Risk factors in underweight older children with sickle cell anemia: a comparison of low- to high-income countries

Children with sickle cell anemia (SCA) are at increased risk of undernutrition from insufficient caloric intake along with higher energy requirements associated with increased protein turnover and synthesis, increased erythropoiesis, elevated myocardial expenditure, and a proinflammatory state.^1-3 Anthropometric values, including weight, height, and body mass index (BMI), indicate nutritional status.^4,5 BMI considers weight relative to height. Children with SCA have lower heights than an age- and sex-matched comparison group.^6-8 Thus, BMI may not optimally capture overall nutritional status, especially chronic malnutrition.⁹ However, weight is a biomarker for both acute and chronic malnutrition.¹⁰ In the general population, in children aged <5 years, a weight-for-age z score < −2 identifies all children with concurrent moderate wasting and stunting,^11,12 a group that is at a 12-times greater risk of mortality in the absence of treatment than children with normal anthropometric measurements.¹³ Weight as a screening measure alone also has the advantage over BMI because it requires only 1 quick measure that is reproducible without requiring specialized training or equipment associated with additional anthropometric measurements.¹⁴ 

The nutritional status of children is a potentially modifiable risk factor for reducing SCA mortality among children living in sub-Saharan Arica, where most children with SCA are born.¹⁵ We recently demonstrated that children aged 5 to 12 years with SCA who are underweight (weight-for-age z score < −1) are at increased risk of mortality in a low-income setting.¹⁶ Among 431 children aged 5 to 12 years with SCA who were followed up prospectively, weight-for-age z score was the sole determinant of death.¹⁶ This association between undernutrition and SCA mortality further supports the observation that underweight children with SCA are at increased risk of hospitalization.¹⁷ However, the risk factors for being underweight, irrespective of living in a low-income vs a high-income setting, remain unexplored.^4,5 We completed a pooled secondary analysis of 2 stroke prevention trials conducted in the low-income setting of northern Nigeria and the high-income countries of North America and Europe to test the hypothesis that there are common biologically associated risks of being underweight (weight-for-age z score < −1) in children aged 5 to 12 years old with SCA.

We conducted a cross-sectional study as a secondary data analysis of deidentified data sets from the baseline data of children with SCA (homozygous hemoglobin S or hemoglobin Sβ⁰ thalassemia) who were screened for 2 previous trials, the Primary Prevention of Stroke in Children with Sickle Cell Disease in Nigeria (SPRING) trial (#NCT02560935) and the Silent Cerebral Infarct (SIT) multicenter clinical trial (#NCT00072761). The legal guardians of all prospective trial participants provided signed informed consent for SPRING or SIT trial screening procedures.

The SPRING trial was a National Institute of Health–funded multicenter phase 3 randomized controlled trial conducted from July 2016 to April 2020 at a low-income region of northern Nigeria at Aminu Kano Teaching Hospital and Murtala Mohammad Specialist Hospital, with referrals from Hasiya Bayero Pediatric Hospital and Muhammad Abdullahi Wase Teaching Hospital, all at Kano, Nigeria, and Barau Dikko Teaching Hospital, Kaduna, Nigeria.^18,19 The institutional review board of Vanderbilt University Medical Center, Nashville, TN, and the respective ethics committees of the local Nigerian participating sites approved the SPRING trial.

The SIT trial was an National Institute of Health–funded multicenter phase 3 randomized, 2-arm controlled clinical trial from December 2004 to November 2013 in 29 clinical centers in the United States, Canada, France, and the United Kingdom.²⁰ Institution-specific institutional review board approval was obtained at each participating institution.

Children aged 5 to 12 years with SCA (hemoglobin SS or hemoglobin Sβ⁰ thalassemia) screened for the SPRING and SIT trials were included in this secondary analysis. Participants aged ≥13 years in the SIT cohort were excluded for comparability between cohorts. Given that the primary outcome was underweight status, children with incomplete data on height, weight, or age were excluded from the study.

We used the World Bank classifications of countries' income levels during the respective studies. Nigeria is a lower-middle–income country.²¹ However, Kano, northern Nigeria, is considered a low-income region. All study sites for SIT were located in high-income countries.

