Browse the corpus

Malaria is a cause of iron deficiency in African children. Malaria and iron deficiency (ID) are common and interrelated public health problems in African children. Observational data suggest that interrupting malaria transmission reduces the prevalence of ID1. To test the hypothesis that malaria might cause ID, we used sickle cell trait (HbAS, rs334 ), a genetic variant that confers specific protection against malaria2, as an instrumental variable in Mendelian randomization analyses. HbAS was associated with a 30% reduction in ID among children living in malaria-endemic countries in Africa (n = 7,453), but not among individuals living in malaria-free areas (n = 3,818). Genetically predicted malaria risk was associated with an odds ratio of 2.65 for ID per unit increase in the log incidence rate of malaria. This suggests that an intervention that halves the risk of malaria episodes would reduce the prevalence of ID in African children by 49%.

This study included ten cohorts of healthy children (in Malawi, Ghana, Burkina Faso, DRC, Kenya (Kilifi and Western), Tanzania, The Gambia, Cameroon and Uganda) living in malaria-endemic countries. Four of these studies (in Malawi, Ghana, DRC and The Gambia) were cross-sectional studies, two were longitudinal (in Tanzania and Kilifi Kenya), two were RCT (in Burkina Faso and Uganda), one was a cluster RCT (in Western Kenya), and one was a cluster survey (in Cameroon). The study also included three cohorts living in countries with no malaria exposure, African American adults from the longitudinal JHS, children from Soweto, South Africa and a survey of school children in Nairobi, Kenya. These studies are described below. The 2015–2016 Malawi Micronutrient Survey (MMS) was conducted as part of the Malawi Demographic and Health Survey. This cross-sectional survey aimed to determine the prevalence of anemia, micronutrient deficiencies (iron and vitamin A), infections and hemoglobinopathies32. The MMS included all children aged 6–59 months from randomly selected clusters and households. Details of the study design are available elsewhere32. Briefly, whole blood collected in tubes containing EDTA was used to test for malaria using a rapid diagnostic test (RDT) (SD Bioline Malaria P. falciparum (HRP2), Alere) and to measure hemoglobin concentrations using the HemoCue 301 system (HemoCue America). Levels of serum ferritin, CRP and AGP were measured using sandwich ELISA (VitMin Laboratory)33. Genotyping of sickle cell trait and α-thalassemia was performed using PCR as previously described32.

laria P. falciparum (HRP2), Alere) and to measure hemoglobin concentrations using the HemoCue 301 system (HemoCue America). Levels of serum ferritin, CRP and AGP were measured using sandwich ELISA (VitMin Laboratory)33. Genotyping of sickle cell trait and α-thalassemia was performed using PCR as previously described32. This study was part of the 2017 Ghana Micronutrient Survey. Details of the cross-sectional study design and ethical approvals are presented in ref.34. Children aged 6–59 months were recruited from three strata (Southern Belt, Middle Belt and Northern Belt) in Ghana, and random selection was performed in each stratum. Blood sampling and anthropometry were conducted during the survey. Malaria testing was performed using RDT (SD Bioline Malaria Ag Pf/Pan RDT kit (Standard Diagnostics); hemoglobin concentrations were measured using HemoCue 301 AB). Sandwich ELISA33 was used to measure serum ferritin, CRP and AGP concentrations. DNA was extracted from blood pellets and used to type sickle cell trait and α-thalassemia using PCR35,36.

sing RDT (SD Bioline Malaria Ag Pf/Pan RDT kit (Standard Diagnostics); hemoglobin concentrations were measured using HemoCue 301 AB). Sandwich ELISA33 was used to measure serum ferritin, CRP and AGP concentrations. DNA was extracted from blood pellets and used to type sickle cell trait and α-thalassemia using PCR35,36. This was part of the VaccGene study, which aimed to identify genetic variants associated with differential response to vaccination in infancy but with ethical approval to undertake analyses to examine the effect of iron status on infection susceptibility. Details of the study design are described elsewhere37. Infants between the ages of 6 and 18 months living in the Banfora region of Burkina Faso were recruited into a phase 1–2b clinical trial to test the safety, immunogenicity and efficacy of an experimental heterologous viral-vectored prime-boost liver-stage malaria vaccine37. Serum ferritin (Chemiluminescent Microparticle Immunoassay, Abbott Architect), hepcidin (DRG Hepcidin 25 (bioactive) High Sensitive ELISA kit (DRG International)), CRP (Multigent CRP Vario assay, Abbott Architect), hemoglobin (Coulter analyzer, Beckman Coulter) and malaria parasitemia (Giemsa-stained thick and thin blood films) were measured at a single time point. Genotyping of sickle cell trait in the VaccGene study is described below.

