A longitudinal analysis of the effect of anemia on health-related quality of life in children with mild-to-moderate chronic kidney disease

Abstract

Background

To evaluate impact of anemia on health-related quality of life (HRQOL) over time in a large pediatric cohort with mild-to-moderate chronic kidney disease (CKD).

Methods

Participants were enrolled in the Chronic Kidney Disease in Children Study (CKiD), a multicenter, longitudinal cohort. HRQOL was measured using the Pediatric Quality of Life Inventory (PedsQL). Anemia was defined as hemoglobin < 5th percentile for age, sex, and race. Two longitudinal analyses were conducted on consecutive visit pairs. Models examined effects of anemia status on both HRQOL score over time and change in HRQOL score between consecutive visits. The sample included 733 children with a median estimated GFR 54 ml/min/1.73 m². Thirty percent of children had anemia at index visit.

Results

Analysis of HRQOL scores revealed the presence of anemia was associated with significantly lower overall HRQOL (β = −2.90 (95% CI = −7.74, −0.21), p = 0.04) and physical functioning (β = −5.72 (−9.49, −2.25), p = 0.001) according to children. On parent ratings, the development of anemia was associated with lower emotional functioning scores (β = −4.87 (−8.72, −0.11), p = 0.045). In the second model, children who developed anemia were rated by caregivers as having more decreased physical functioning than children who remained anemia-free (β = −3.30 per year (−5.83, −0.76), p = 0.01). Caregivers did not observe declines in their children’s other PedsQL subscales in the presence of developed anemia. Children with resolved or persistence did not show improvement or decline in any aspect of HRQOL functioning relative to non-anemic subjects.

Conclusions

In children with CKD, anemia has an adverse effect on HRQOL which persists over time but does not appear to be progressive.

Background

One of the most common and debilitating problems of patients with chronic kidney disease (CKD) is anemia [1]. Anemia often develops as CKD progresses and can be associated with a decrease in energy and fatigue. The presence of anemia regardless of disease progression has been associated with a lower quality of life in a variety of illnesses, including Crohn’s disease and chemotherapy-induced anemia [2, 3]. In adult patients with CKD, anemia has also been associated with lower health-related quality of life (HRQOL) [4–8], which may improve with treatment [9–11]. However, HRQOL improving with the resolution of anemia has not been universal [12]. Among the latter researchers, there is speculation that the treatment of anemia may in and of itself cause distress. Indeed, medication burden and injections can lead to a poorer quality of life, which could mitigate the potential benefits of anemia resolution.

A paucity of literature exists to evaluate the relationship between anemia and quality of life in children. It has been shown that children with CKD tend to have poorer HRQOL than their healthy counterparts, as well as those children with some but not all chronic health conditions [13]. In a cross-sectional study of adolescents with CKD, Gerson and colleagues found that the parents reported worse HRQOL for the children who were anemic [14]. Baek and colleagues [15] also identified anemia as a co-morbidity of children with pre-dialysis CKD that negatively impacts HRQOL. These studies, along with most of the HRQOL pediatric literature, utilize a cross-sectional design. Since hemoglobin levels often fluctuate over time in patients with CKD, it is more desirable to evaluate the impact of anemia longitudinally to evaluate risk factors for CKD progression [16].

Using the Chronic Kidney Disease in Children (CKiD) Study longitudinal cohort, we sought to identify trends in HRQOL for patients who either have a new onset of anemia, persistent anemia, or resolution of anemia relative to patients without anemia. By using two complementary analyses, we sought to determine how anemia affects different aspects of HRQOL over time while comparing results from children and parent reports.

Methods

Study population

Participants were children enrolled in the CKiD, a multicenter, longitudinal, observational cohort study of pediatric patients with CKD at centers across North America [17]. The study predominantly recruits patients with CKD stages one through four. The CKiD study protocol was approved by the Institutional Review Boards at each participating center. Parents or guardians of study participants provided informed consent. HRQOL was measured by the PedsQL (described below), which is administered to study participants and their parents at 6 months after the baseline visit and at each subsequent annual CKiD study visit as well. Hemoglobin is collected at each annual visit. This analysis included patients who contributed both hemoglobin and PedsQL scores for at least two consecutive visits so that change in anemic status and HRQOL over 1-year intervals could be observed. A total of 773 participants contributed a median of 3.2 years of follow-up, creating a sizable database of longitudinal observations.

