interstitial renal dis- analysis. The factors that were moderately (P ⫽ 0.20)
ease, (3) hereditary renal disease, or (4) miscellaneous. associated with low SA on univariate analysis were then
 Gulati et al: Children with hypoalbuminemia on CPD 2363

considered for inclusion into a multiple model at a subse- There were four (2.9%) deaths. Two children died
quent stage. Factors that were highly correlated, with from unrelated causes, leukemia in one case and gastro-
Pearson correlation coefficient ⬎0.5, were not entered intestinal bleed in the other. Only one of them had hypo-
into the model together. Within a group of correlated albuminemia. Two other children, both of whom had
variables, the variables that were considered to be clini- hypoalbuminemia, died because of technique failure:
cally most important were selected to go into the multiple one each due to peritonitis and membrane failure. Of
model. A best subsets selection was also performed. This the 131 surviving patients, 7 (5.2%) continued on CCPD.
method selected the best two factors and higher dimen- Thirteen (9.6%) were transferred to adult nephrology
sion models. The final model consisted of variables meet- units for continuation of CPD. Ninety-five (70.3%) re-
ing the 5% level of significance, those influencing the ceived a transplant, and 16 (11.8%) were transferred to
significance of other variables and those deemed to be hemodialysis as a result of technique failure. The causes
of clinical relevance. For example, SA pre-PD and at six of technique failure were repeated peritonitis in 10 pa-
months closely correlated with the SA at one-month tients, access failure in 3, obesity in 2, and patient/family
post-PD. Hence, both of these were excluded from the choice secondary to burnout in 1.
multiple regression model. Similarly, separate univariate Based on SA levels at last follow-up, the two groups
and multiple regression analysis was done to evaluate were compared using several variables. Univariate analy-
the clinical impact of hypoalbuminemia in these children sis indicated that children in group I (N ⫽ 54, SA ⬍35
using a separate set of independent variables. Calcula- g/L) were significantly younger at initiation of PD (P ⫽
tions were performed using the SAS 6.12 (SAS, Cary, 0.05) than those in group II (N ⫽ 81, SA ⫽ 35 g/L) and
NC, USA) statistical analysis software. A P value ⬍0.05 were more likely to have low SA at one-month PD (P ⫽
was considered significant. 0.001) and six-months PD (P ⬍ 0.0001). They also had
tendency for multiple (⬎2 episodes) of peritonitis (P ⫽
0.018). The groups did not differ in gender, modality of
RESULTS
CPD, body surface area, initial BMI, and presence of
Of the 180 children, 135 satisfied the inclusion criteria hypertension or acidosis (Table 1). Using multiple logis-
and constituted the study group. There were 79 boys. The tic regression, only two factors predicted hypoalbumi-
mean age at initiation of CPD in the study group was nemia in these children: low SA at one-month PD (P ⫽