Demographic and baseline laboratory values were collected. Study nurses or physicians performed anthropometric measurements, including height (cm) and weight (kg) according to standard protocol²² as part of the initial research visit. BMI (kg/m²) was calculated from height and weight measurements. We converted anthropometric measurements to age- and sex-specific z scores based on the World Health Organization growth reference.^4,5 To calculate weight-for-age z scores for children aged >10 years, we used the Canadian Pediatric Endocrine Group growth charts. The Canadian Pediatric Endocrine Group growth charts extend the weight-for-age z scores using the same core data set as the World Health Organization reference for school-aged children and adolescents.²³ 

Underweight, stunting, and wasting were classified based on weight-for-age, height-for-age, and BMI z scores, respectively.^4,5 Degrees of undernutrition were delineated into z scores between ≤ −2 and < −1, ≤ −3 and < −2, and < −3 as mild, moderate, and severe, respectively. Our primary analysis focuses on underweight, defined as weight-for-age z score < −1, given the association with mortality in children aged <5 years²⁴ and children aged 5 to 12 years with SCA.¹⁶ 

Summary statistics for continuous variables were summarized as means and standard deviations or as medians and interquartile ranges for variables not normally distributed. Categorical variables and prevalence were reported as numbers and percentages. A χ² test was used for percentages, a t test for means, and a Mann-Whitney U test for medians. We used multivariable linear and logistic regression to assess several biological factors (age, sex, hemoglobin, and white blood cell count) likely associated with weight-for-age z score and underweight status. Odds ratios (ORs) with a 95% confidence interval (CI) were used to characterize risk factors associated with being underweight. An interaction term was added for hemoglobin level with cohort based on initial model results. A 2-sided P value < .05 was considered significant. Data analysis was performed using SPSS 28.0 (IBM, Armonk, NY) and Stata 15 (StataCorp, College Station, TX).

Of the children with SCA screened for the SPRING and SIT trials, 97% (928 of 934 total participants) and 98% (1093 of 1119 total participants), respectively, had complete anthropometric values, and age and sex data to calculate anthropometric z scores, and were included in the analysis (Table 1). The median age of children with SCA in the SIT trial (8.5 years) was higher than that of the SPRING cohort (8.1 years; P < .001). In both cohorts, approximately half of the children were male (SPRING: 49.0%; SIT: 50.8%; P = .434). The mean hemoglobin level was lower in the SPRING cohort (7.5 g/dL) than in the SIT cohort (8.1 g/dL; P < .001), whereas the mean white blood cell count was higher in the SPRING cohort (14.7 × 10⁹/L vs 12.4 × 10⁹/L, respectively; P < .001).

All measures of nutritional status were significantly lower in the SPRING cohort than in the SIT cohort. (Tables 1 and 2). The mean weight-for-age z scores were negative for both the SPRING (−2.21) and SIT cohorts (−0.33; P < .001). In the SPRING cohort, 87.9% of participants were underweight, with 22.7% classified as severely underweight (weight-for-age z score < −3). In comparison, in the SIT cohort, 25.7% of participants were underweight, with 0.7% classified as severely underweight.

In the combined cohorts, 27.2% were underweight, with stunting and wasting, representing 49.4% and 8.3% of the SPRING and SIT cohorts, respectively. All participants with wasting and stunting were also underweight. Conversely, except for 8 participants (0.4%), all participants who were underweight were either stunted, wasted, or both (Table 3).

In both the SPRING and SIT cohorts, the median age of children classified as underweight was older than those children who were not underweight (SPRING: 8.3 vs 6.7 years, P < .001; SIT: 9.5 vs 8.1 years, P < .001). Compared with children who were not underweight, children who were underweight had higher white blood cell counts and lower hemoglobin levels in the combined cohort (Table 4). The association between lower hemoglobin and being underweight remained significant in the SIT cohort (P < .001). However, there was no significant difference in hemoglobin levels or white blood cell count in the SPRING trial between participants who were underweight and those who were not underweight (P = .738 and P = .140, respectively; Table 4).

Predetermined plausible biological factors postulated to be associated with being underweight (weight-for-age z score < −1) were analyzed in a multivariable logistic regression in the combined cohorts. The model also included an interaction term between hemoglobin level and cohort. Among all participants, for every additional year of age, the odds increased of being underweight (OR, 1.24; 95% CI, 1.17-1.31; P < .001; Table 5; Figure 1). Increasing hemoglobin level was associated with decreased odds of being underweight in each cohort (combined cohort: OR, 0.67; 95% CI, 0.6-0.75; P < .001; Table 5). There was an interaction between hemoglobin level and cohort (P = .007), in which the association between the hemoglobin level and being underweight was dependent on the cohort, with an increase in the hemoglobin level having a greater effect for lower hemoglobin levels in the SIT cohort but at higher hemoglobin levels for the SPRING cohort (Table 5; Figure 2). The participants in the SPRING cohort were more likely to be underweight at all hemoglobin levels (Figure 2).