Sensitive ELISA kit (DRG International)), CRP (Multigent CRP Vario assay, Abbott Architect), hemoglobin (Coulter analyzer, Beckman Coulter) and malaria parasitemia (Giemsa-stained thick and thin blood films) were measured at a single time point. Genotyping of sickle cell trait in the VaccGene study is described below. Children were recruited from rural villages in Bungoma, Kakamega and Vihiga counties in western Kenya, using a cluster study design as part of the water, sanitation and handwashing benefits RCT (WASH Benefits trial). Details of the study design were published elsewhere38. Samples collected from the environmental enteropathy endline survey were used. Venous blood samples were used to test for malaria parasitemia using RDT (SD Bioline Malaria P. falciparum (HRP2), Alere) and hemoglobin concentrations (HemoCue Hb 301). Serum ferritin and CRP levels were assayed using sandwich ELISA33. Serum hepcidin-25 levels were quantified by using a competitive ELISA kit (PenLabs). Genotyping of sickle hemoglobin types and α-thalassemia was conducted using PCR35,36.

laria P. falciparum (HRP2), Alere) and hemoglobin concentrations (HemoCue Hb 301). Serum ferritin and CRP levels were assayed using sandwich ELISA33. Serum hepcidin-25 levels were quantified by using a competitive ELISA kit (PenLabs). Genotyping of sickle hemoglobin types and α-thalassemia was conducted using PCR35,36. This study used data collected during a nutrition cross-sectional survey of mothers and their children aged 6–59 months in rural Sud Kivu and Kongo Central provinces in the DRC as described elsewhere39. Venous blood samples were used to test for malaria parasitemia using RDT (CareStart Malaria Screen, Access Bio). Serum ferritin, CRP and AGP levels were assayed using sandwich ELISA33. Hemoglobin typing for sickle cell trait was conducted using pyrosequencing, while PCR was used to detect α-thalassemia as described elsewhere39.

ood samples were used to test for malaria parasitemia using RDT (CareStart Malaria Screen, Access Bio). Serum ferritin, CRP and AGP levels were assayed using sandwich ELISA33. Hemoglobin typing for sickle cell trait was conducted using pyrosequencing, while PCR was used to detect α-thalassemia as described elsewhere39. This was an ongoing rolling longitudinal study designed to evaluate immunity to malaria in children and is described elsewhere40. Within this cohort, children were followed to 8 years of age with weekly follow-ups and annual cross-sectional surveys during which anthropometry measurements were made and blood samples were collected. Serum ferritin (Chemiluminescent Microparticle Immunoassay, Abbott Architect), hepcidin (DRG Hepcidin 25 (bioactive) High Sensitive ELISA kit (DRG International)), CRP (Multigent CRP Vario assay, Abbott Architect), hemoglobin (Coulter analyzer, Beckman Coulter) and malaria parasitemia (Giemsa-stained thick and thin blood films) were measured from blood samples collected at a single cross-sectional survey based on the availability of plasma samples archived at 80°C. Genotyping of hemoglobin types and α-thalassemia was conducted by PCR35,36 using DNA extracted with the Qiagen DNA Blood Mini kit (Qiagen).

emsa-stained thick and thin blood films) were measured from blood samples collected at a single cross-sectional survey based on the availability of plasma samples archived at 80°C. Genotyping of hemoglobin types and α-thalassemia was conducted by PCR35,36 using DNA extracted with the Qiagen DNA Blood Mini kit (Qiagen). Children were enrolled at delivery into the MOMS Project longitudinal birth cohort at Muheza District Hospital in northeastern Tanzania between 2002 and 2006. Children were assessed for malaria parasitemia every 2 weeks during infancy and monthly thereafter, as well as at the time of any illness. Singleton children without evidence of human immunodeficiency virus in themselves or in their mothers during follow-up were included in studies of iron status and risk of malaria infection as described elsewhere41. Blood samples were collected at 3, 6 and 12 months of age and then once every 6 months in years 2 and 341. In this study, we used data from samples collected at a single time point, each child’s oldest time point, as older children were likely to have experienced more malaria episodes. The presence of P. falciparum parasitemia was determined using Giemsa-stained thick blood smears, while hemoglobin levels were measured using an impedance-based analyzer (Abbott Cell Dyn 1200). Plasma ferritin and CRP levels were assayed using a multiplex bead-based platform (Bio-Rad) and custom assay kits, and hemoglobin was typed by electrophoresis (Helena Laboratories)41. Genotyping for α-thalassemia was conducted as described by Chong et al.36.

red using an impedance-based analyzer (Abbott Cell Dyn 1200). Plasma ferritin and CRP levels were assayed using a multiplex bead-based platform (Bio-Rad) and custom assay kits, and hemoglobin was typed by electrophoresis (Helena Laboratories)41. Genotyping for α-thalassemia was conducted as described by Chong et al.36. All children aged 2–6 years old were recruited from ten rural villages in the West Kiang region of The Gambia during the malaria season (July to August 2001)42. We used cross-sectional data collected at the start of the malaria season. All children had a clinical examination, anthropometric measurements and a 3-d course of mebendazole for possible hookworm infection. A blood sample was collected for complete blood count, observations with a malaria slide, measurements of ferritin (Microparticle Enzyme Immunoassay (Abbott Architect)), hepcidin (Hepcidin-25 (human) EIA Kit (Bachem) and ACT (immunoturbidimetry, Cobas Mira Plus Bio-analyzer, Roche) levels and DNA extraction. Children with a temperature >37.5°C had a malaria blood film, appropriate clinical treatment and a blood sample 2 weeks later after recovery from illness. Genotyping of sickle cell was performed on amplified DNA as detailed elsewhere42. The Gambian Bachem hepcidin values were harmonized by converting to the old DRG hepcidin assay values ((0.266×Bachem values)+1.633) and then to the new High Sensitive DRG hepcidin assay values ((1.989× old DRG values)-3.24) as previously validated43.