10.1 ⫾ 5.6 years. Of the study group patients, 61 (45.2%) 0.003) and multiple episodes of peritonitis (P ⫽ 0.037;
had ESRD secondary to chronic glomerulonephritis, 52 Table 2). The estimated probability of having low SA
(38.5%) had obstructive/interstitial nephropathy, 12 (8.9%) on follow-up was higher if there was a low SA at one
had hereditary diseases (polycystic kidney disease, Al- month after initiation of PD and more than two episodes
port syndrome, nephronophthisis), and 10 (7.4%) had of peritonitis (Fig. 1). However, there was no correlation
miscellaneous causes (Wilm’s tumor, renal vein throm- of either the SA pre-PD or SA at one-month PD with
bosis). At last follow-up, after a mean duration of CPD peritonitis. The number of children with multiple epi-
of 573 ⫾ 437 days (120 to 2960 days), 54 (40%) children sodes of peritonitis in the low pre-PD SA group (⬍35
were observed to have hypoalbuminemia. Of the study g/L) was similar to those who had normal pre-PD SA
population, 63 subjects (46.7%) were on CAPD, 34 (4 out of 45 vs. 20 out of 90, P ⫽ 0.58). Similarly, the
(25.2%) were on CCPD, and 38 (28.1%) initially on number of children with multiple episodes of peritonitis
CAPD were transferred to CCPD. Of the 135 children, in low SA at one-month PD was similar to those who
18 returned to PD after a failed transplant. The subse- had normal SA at one-month PD (12 out of 52 vs. 18
quent period of PD was not considered for analysis. out of 83, P ⫽ 0.73). We also found that SA at one-
Fifty-six patients (41.5%) had no episodes of peritonitis, month PD correlated very well with both the pre-PD
30 (22.2%) had one episode, 20 (14.8%) had two epi- SA (R2 ⫽ 32.2%, P ⬍ 0.00001) and the SA at six-months
sodes, and 29 (21.5%) had more than two episodes. Fifty- PD (R2 ⫽ 34.2%, P ⬍ 0.00001). When the mean SA
two (38.8%) had other infections (urinary tract infection, during the study period was correlated with technique
sepsis, exit site infection). Sixty-five (48.1%) of the 135 failure, the incidence of technique failure was significantly
children required antihypertensive therapy. During this higher in children with low mean SA (⬍35 g/L) as com-
period on PD, these children were admitted to the hospi- pared with those with normal mean SA levels (10 out
tal for various reasons: PD training reinforcement, blood of 48 vs. 6 out of 87, P ⫽ 0.02).
transfusions, exit site infections (feeding tube and PD When the clinical impact of hypoalbuminemia was
catheter), peritonitis, and other infections. Of the study evaluated by univariate analysis, there was a significantly
group children, 25 (18.8%) had no hospitalizations, 21 higher incidence of failed CPD in children who had hypo-
(15.8%) had one hospital admission, 25 (18.8%) had two albuminemia (P ⫽ 0.014). Groups I and II did not differ
admissions, and 64 (47.4%) had three or more admis- in serum cholesterol level, number of hospital admis-
sions during the course of PD. sions, number of other infections, follow-up BMI, and
2364 Gulati et al: Children with hypoalbuminemia on CPD