Measures

Pediatric quality of life scales

The 23-item PedsQL™ 4.0 Generic Core Scales encompass measures assessing physical, emotional, social, and school functioning. In addition, a total score can be computed which measures overall HRQOL as encompassed in the WHO definition [18]. Respondents are asked to indicate the frequency with which each situation has been a problem in the past month using a 5-point categorical scale (never, almost never, sometimes, often, almost always). The reliability and validity of the PedsQL have been consistently affirmed in studies of children with CKD and other chronic medical conditions [19–22]. Scores range from 0 to 100, with higher scores reflecting better HRQOL.

Statistical methods

In order to comprehensively evaluate the relationship between anemia and quality of life, we used two different analytic methods. The first analysis aimed to examine the effects of the time-varying anemia status on the HRQOL scores itself, adjusting for relevant covariates. The second analysis aimed to examine the effects of dynamic anemia status on visit-to-visit change in HRQOL. In both approaches, the unit of analysis was defined as a visit pair—two consecutive study visits (here denoted “index visit” and “subsequent visit”) typically 1 year apart, at which both anemia and HRQOL data were available. Each subject could contribute to multiple such visit pairs to the analysis.

Anemia at each visit was defined dichotomously as hemoglobin value less than the 5th percentile for age, sex, and race according to CDC guidelines. Dynamic anemia status for a visit pair was classified as “none” (no anemia at either visit), “developed” (no anemia at index visit, anemia at subsequent visit), “persistent” (anemia at both visits), or “resolved” (anemia at index visit, no anemia at subsequent visit). The use of iron supplements and erythropoietin stimulating agents (ESA) was not considered in the definition of anemia but was controlled for in the models given that both ESA injections and added pill burden could also affect HRQOL. For each HRQOL subscale, change in HRQOL was defined as the value at the subsequent visit minus the value at the index visit.

The first analysis was designed to investigate the level of HRQOL. Because HRQOL itself was non-normally distributed (both skewed and bounded), quantile regression models were fit at the 25th percentile of HRQOL to examine the effects of dynamic anemia status, adjusted for other covariates. The 25th percentile was chosen because the skewness of the responses results in greater variability in the lower tail, which may thus be more sensitive to covariate effects. The outcome for these models was the HRQOL level at the second visit within a pair. Bootstrapping at the subject level with 2000 replicates was performed to adjust standard errors for repeated observations because each subject could contribute multiple visit pairs.

The second analysis examined the change in HRQOL across each pair of visits. Enhanced box plots were used to visualize the distributions of these changes. This visualization is particularly useful for accurately depicting the tails of a distribution. The shape of the box plot helps indicate the variability of the data and displays where the data are clustered. The width of the box plot is proportional to the value of the percentile shown (e.g., the 25th and 75th percentiles are half as wide as the 50th percentile). These plots also include pairwise tests of each adjacent pair of distribution, with each p value taken from an unadjusted linear mixed model with a random intercept. We tested only a targeted subset of pairwise comparisons to reduce type 1 error risk and highlight differences that would be most meaningful from a clinical perspective. Because the change in HRQOL was distributed normally, we utilized linear mixed models with a random intercept at the subject level to accommodate for repeated observations.

Covariates for the models were chosen based on variables that are well-known from previous HRQOL studies in the literature [14, 23]. Primary exposure variables included the presence of anemia at the index visit, developed anemia, and resolved anemia. Covariates for adjustment included use of any iron supplement, use of an ESA, age, duration of CKD, sex, race, ethnicity, glomerular etiology, estimated GFR (eGFR [24]), urine protein/creatinine ratio, height Z score corrected for age and sex, and maternal education.

All analyses were performed in SAS 9.4 (SAS Institute, Cary, NC). Graphics were produced in R 3.5.1. P values < 0.05 were considered to be statistically significant.