We then constructed a multivariable logistic regression with the same covariates except group, to assess the model in each cohort separately (the interaction term was not required). Older age (SPRING: OR, 1.38; 95% CI, 1.24-1.53; P < .001; SIT: OR, 1.19; 95% CI, 1.11-1.27; P < .001) and lower hemoglobin level (SPRING: OR, 0.82; 95% CI, 0.68-0.99; P = .042; SIT: OR, 0.59; 95% CI, 0.51-0.68; P. <.001) remained associated with being underweight (weight-for-age z score ≥ −1; supplemental Table 1).

Multivariable linear regression was used to assess the association of age, sex, hemoglobin, and white blood cell count with weight-for-age z score at baseline. Age (β = −0.14; P < .001), male sex (β = 0.12; P = .008), and hemoglobin level (β = 0.22; P < .001) were associated with weight-for-age z score for the combined cohort (supplemental Table 2). Age remained significant in both the SPRING and SIT cohorts (SPRING: β = −0.17, P < .001; SIT: β = −0.11, P < .001; supplemental Table 2). Hemoglobin level also remained significant in both cohorts (SPRING: β = 0.12, P < .001; SIT: β = 0.23, P < .001; supplemental Table 2). However, male sex was associated with weight-for-age z score in the SIT cohort (β = 0.24, P < .001; supplemental Table 2).

Despite the association of being underweight with increased mortality¹⁶ and hospitalizations,¹⁷ no current studies, to our knowledge, discuss the risk factors for being underweight in children aged >5 years with SCA in both low- and high-income settings. We demonstrate that older age and lower hemoglobin level in children aged 5 to 12 years with SCA in both low- and high-income countries are risk factors for being underweight (weight-for-age z score < −1). As anticipated, the z scores of all anthropometric indices were notably lower among children in the low-income cohort. The presence of common risk factors for undernutrition in both settings underscores the potential for a common biological association for undernutrition in SCA.

Anthropometrical indices to classify children as underweight are proxies for the physiological and functional consequences of the underlying processes of undernutrition and weight loss or the slow rate of weight gain, reflecting a range of possible causative factors, both biological (at child level) and socioeconomic (household and society levels).¹⁰ Children with SCA have higher resting energy expenditure than children without SCA, suggesting that the need for dietary intake of energy and protein for children with SCA is somewhat increased compared with that for children without SCA.^25-27 Children in the low-income cohort were more likely to be underweight, with the dietary and feeding practices remaining suboptimal for these children.⁶ There are socioenvironmental factors contributing to the increased malnutrition in the low-income setting, in addition to the biological factors we have investigated. In our low-income cohort, a high percentage of participants were underweight, stunted, and wasted (49.4%). Similar to children aged <5 years without hemoglobinopathies,²⁸ this group may be at higher risk of death.

Consistent with prior studies, children with lower hemoglobin levels were more likely to be underweight in both cohorts.^6,29 However, these previous studies were in younger children aged <5 years ⁶ or with small sample sizes (n = 100),²⁹ and not in both low- and high-income settings.^6,29 In a secondary analysis of the Dissemination and Implementation of Stroke Prevention Looking at the Care Environment cohort (3305 children with SCA and at least 2 anthropometric assessments during study follow-up), low hemoglobin level was associated with low BMI z scores.³⁰ However, the association with weight-for-age z scores was not addressed, and all study centers were located in the United States, a high-income setting.³⁰ Chronic anemia is associated with SCA and decreased growth; long-term red blood cell transfusion therapy designed to increase the baseline hemoglobin levels have been noted to increase weight, height, and BMI z scores for children with SCA.^30,31 Some studies suggest that hydroxyurea has no detrimental effects on growth and that it can potentially improve growth in children with SCA.^30,32-35 However, to our knowledge, no trial has tested whether hydroxyurea improves growth, potentially through increasing hemoglobin levels,³⁶ as a primary hypothesis.