Table 1. Factors predictive of low serum albumin levels

Hypoalbuminemia Normal serum albumin

Factors (N ⫽ 54) (N ⫽ 81) P value
Age years 8.95 ⫾ 5.8 10.85 ⫾ 5.3 0.05
Gender M:F 35:19 44:37 0.22
Duration of CPD days 637 ⫾ 505 531 ⫾ 384 0.17
Body surface area m2 0.96 ⫾ 0.43 1.33 ⫾ 2.8 0.34
Hypertension 26/54 39/81 0.99
Serum bicarbonate 26 ⫾ 3.8 25.5 ⫾ 5.8 0.57
Serum albumin pre-PD 33.9 ⫾ 7.0 35.8 ⫾ 6.4 0.19
Serum albumin
1 month PD 33.7 ⫾ 5.7 37 ⫾ 5.0 0.001
6 months PD 33.2 ⫾ 5.1 38.8 ⫾ 8.3 0.00001
Initial BMI kg/m2 18 ⫾ 3 17 ⫾ 3 0.26
Episodes of peritonitis
⫽2 36 69 0.018
⬎2 18 12
Modality
CAPD 28 35 0.51
CCPD 11 23
Both 15 23
Abbreviations are: CPD, continuous peritoneal dialysis; BMI, body mass index; CAPD, continuous ambulatory peritoneal dialysis; CCPD, continuous cyclic
peritoneal dialysis.

Table 2. Predictors of low serum albumin using multiple

regression model

95% CI
Predictor Coefficient P value Odds ratio Upper Lower
Age ⫺0.03 0.29 0.96 0.90 1.03
Serum albumin
1 month PD ⫺0.11 0.003 0.89 0.83 0.96
Peritonitis 1.05 0.037 2.58 1.06 6.28
PD is peritoneal dialysis.

growth as evaluated by the delta height standard devia-

tion scores (Table 3). On multivariate analysis, children
with hypoalbuminemia (group I) had a significantly higher Fig. 1. Estimated probability of low serum albumin (SA) at follow-
up: Relationship with SA at one-month peritoneal dialysis (PD) and
incidence of technique failure than those in group II episodes of peritonitis. Symbols are: (⫹) ⬎2 episodes of peritonitis;
(P ⫽ 0.047, odds ratio 3.34, 95% CI 1.02 to 10.97; Table 4). (䊊) ⱕ2 episodes of peritonitis.