Results

Descriptive statistics of the 773 participants included in the sample are shown in Table 1, with data taken from their first index visit. Participants were 61% male, with a median age of 11 years. All patients had CKD but patients on dialysis were excluded. The median duration of CKD was 8 years and the median eGFR was 54 ml/min/1.73 m². Median hemoglobin was 12.8 g/dL, while age-sex-race-adjusted Z scores indicated this to be somewhat below expected levels (median = −0.7); 30% of participants were below the age-sex-race specific 5th percentile threshold characterizing the presence of anemia at the index visit. Iron supplements were in use by 28% of participants, while erythropoietin-stimulating agents were used by only 6%. Participants contributed a median of 3 visit pairs to the analysis (i.e., 4 years of data).

Table 1.

Description of participants included in this analysis at their first index visit (N = 773)

Characteristic	Median [IQR] or n (%)
Age, years	11 [8, 15]
Duration of CKD	8 [4, 12]
Male sex	471 (61%)
African American race	170 (22%)
Hispanic ethnicity	110 (14%)
Glomerular etiology	226 (29%)
eGFR, ml/min/1.73 m2	54 [40, 67]
Urine protein/creatinine ratio	0.3 [0.1, 0.9]
Height Z score (for age and sex)	− 0.5 [− 1.3, 0.3]
Maternal education
High school or less	299 (40%)
Some college	212 (28%)
College or more	243 (32%)
Hemoglobin (g/dL)	12.8 [11.8, 13.8]
Hemoglobin Z score (for age, sex, and race)	− 0.7 [− 1.9, 0.5]
Anemia (hemoglobin < 5th percentile for age, sex, and race)	231 (30%)
Use of iron supplement	216 (28%)
Use of ESA	50 (6%)
Number of visit pairs analyzed	3 [2, 5]

CKD, chronic kidney disease; eGFR, estimated glomerular function rate; ESA, erythropoietin-stimulating agent; IQR, interquartile range

Effect of anemia status on HRQOL score

Table 2 compares the HRQOL scores between groups with and without prevalent anemia at their first index visit. Medians are the best way to describe HRQOL and are more relevant to our analysis since the distribution of HRQOL is skewed and bounded. However, means for our sample are also reported in order to compare to the normative means available in the literature [13]. A visual examination of the mean HRQOL values for both parent proxies and child self-ratings shows that patients without anemia were similar to normative means in the majority of categories. The one exception was the school subscale, which skewed somewhat lower than the normative mean such that participants with anemia tended to have lower HRQOL scores compared to patients without anemia, and when juxtaposed to the normative means.

Table 2.

Description of participants HRQOL scores at their first index visit (N = 773) compared to established norms of healthy children

PedsQL SCORE	Median [IQR]		Mean ± SD		Normative mean**
	No anemia (n = 579)	anemia (n = 194)	No anemia (n = 579)	anemia (n = 194)
Parent proxy rating
Overall	82 [68, 91]	77 [64, 88]	78.4 ± 15.8	75.5 ± 16.3	81
Physical	91 [72, 97]	84 [66, 94]	82.2 ± 19.8	79.5 ± 19.5	83
Emotional	80 [65, 90]	80 [60, 90]	77.1 ± 18.1	75.4 ± 19.4	80
Social	90 [70, 100]	80 [65, 100]	81.9 ± 19.7	79.8 ± 19.4	82
Child self-rating
School	70 [55, 85]	65 [50, 80]	69.8 ± 20.6	64.5 ± 21.8	77
Overall	82 [71, 89]	75 [64, 88]	78.9 ± 13.6	74.3 ± 15.9	83
Physical	88 [75, 94]	81 [66, 94]	83.4 ± 15.2	78.4 ± 17.5	87
Emotional	80 [65, 90]	75 [60, 90]	76.9 ± 17.6	74.3 ± 20.3	78
Social	90 [80, 100]	80 [65, 100]	85.1 ± 17.7	79.2 ± 19.7	84
School	70 [55, 80]	65 [50, 75]	67.9 ± 18.8	63.2 ± 19.4	80

