Longitudinal Relationship Between Anemia and Statural Growth Impairment in Children and Adolescents With Nonglomerular CKD: Findings From the Chronic Kidney Disease in Children (CKiD) Study

Abstract

Rationale & objective: Anemia and statural growth impairment are both prevalent in children with nonglomerular chronic kidney disease (CKD) and are associated with poor quality of life and increased morbidity and mortality. However, to date no longitudinal studies have demonstrated a relationship between anemia and statural growth in this population.

Study design: The CKD in Children (CKiD) study is a multicenter prospective cohort study with over 15 years of follow-up observation.

Setting & participants: CKiD participants younger than 22 years with nonglomerular CKD who had not reached final adult height.

Exposure: Age-, sex-, and race-specific hemoglobin z score.

Outcome: Age- and sex-specific height z score.

Analytical approach: The relationship between hemoglobin and height was quantified using (1) multivariable repeated measures paired person-visit analysis, and (2) multivariable repeated measures linear mixed model analysis. Both models were adjusted for age, sex, body mass index, estimated glomerular filtration rate, acidosis, and medication use.

Results: Overall, 67% of the 510 participants studied had declining hemoglobin z score trajectories over the follow-up period, which included 1,763 person-visits. Compared with average hemoglobin z scores of≥0, average hemoglobin z scores of less than -1.0 were independently associated with significant growth impairment at the subsequent study visit, with height z score decline ranging from 0.24 to 0.35. Importantly, in 50% of cases hemoglobin z scores of less than -1.0 corresponded to hemoglobin values higher than those used as cutoffs defining anemia in the KDIGO clinical practice guideline for anemia in CKD. When stratified by age, the magnitude of the association peaked in participants aged 9 years. In line with paired-visit analyses, our mixed model analysis demonstrated that in participants with baseline hemoglobin z score less than -1.0, a hemoglobin z score decline over the follow-up period was associated with a statistically significant concurrent decrease in height z score.

Limitations: Limited ability to infer causality.

Conclusions: Hemoglobin decline is associated with growth impairment over time in children with mild to moderate nonglomerular CKD, even before hemoglobin levels reach the cutoffs that are currently used to define anemia in this population.

Keywords: Adolescent; anemia; children; chronic kidney disease (CKD); growth impairment; height z score; hemoglobin; pediatric CKD; short stature.

Figure 1.. Selection of study population for paired person-visit analysis and for mixed-effects model analyses.

Study population selection presented as number of participants (number of person-visit pairs) for boxes 1-7 and as number of participants (number of person-visits) for the last box 8, which shows the study population for the mixed model analysis.

Figure 2.. Results of paired person-visit analysis of the relationship between average hemoglobin z-score and subsequent height z-score.

Exposure was defined as categorical hemoglobin z-score averaged between visits i and i+1. Outcome was defined as height z-score at visit i+1. Visit pairs that consisted of visits occurring between 0.75 to 1.5 years apart were included. Black represents unadjusted estimates and corresponding 95% CIs, and gray represents adjusted estimates and corresponding 95% CIs. Unadjusted: 1763 observations; adjusted: 1644 observations (119 missing covariates) * indicates statistical significance.

Figure 3.. Paired person-visit analysis of the relationship between average hemoglobin z-score and subsequent height z-score stratified by age.

Exposure was defined as continuous hemoglobin z-score averaged between visits i and i+1. Outcome was defined as height z-score at visit i+1. Visit pairs that consisted of visits occurring between 0.75 to 1.5 years apart were included. Black represents unadjusted estimates and corresponding 95% CIs, and gray represents adjusted estimates and corresponding 95% CIs. Unadjusted: 1754 observations; adjusted: 1644 observations (110 missing covariates)