DISCUSSION
Hypoalbuminemia has been correlated with mortality Studies on the clinical significance of hypoalbumi-
and morbidity in adult patients on both hemodialysis nemia in adults on CPD have yielded varying results.
and PD [1, 2, 4]. The amount of albumin in the body The prevalence of hypoalbuminemia has been reported
and concentration of albumin in serum are determined to vary between 26 and 51% [7, 9, 17]. Data on the
by (1) albumin synthesis, which is affected by dietary factors predictive of a low SA in CPD patients are con-
protein intake and hepatic protein synthesis; (2) protein flicting. The factors most consistently reported to be
catabolism and external losses; and (3) extravascular al- associated with it are advanced age and diabetes. Al-
bumin distribution [14]. In addition to other factors that though a few of these studies have demonstrated a corre-
affect SA metabolism in advanced renal failure, CPD lation between low SA and high/high-average peritoneal
patients are exposed to another influence on SA unique membrane transport characteristics, normalized protein
to CPD—protein losses in the dialysate—which are catabolic rate, and the dialysis dose [1, 6, 9], most others
made worse with peritonitis, particularly if it is severe have failed to demonstrate such a relationship [5, 8, 18].
or persistent [8, 15]. A lower dietary protein intake has In contrast, few studies have questioned the significance
been reported in adult patients on CPD than in those of the small reductions in SA. One study reported that
on hemodialysis [16]. the low SA at onset merely reflected the presence of
 Gulati et al: Children with hypoalbuminemia on CPD 2365

Table 3. Clinical impact of low serum albumin levels

Hypoalbuminemia Normal serum albumin

Factors (N ⫽ 54) (N ⫽ 81) P value
Serum cholesterol mmol/L 5.19 ⫾ 2.01 5.59 ⫾ 4.29 0.53
Hospitalizations
⫽2 26 48 0.27
⬎2 29 33
Other infections 25/54 27/81 0.10
Delta height standard deviation scores 0.10 ⫾ 0.68 0.07 ⫾ 0.74 0.15
Follow-up BMI kg/m2 19 ⫾ 3 18 ⫾ 4.9 0.52
Technique failure 11/52 5/79 0.014
BMI is body mass index.

Table 4. Outcome of low serum albumin using stepwise multiple regression model
95% CI
Predictor Coefficient P value Odds ratio Upper Lower
Technique failure 1.20 0.047 3.34 1.02 10.97
Serum cholesterol ⫺0.07 0.36 0.92 0.78 1.10
Hospitalizations 0.04 0.16 1.04 0.98 1.10
Other infections 0.57 0.16 1.76 0.80 3.89
Delta height standard deviation scores 0.32 0.31 1.38 0.74 2.57

systemic disease like diabetes, which determined the pa- Thus, the low SA could be related to systemic disease
tient’s survival [19]. or the underlying uremic state. It is noteworthy that SA
Hypoalbuminemia as defined by low SA on last follow- one month after initiation of PD correlated more closely
up was present in 40% of children on CPD. SA at the with albumin on last follow-up than the SA before initia-
last follow-up was taken as the cut-off value because tion of PD. It is believed that SA is an inverse acute
early values reflect the start, whereas end values reflect phase reactant; the pattern of cytokine activation associ-
the effect of CPD (nutrition, dilution, loss in dialysate). ated with acute phase reaction could increase vascular
In addition, there are known transient fluctuations in SA and peritoneal permeability [7, 21].
that might have no prognostic value over the longer Repeated peritonitis (⬎2 episodes) were also ob-
period. However, we observed that there was a linear served to be predictive of low SA on follow-up in these
correlation between pre-PD SA and SA values at one- children (P ⫽ 0.018), probably because the effects of
month PD as well as between SA at one-month PD uncomplicated and recurrent peritonitis differ. Uncom-
and at six months PD. Furthermore, there was a close plicated peritonitis is characterized by clinical improve-
correlation of mean SA during the study period with ment and declining dialysate white cell counts. On the
technique failure. The incidence of technique failure was other hand, recurrent peritonitis causes greater dialysate
significantly higher in children with a low mean SA (10 albumin losses, decreased protein intake secondary to
out of 48) in common to those with normal mean SA (6 anorexia, and decreased hepatic protein synthesis medi-
out of 87, P ⫽ 0.02). In a pilot study by Scolnik and Balfe, ated by inflammatory cytokines, lipid mediators, and
persistent hypoalbuminemia and hypercholesterolemia hormones [5, 15, 22]. The end result is hypoalbuminemia.
were reported to be prevalent in 43% (23 out of 53) of Hypoalbuminemia (and possibly hypogammaglobuli-
children on CPD [12]. Only one other study of children nemia) can also predispose to the development of perito-
addressed this issue, and it involved only 15 subjects. nitis. This appears to be unlikely at least in the present
Low SA (⬍35 g/L) was present in 8 out of 15 (53.3%) study, as we observed that there was no correlation of
of the children at onset of CPD. Only 5 out of 15 children both the SA pre-PD and SA at one-month PD with
had a follow-up of ⬎18 months. The factors predictive peritonitis. However, a significantly greater number of
of a low SA were age of less than years and peritoneal children with low SA at last follow-up had more than
protein losses [20]. Neither of these studies addressed the two episodes of peritonitis as compared with those with
clinical impact of hypoalbuminemia. We also observed, normal SA at last follow-up. This suggests that possibly
using univariate analysis, that children with low SA were repeated peritonitis and consequent albumin losses con-
likely to be younger at initiation of PD (P ⫽ 0.05) and tributed to the development of hypoalbuminemia.
more likely to have hypoalbuminemia at one-month PD Like Cueto-Manzano et al, we did not find nutritional
(P ⫽ 0.001) and also at six months PD (P ⬍ 0.00001). status as assessed by the initial BMI to be predictive of
2366 Gulati et al: Children with hypoalbuminemia on CPD