HRQOL, health-related quality of life; IQR, interquartile range; SD, standard deviation

^**Varni JW, Burwinkle TM, Rapoll MA, Seid M, Skarr D. The PedsQL 4.0 as a pediatric population health measure: Feasibility, reliability, and validity. Ambul Pediatr. 2003;3(6):329–341

Table 3 summarizes the results of adjusted quantile regression models, focusing on the effect of anemia status on the 25th percentile of each HRQOL distribution. Index anemia was associated with significantly lower overall HRQOL and physical functioning according to children, while developed anemia was associated only with lower emotional functioning on the parent ratings (see Table 3). Iron supplements were associated with lower parent-rated social HRQOL (β = −4.81 (−8.96, −0.68), p = 0.02; data not shown).

Table 3.

Quantile regression results: estimated effects on the 25th percentile of HRQOL level

Outcome	Parent proxy rating		Child self-rating
Predictor	Estimate (95% CI)	p value	Estimate (95% CI)	p value
Overall HRQOL
Index anemia	− 2.29 (− 6.50, 1.41)	0.20	− 2.90 (− 7.74, − 0.21)	0.04
Developed anemia	− 2.53 (− 6.96, 0.51)	0.09	− 2.86 (− 5.19, 1.33)	0.23
Resolved anemia	− 1.04 (− 4.44, 3.22)	0.77	0.95 (− 2.84, 5.75)	0.50
Physical HRQOL
Index anemia	− 4.10 (− 9.85, 2.15)	0.21	− 5.72 (− 9.49, − 2.25)	0.001
Developed anemia	− 4.88 (− 9.96, 0.17)	0.07	−3.27 (−6.64, 0.19)	0.06
Resolved anemia	− 1.97 (− 7.52, 4.10)	0.56	0.59 (− 3.19, 5.76)	0.60
Emotional HRQOL
Index anemia	− 3.31 (− 7.44, 1.58)	0.23	− 2.34 (−6.52, 1.63)	0.29
Developed anemia	− 4.87 (− 8.72, − 0.11)	0.045	− 0.43 (− 4.63, 3.21)	0.85
Resolved anemia	− 0.16 (− 6.02, 5.21)	0.86	0.72 (− 4.31, 5.86)	0.78
Social HRQOL Index anemia	− 3.15 (− 8.48, 2.86)	0.32	− 3.62 (− 8.84, 0.79)	0.11
Developed anemia	− 3.02 (− 8.88, 2.48)	0.30	− 1.52 (− 5.81, 3.10)	0.58
Resolved anemia	0.60 (− 5.39, 6.50)	0.84	3.87 (− 1.44, 9.31)	0.15
School HRQOL
Index anemia	0.92 (− 4.11, 4.57)	0.87	− 1.82 (− 5.04, 2.43)	0.56
Developed anemia	− 0.75 (− 5.07, 2.95)	0.59	0.00 (− 3.48, 4.95)	0.81
Resolved anemia	− 0.96 (− 6.66, 3.70)	0.60	− 0.26 (− 3.88, 4.37)	0.94

CI, confidence interval; HRQOL, health-related quality of life

Effect of anemia status on change in HRQOL

Distributions of change in parent- and child-rated overall and physical HRQOL are shown in Fig. 1. Each anemic status group’s distribution is shown in a percentile boxplot, displaying percentiles from the 2.5th to the 97.5th as labeled, with the median value displayed in the center. Along the bottom, p values test the comparison between adjacent groups (i.e., none vs. developed, developed vs. persistent, persistent vs. resolved), each from a separate linear mixed model with random intercept and no covariates aside from anemia status. Visit pairs in which participants developed anemia showed a more negative change in parent-rated overall HRQOL (p = 0.049) and physical functioning (p = 0.004), but there was no difference for the emotional, social, or school subscales (figures not shown). This trend was not duplicated in child ratings, nor was any difference observed on any scale between developed and persistent anemia or persistent and resolved anemia.

Fig. 1.