Older age was associated with an increased risk of being underweight in both cohorts. The majority of undernutrition literature focuses on children aged <5 years.^37,38 In a longitudinal study by Zemel et al, in children from birth to aged 18 years, age at enrollment was negatively associated with weight-for-age z scores, and from study entry to the last visit, 59% of children with SCA declined in weight-for-age z score.³⁹ Another study in children aged 6 months to 15 years demonstrated weight-for-age z score reduction across all age groups.⁴⁰ Similarly, in Tanzania, in a cohort of patients with SCA with ages ranging from 0.5 to 48 years, the most significant growth deficits were observed during adolescence.¹⁷ Our findings and those of others suggest that growth faltering is associated with increasing age in children with SCA. Growth faltering with age may represent a continued insult from SCA or reflect the effects of increased resting energy expenditure and suboptimal food intake perpetuated by or superimposed on SCA.

As expected, our study has limitations as a secondary analysis of participants screened for 2 large randomized clinical trials in children with SCA. Because of the cross-sectional design, we cannot infer that the relationship between the biological factors is causal. Weight-for-age z score as a continuous variable was also significantly related to lower hemoglobin levels and older age among both individual and combined cohorts, consistent with a dose response. Lower hemoglobin levels may result from the same pathophysiology that causes poor nutritional status, and longitudinal studies should evaluate the change in hemoglobin and weight status over time. Dietary intake, maternal characteristics, access to health care, and household food security should be explored in future work on nutritional status in children with SCA. However, the strength of our study was the ability to combine research-quality anthropometric measures in 2 large populations of children with SCA that differed in location, low-income vs high-income settings.

The consistency of risk factors in low- and high-income settings provides substantial evidence for disease-specific risk factors associated with being underweight. Our findings from this study contribute to our understanding of the factors associated with underweight status in children aged 5 to 12 years with SCA in both low- and high-income settings. A low hemoglobin level is a potentially modifiable risk factor for preventing underweight status in older children with SCA in both low- and high-income settings.

The authors thank Bilya Sani Musa, Mustafa Nateqi, Khadijah Bulama, Murtala Umar, Charity Dooshima Agba-Dooyum, Fahad Usman, Abdulrasheed Sani, Awwal Gambo, Jamila Ibrahim, Gloria Bahago, Jamil Galadanci, Leshana Saint Jean, and Jennifer C. Beck for providing the successful conduct of the research reported in this publication.

The SPRING trial was supported by National Institutes of Health (NIH) grant R01NS094041, NIH/National Center for Advancing Translational Sciences grant UL1 TR000445, and the generous donation of the Phillips family. The SIT trial was supported by grants from the National Institute of Neurological Disorders and Stroke (5U01NS042804 and 3U01NS042804 [American Recovery Reinvestment Act supplementary grant]) (M.R.D.). L.J.K. was supported by the Fogarty International Center; the National Heart, Lung, and Blood Institute of the NIH (D43 TW009337), and the National Institute of Diabetes and Digestive and Kidney Diseases of the NIH (T32DK007673). The sponsors did not have any role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. The findings and conclusions in this paper are those of the authors and do not represent the official position of the NIH.

Contribution: L.J.K., M.R., and M.R.D. conceptualized and designed the secondary analysis, had full access to all the data in the study, took responsibility for data integrity and data analyses accuracy, and wrote the initial manuscript; S.U.A. and S.G. conducted the SPRING trial; M.R.D. served as the principal investigator of the SIT trial; L.J.K. and M.R. performed the analyses; M.R.D. and L.J.K. interpreted the results; and all authors reviewed the manuscript and approved its submission.

Conflict-of-interest disclosure: M.R.D. and his institution sponsor 2 externally funded research investigator-initiated projects; Global Blood Therapeutics (GBT) will provide funding for the cost of the clinical studies. GBT was not a cosponsor of either study; did not receive any compensation for the conduct of these 2 investigator-initiated observational studies. M.R.D. is a member of the GBT advisory board for a proposed randomized controlled trial, for which he receives compensation; serves on the steering committee for a Novartis-sponsored phase 2 trial to prevent priapism in men; was a medical adviser in developing the CTX001 Early Economic Model; provided medical input on the economic model as part of an expert reference group for the Vertex/CRISPR CTX001 Early Economic Model in 2020; and consulted for the Formal Pharmaceutical company on sickle cell disease in 2021 and 2022. The remaining authors declare no competing financial interests.

Correspondence: Lauren Jane Klein, D. Brent Polk Division of Pediatric Gastroenterology, Hepatology, and Nutrition, Department of Pediatrics, Monroe Carell Jr. Children's Hospital at Vanderbilt and Vanderbilt Institute for Global Health, 2525 West End Ave, Suite 750, Nashville, TN 37203; e-mail: lauren.klein@vumc.org.