a low SA, perhaps because our patients were followed up lated with higher mortality in adults [1, 5] and has been
closely by the dietitian and because our hospital adhered reported to predict other adverse clinical outcomes in
to a policy of aggressive nutritional supplementation us- CPD [1, 2, 17, 26].
ing tube feeding [23]. Furthermore, SA is influenced Teehan et al, in a study of 51 patients over a five-
strongly by non-nutritional factors, such as comorbidity year period, found that SA predicted mortality in CAPD
and peritoneal albumin losses. BMI is not a comprehen- patients better than kinetic studies [26]. Blake et al, in
sive marker of nutritional status as it depends on weight, a study of 78 adult patients, found that SA correlated
which can fluctuate markedly with fluid overload in these with days hospitalized, fatigue index, nerve conduction,
children. It also does not take into account the amount and technique failure [1]. In our study as well, after a
of lean body mass versus fat mass. Perhaps using an mean duration of CPD of 573 ⫾ 473 days, there was
adapted form of Subjective Global Assessment would a significantly higher incidence of technique failure in
be a better index of nutrition in these children [24]. The children with low SA (P ⫽ 0.047, odds ratio ⫽ 3.34,
CANUSA study also observed that SA was not influ- 95% CI ⫽ 1.02 to 10.97). Interestingly, both the patients
enced by the dialysis dose or nutrition but remained a whose deaths were related to peritonitis and membrane
strong predictor of outcome [24]. This is possibly due to failure also had hypoalbuminemia. In our study, re-
fact that SA is influenced more strongly by non-nutri- peated peritonitis was the most common cause of tech-
tional factors such as comorbidity and peritoneal albu- nique failure: 9 of the 10 patients were hypoalbuminemic.
min losses. Infections remain an important comorbidity In three other children, it was secondary to failure of
as they stimulate protein catabolism and increase protein ultrafiltration, that is, access failure (only one of them
losses. Studies reveal that while SA is definitely affected had normal SA levels). Hypoalbuminemia could have
by malnutrition, it is also independently affected by a predisposed to both of these factors. Obesity in two
high dialysate to plasma-creatinine ratio, increasing age, patients and patient/family choice secondary to burnout
hypervolemia and inflammation [17–24]. Most of them in another patient were the other causes of technique
are independent predictors of mortality as well. Thus, failure. All three of these patients had normal SA levels.
SA is not merely a nutritional marker but is of far greater The precise reason why hypoalbuminemia is associated
prognostic significance. technique failure is unknown, although it could be due
Acidosis was not associated with the low SA levels, a to the fact that SA is an inverse acute phase reactant.
finding that might be explained by the fact that few of Hypoalbuminemia (and possibly hypoglobulinemia) pre-
our patients were acidotic and the mean serum bicarbon- disposes to recurrent infectious (exist site and peritoni-
ate level in these children was 23 ⫾ 2.3 mmol/L. In tis). However, we did not observe any correlation of low
studies done in adults, the data on correlation of low SA SA with pre-PD SA or follow-up BMI, which was used
with peritoneal membrane transport characteristics and as an index of nutrition assessment. In contrast to earlier
urea kinetic indices have been variable and conflicting research, no correlation was found between hospitaliza-
[1, 5, 9, 18, 23, 25]. Our study did not analyze the correla- tion and low SA, perhaps because indications for hospi-
tion among creatinine clearance, peritoneal equilibration talization varied over the 22-year study period. Children
test (PET) results, and SA, as the patients were studied were once routinely admitted for blood transfusions, but
over a long period, and some of them before the intro- with the availability of recombinant human erythropoie-
duction of adequacy and PET studies. Although urine tin, this is no longer true. Similarly, in earlier times,
protein excretion data were not available, children with children with peritonitis would invariably be admitted;
massive protein losses such as those with nephrotic syn- now, with greater experience, many are being managed
drome (focal segmental glomerulosclerosis and congeni- as outpatients. Although the mean serum cholesterol in
tal nephrotic syndrome) were excluded in order to mini- our study group was high, no correlation was observed
mize the effect of residual proteinuria on SA. On multiple between low SA and cholesterol, possibly because ele-
logistic regression, the only two factors predictive of vated serum glucose levels could have an effect on the
hypoalbuminemia in these children were low SA at one cholesterol values [12].
month after PD (P ⫽ 0.003) and multiple episodes of In summary, of the 135 children on CPD over the last
peritonitis (P ⫽ 0.037). Of note is the fact that the low 22 years in this study, 54 (40%) were observed to have
SA at one month had an independent and also greater hypoalbuminemia. Low SA at one month after starting
predictive value for low SA on follow-up than peritonitis. PD and recurrent peritonitis are predictive of hypoal-
Thus, low SA on follow-up was unlikely to be the sole buminemia in children on CPD. These children have an
result of recurrent peritonitis. increased incidence of CPD technique failure. Thus, low
Serum albumin predicts survival because it reflects not SA at one month is a harbinger of a high incidence of
only nutrition but also systemic disease. No pediatric peritoneal dialysis failure, and it is possible that protein
data are available on its clinical significance and prognos- supplementation may prevent peritoneal and PD failure.
tic value. The SA concentration ⬍35 g/L has been corre- Prospective studies are required to validate the causes
 Gulati et al: Children with hypoalbuminemia on CPD 2367