Distributions of change in HRQOL for parent- and child-rated overall and physical scales. Percentile boxplots of the change in HRQOL scores across pairs of visits based on anemic status across the pair. The median of each group is displayed in the center, and the p values test difference in the means of adjacent distributions using an unadjusted linear mixed model with a random intercept at the subject level

Fully adjusted linear mixed models for these HRQOL change outcomes are summarized in Table 4. “Index anemia” indicates the presence of anemia at the index visit regardless of anemia status at the subsequent visit; “resolved anemia” is thus the effect of resolving anemia compared to having persistent anemia. The effect of developed anemia on parent-rated physical HRQOL, as noted above in the unadjusted analysis of Fig. 1, was again observed after covariate adjustment; however, all other relationships between anemia status and change in HRQOL outcomes were non-significant.

Table 4.

Linear mixed model results: estimated effects on the mean of change in HRQOL

Outcome	Parent proxy rating		Child self-rating
Predictor	Estimate (95% CI)	p value	Estimate (95% CI)	p value
Change in overall HRQOL
Index anemia	− 0.11 (− 1.68, 1.46)	0.89	− 0.36 (− 1.75, 1.04)	0.61
Developed anemia	− 1.42 (− 3.21, 0.37)	0.12	− 0.05 (− 1.70, 1.59)	0.95
Resolved anemia	0.16 (− 2.06, 2.38)	0.89	0.28 (− 1.70, 2.26)	0.78
Change in physical HRQOL
Index anemia	− 0.81 (− 3.03, 1.41)	0.47	− 0.39 (− 2.01, 1.24)	0.64
Developed anemia	− 3.30 (− 5.83, − 0.76)	0.01	0.84 (− 1.08, 2.76)	0.39
Resolved anemia	− 1.17 (− 4.31, 1.98)	0.47	0.25 (− 2.06, 2.55)	0.83
Change in emotional HRQOL
Index anemia	− 0.29 (− 2.25, 1.67)	0.77	0.52 (− 1.66, 2.69)	0.64
Developed anemia	− 0.42 (− 2.66, 1.81)	0.71	0.78 (− 1.79, 3.35)	0.55
Resolved anemia	1.54 (− 1.23, 4.32)	0.28	− 0.78 (− 3.87, 2.31)	0.62
Change in social HRQOL
Index anemia	− 0.11 (− 2.21, 1.99)	0.92	− 0.67 (− 2.60, 1.25)	0.49
Developed anemia	− 1.98 (− 4.38, 0.41)	0.10	− 1.41 (− 3.69, 0.86)	0.22
Resolved anemia	1.17 (− 1.81, 4.15)	0.44	1.75 (− 0.99, 4.48)	0.21
Change in school HRQOL
Index anemia	0.95 (− 1.24, 3.15)	0.40	− 0.82 (− 2.89, 1.25)	0.44
Developed anemia	0.97 (− 1.53, 3.48)	0.45	− 1.08 (− 3.53, 1.36)	0.39
Resolved anemia	0.07 (− 3.04, 3.18)	0.96	− 0.21 (− 3.14, 2.73)	0.89

CI, confidence interval; HRQOL, health-related quality of life

Covariate relationships with HRQOL

We observed generally positive effects of increasing age and CKD duration, height Z score, and maternal education as well as the generally negative effects of urine protein/creatinine ratio and African American race. Table 5 displays the complete results of the quantile regression models for parent- and child-rated overall HRQOL scores. Only two covariates were associated with a visit-to-visit change in HRQOL in linear mixed models. Older age was associated with a more positive change in parent-rated overall HRQOL (β = 0.28, p = 0.008) and social HRQOL (β = 0.35, p = 0.01), and maternal education of college or more was associated with a more negative change in child-rated school HRQOL (β = −1.83, p = 0.04; all data not shown). No other significant effects were observed in any of the ten models for change.

Table 5.

Full quantile regression results for parent- and child-rated Characteristic Parent proxy rating 697 visits Child self-rating 650 visits overall HRQOL

Characteristic	Parent proxy rating 697 visits		Child self-rating 650 visits
	Estimate (95% CI)	p value	Estimate (95% CI)	p value
Index anemia	− 2.29 (− 6.50, 1.41)	0.20	− 2.90 (− 7.74, − 0.21)	0.04
Developed anemia	− 2.53 (− 6.96, 0.51)	0.09	− 2.86 (− 5.19, 1.33)	0.23
Resolved anemia	− 1.04 (− 4.44, 3.22)	0.77	0.95 (− 2.84, 5.75)	0.50
Iron supplement	− 2.25 (− 5.45, 0.50)	0.09	− 2.23 (− 6.24, 0.58)	0.11
Epogen	− 0.88 (− 5.24, 5.07)	0.96	− 1.57 (− 6.80, 5.81)	0.81
Age, years	− 0.43 (− 0.87, 0.24)	0.25	0.50 (0.01, 0.99)	0.048
Duration of CKD, years	0.39 (− 0.18, 0.81)	0.18	0.50 (0.11, 0.91)	0.02
Male sex	− 0.54 (− 3.79, 2.69)	0.76	1.44 (− 1.30, 4.27)	0.28
African American race	− 9.09 (− 12.67, − 4.88)	<0.001	− 4.50 (− 7.96, − 0.31)	0.04
Hispanic ethnicity	0.61 (− 4.47, 5.23)	0.83	3.86 (− 0.85, 7.77)	0.10
Glomerular dx	7.89 (1.44, 12.68)	0.015	5.06 (1.44, 9.80)	0.008
eGFR, per 10 ml/min/1.73 m2	− 0.22 (− 1.33, 1.09)	0.81	− 0.43 (− 1.64, 0.66)	0.42
uP/C ratio, per doubling	− 0.80 (− 1.54, 0.03)	0.06	− 0.49 (− 1.33, 0.22)	0.16
Height Z score	2.83 (1.47, 4.05)	< 0.001	2.16 (1.09, 3.29)	< 0.001
M.E.: some college	5.32 (0.49, 8.85)	0.03	0.84 (− 2.56, 4.32)	0.65
M.E.: college or more	8.23 (3.92, 11.96)	< 0.001	3.92 (0.33, 7.03)	0.03

CI, confidence interval; CKD, chronic kidney disease; eGFR, estimated glomerular function rate; HRQOL, health-related quality of life; M.E., maternal education; P/C, protein:creatinine ratio

Discussion

To the best of our knowledge, this study is the first longitudinal analysis to look at the relationship between HRQOL and anemia in a pediatric population with mild-to-moderate CKD. Anemia is well-known to worsen as CKD progresses but the impact on change in HRQOL is largely unknown in children. Overall, we found anemia can be associated with a decrease in HRQOL, which was expected. As expected, the CKD patients with anemia consistently had lower HRQOL scores in most categories than patients without anemia. This largely remained the case when HRQOL mean ratings for patients with CKD without anemia were juxtaposed to the available normative means. The one exception was on the PedsQL School Subscale where CKD patients had lower ratings on and wider variability, and this was particularly the case for individuals with anemia. It is not clear as to why ratings on the School Subscale were lower, although it is speculated that patients with CKD have more missed school days due to associated illnesses and an increased number of health visits, and this may be even more so for children with CKD and anemia.

The relationship between anemia and HRQOL was more complicated in a longitudinal design. Our first goal was to assess if anemia influenced HRQOL at a specific visit. Similar to previous studies [14, 15], anemia was associated with a lower overall HRQOL as well as lower parental ratings on the physical functioning subscale in the quantile regression. Since children with anemia often have fatigue and decreased energy, this relationship can be supported clinically. Another explanation is that oxidative stress and inflammation related to early-onset anemia could have an impact on HRQOL. This relationship persists regardless of GFR.