of low SA in these children and to devise strategies to 11. Nelson P, Strover J: Principles of nutritional assessment and
management in children with renal insufficiency, in End Stage
correct it (increasing dietary protein intake or aggressive Renal Disease in Children, edited by Fine RN, Philadelphia, W.B.
ultrafiltration). It also would be important to evaluate Saunders, 1984, pp 209–226
whether correction of low SA affects outcome (technique 12. Scolnik D, Balfe JW: Initial hypoalbuminemia and hyperlipid-
emia persist during chronic peritoneal dialysis in children. Perit
failure or mortality). Dial Int 13:136–139, 1993
13. Teehan BP: Urea kinetic modelling is an appropriate assessment
ACKNOWLEDGMENT of adequacy. Semin Dial 5:189–192, 1992
14. Kaysen GA, Schoenfeld PY: Albumin homeostasis in patients
This article was prepared with the assistance of Editorial Services, undergoing continuous ambulatory peritoneal dialysis. Kidney Int
The Hospital for Sick Children, Toronto, Canada. 25:107–114, 1984
15. Tzamaloukas AH, Obermiller LE, Gibel LJ, et al: Peritonitis
Reprint requests to Dr. J. Williamson Balfe, The Hospital for Sick associated with intra-abdominal pathology in continuous ambula-
Children, 555 University Avenue, Toronto, Ontario, Canada M5G 1X8. tory peritoneal dialysis patients. Perit Dial Int 13(Suppl):S335–
E-mail: bill.balfe@sickkids.on.ca S337, 1993
16. Cianciaruso B, Brunor G, Kipple JD, et al: Cross-sectional com-
parison of malnutrition in continuous ambulatory peritoneal dial-
REFERENCES ysis and hemodialysis patients. Am J Kidney Dis 26:475–486, 1995
1. Blake PG, Flowerdew G, Blake RM, Oreopoulos DG: Serum 17. Malhotra D, Murata GH, Tzamaloukas AH: Serum albumin
albumin in patients on continuous ambulatory peritoneal dialysis: in peritoneal dialysis: Clinical significance and important influences
Predictors and correlations with outcomes. J Am Soc Nephrol on its levels. Int J Artif Organs 20:251–254, 1997
3:1501–1507, 1993 18. Blake PG, Sombolos K, Abraham G, et al: Lack of correlation
2. Struijk DG, Krediet RT, Koomen GC, et al: The effect of serum between urea kinetic indices and clinical outcomes in CAPD pa-
albumin at the start of continuous ambulatory peritoneal dialysis tients. Kidney Int 39:700–706, 1991
treatment on patient survival. Perit Dial Int 14:121–126, 1994 19. Fine A, Cox D: Modest reduction of serum albumin in continuous
3. Kaysen GA, Stevenson FT, Depner TA: Determinants of albumin ambulatory peritoneal dialysis patients is common and of no appar-
concentration in hemodialysis patients. Am J Kidney Dis 29:658– ent clinical consequence. Am J Kidney Dis 20:50–54, 1992
668, 1997 20. Broyer M, Niaudet P, Champion G, et al: Nutritional and meta-
4. Owen WF Jr, Lew NL, Liu Y, et al: The urea reduction ratio and bolic studies in children on continuous ambulatory peritoneal dial-
serum albumin concentration as predictors of mortality in patients ysis. Kidney Int 24(Suppl 15):S106–S110, 1983
undergoing hemodialysis. N Engl J Med 329:1001–1006, 1993 21. Yeun JY, Kaysen GA: Acute phase proteins and peritoneal dialy-
5. Malhotra D, Tzamaloukas AH, Murata GH, et al: Serum albu- sate albumin loss are the main determinants of serum albumin in
min in continuous peritoneal dialysis: Its predictors and relation- peritoneal dialysis patients. Am J Kidney Dis 30:923–927, 1997
ship to urea clearance. Kidney Int 50:243–249, 1996 22. Young GA, Young JB, Young SM, et al: Nutrition and delayed
6. Nolph KD, Moore HL, Prowant B, et al: Cross sectional assess- hypersensitivity during continuous ambulatory peritoneal dialysis
ment of weekly urea and creatinine clearances and indices of nutri- in relation to peritonitis. Nephron 43:177–186, 1986
tion in continuous ambulatory peritoneal dialysis patients. Perit 23. Cueto-Manzano AM, Espinosa A, Hernandez A, et al: Perito-
Dial Int 13:178–183, 1993 neal transport kinetics correlate with serum albumin but not with
7. Han DS, Lee SW, Kang SW, et al: Factors affecting low values the overall nutritional status in CAPD patients. Am J Kidney Dis
of serum albumin in CAPD patients. Adv Perit Dial 12:288–292, 30:229–236, 1997
1996 24. McCusker FX, Teehan BP, Thorpe KE, et al: How much perito-
8. Heimburger O, Bergstrom J, Lindholm B: Albumin and amino neal dialysis is required for the maintenance of a good nutritional
acid levels as markers of adequacy in continuous ambulatory peri- state? Canada-USA (CANUSA) Peritoneal Dialysis Study Group.
toneal dialysis. Perit Dial Int 14:S123–S132, 1994 Kidney Int 50(Suppl 56):S56–S61, 1996
9. Marcus RG, Chaing E, Dimaano F, Uribarri J: Serum albumin: 25. Margetts PJ, McMullin JP, Rabbat CG, Churchill DN: Perito-
Associations and significance in peritoneal dialysis. Adv Perit Dial neal membrane transport and hypoalbuminemia: Cause or effect?
10:94–98, 1994 Perit Dial Int 20:14–18, 2000
10. Du Bois D, Du Bois EF: A formula to estimate the approximate 26. Teehan BP, Schleifer CR, Brown JM, et al: Urea kinetic analysis
surface if height and weight be known. Arch Intern Med 17:863–871, and clinical outcome on CAPD: A five year longitudinal study.
1916 Adv Perit Dial 6:181–185, 1990