However, there were differences when comparing scores from children and parents which differ from previous research [15, 20]. The association between anemia and physical functioning HRQOL subscale was only found when analyzing children’s ratings. On the other hand, parents reported more impact on emotional functioning. The effect was small in magnitude but still significant. Previous adult studies have shown kidney disease can affect emotional functioning and result in a greater number of patients with depression and anxiety [25]. One explanation of our finding is children might have physical complaints but not show outward signs of distress that parents can see or that children may not verbalize their discomfort to their caregiver. Another possible explanation is parents might be more sensitive to emotional functioning simply due to their age/maturity. Al-Uzri and colleagues [23] observed a lack of agreement between parents and children. In their study, looking at short stature and HRQOL, they found parents were more likely to rate children lower on emotional functioning compared to children themselves when looking at multiple variables that affect HRQOL. Both studies emphasize the importance of querying both parents and children since the impact of various factors on HRQOL can be reported differently.

In the second model, the goal was to assess the effect of anemia on the change in HRQOL over time. Only children who developed anemia had a decrease in HRQOL related to physical functioning according to parent ratings. However, the overall HRQOL and other subscales were not affected by anemic status in the adjusted model. Even though anemia affects the level of HRQOL in the first analysis, this effect was not progressive over time. This finding could be due to fluctuations in hemoglobin between study visits, which could confound the results. Medications could also have been added in-between visits, which have been shown to have varying effects on HRQOL [26, 27]. In this study, there was not a significant effect of anemia medications on HRQOL. Overall, anemia affects the level of HRQOL but perhaps the effect is not strong enough or acute enough to affect the trajectory of HRQOL over the concurrent time period.

In the same analysis, we were also able to examine the effect of covariates. As expected from previously published CKiD findings [23, 28], age, CKD duration, and height Z score had positive effects on HRQOL. However, no factors were associated with a change in the quality of life in the adjusted models. One possible explanation is the high HRQOLs have a ceiling effect; if HRQOL is already very high, a person has little opportunity to increase it. In other words, such a group will have more (relatively) negative change compared to a group that starts lower and thus has more opportunity for “catch-up” HRQOL trajectory. Another important covariate finding was that older patients who have a longer duration of CKD have a higher HRQOL. Our hypothesis is that these patients are more likely to become adjusted to their disease, thus improving their HRQOL scores.

There are many limitations to this study. Both hemoglobin and HRQOL may have considerable variability over a year. In some cases, the hemoglobin at the time of the visit might not represent the hemoglobin trend. Second, the PedsQL used to assess HRQOL relies on self-reports from parents and children. Even though it has been used in pediatric CKD, it is a general assessment of HRQOL and is not disease-specific. Additionally, unlike other CKiD studies, we defined anemia solely by hemoglobin level in order to better identify the effect of anemia regardless of medications. Medication usage by itself affects HRQOL and was used as a covariate. We defined anemia dichotomously rather than examining the continuous level and change in hemoglobin value; this alternative analysis could potentially be more powerful, but may not be as sensitive to detriments in HRQOL that would only be expected at the lowest percentiles of hemoglobin. Since we did not have the duration of anemia for patients at study onset, we were also unable to look at the length of anemia for the majority of patients. Another limitation of our study was the mean follow-up duration was 3.2 years. Changes in HRQOL might need a longer duration of follow-up. However, this time period did allow us to include an average of 4 study visits per participant. Another important factor is that there can be variable effects of depression and anxiety on HRQOL. A recent study by Johnson et al. [29] found that anemia and depression were not correlated in pediatric CKD. In our analysis, we did not include these psychiatric variables because they were not measured at the same frequency of visits of HRQOL which would thus complicate the analysis. Finally, due to the large sample size, we were able to find smaller differences that might not be as important clinically.

In summary, the adjusted analyses show anemia does affect HRQOL in a pediatric CKD population as prior reports in the literature have shown. However, anemia does not affect HRQOL trajectories over time. Compared to other studies, we focused on the impact of change in HRQOL. HRQOL differs for parents and children’s ratings, making it important for clinicians to address both parents and children during their visits. Further longitudinal studies on HRQOL are needed to better understand changes over time and identify ways to intervene and help patients.

Abbreviations

CKD

chronic kidney disease

CKiD

Chronic Kidney disease in Children Study

ESA

erythropoietin stimulating agents

GFR

glomerular filtration rate

HRQOL

health-related quality of life