f sickle cell was performed on amplified DNA as detailed elsewhere42. The Gambian Bachem hepcidin values were harmonized by converting to the old DRG hepcidin assay values ((0.266×Bachem values)+1.633) and then to the new High Sensitive DRG hepcidin assay values ((1.989× old DRG values)-3.24) as previously validated43. Children 12–59 months of age were recruited to a cluster survey that aimed to determine the prevalence of inherited hemoglobin disorders in Yaoundé and Douala, Cameroon, as described elsewhere44. Venous blood samples were collected for malaria testing using RDT (SD Bioline Malaria Ag Pf/Pan, Standard Diagnostics) and hemoglobin measurement using a photometer (HemoCue). Plasma ferritin, CRP and AGP levels were assayed by ELISA33. Hemoglobin genotypes were determined by HPLC using an ultra-Resolution Variants Analyzer (Trinity Biotech), while α-thalassemia type was determined by PCR44.

ine Malaria Ag Pf/Pan, Standard Diagnostics) and hemoglobin measurement using a photometer (HemoCue). Plasma ferritin, CRP and AGP levels were assayed by ELISA33. Hemoglobin genotypes were determined by HPLC using an ultra-Resolution Variants Analyzer (Trinity Biotech), while α-thalassemia type was determined by PCR44. EMaBS is a prospective birth cohort that was originally designed as an RCT to test whether anthelminthic treatment during pregnancy and early childhood was associated with differential response to vaccination or incidence of infections, such as pneumonia, diarrhea or malaria (http://emabs.lshtm.ac.uk/)45. This cohort was part of the VaccGene study as described elsewhere46. Blood samples were collected in Vacutainer tubes containing EDTA at birth and at subsequent birthdays up to 5 years of age. Anthropometry and biomarkers of iron and inflammation were measured in samples from a single annual visit based on the availability of stored samples. Serum ferritin (Chemiluminescent Microparticle Immunoassay, Abbott Architect), CRP (Multigent CRP Vario assay, Abbott Architect), hepcidin (DRG Hepcidin-25 (bioactive) High Sensitive ELISA kit (DRG International)), hemoglobin (Coulter analyzer, Beckman Coulter) and malaria parasitemia (Giemsa-stained thick and thin blood films) were measured. Genotyping of sickle cell trait in the VaccGene study is described below.

This study included ten cohorts of healthy children (in Malawi, Ghana, Burkina Faso, DRC, Kenya (Kilifi and Western), Tanzania, The Gambia, Cameroon and Uganda) living in malaria-endemic countries. Four of these studies (in Malawi, Ghana, DRC and The Gambia) were cross-sectional studies, two were longitudinal (in Tanzania and Kilifi Kenya), two were RCT (in Burkina Faso and Uganda), one was a cluster RCT (in Western Kenya), and one was a cluster survey (in Cameroon). The study also included three cohorts living in countries with no malaria exposure, African American adults from the longitudinal JHS, children from Soweto, South Africa and a survey of school children in Nairobi, Kenya. These studies are described below.

The 2015–2016 Malawi Micronutrient Survey (MMS) was conducted as part of the Malawi Demographic and Health Survey. This cross-sectional survey aimed to determine the prevalence of anemia, micronutrient deficiencies (iron and vitamin A), infections and hemoglobinopathies32. The MMS included all children aged 6–59 months from randomly selected clusters and households. Details of the study design are available elsewhere32. Briefly, whole blood collected in tubes containing EDTA was used to test for malaria using a rapid diagnostic test (RDT) (SD Bioline Malaria P. falciparum (HRP2), Alere) and to measure hemoglobin concentrations using the HemoCue 301 system (HemoCue America). Levels of serum ferritin, CRP and AGP were measured using sandwich ELISA (VitMin Laboratory)33. Genotyping of sickle cell trait and α-thalassemia was performed using PCR as previously described32.

This study was part of the 2017 Ghana Micronutrient Survey. Details of the cross-sectional study design and ethical approvals are presented in ref.34. Children aged 6–59 months were recruited from three strata (Southern Belt, Middle Belt and Northern Belt) in Ghana, and random selection was performed in each stratum. Blood sampling and anthropometry were conducted during the survey. Malaria testing was performed using RDT (SD Bioline Malaria Ag Pf/Pan RDT kit (Standard Diagnostics); hemoglobin concentrations were measured using HemoCue 301 AB). Sandwich ELISA33 was used to measure serum ferritin, CRP and AGP concentrations. DNA was extracted from blood pellets and used to type sickle cell trait and α-thalassemia using PCR35,36.

This was part of the VaccGene study, which aimed to identify genetic variants associated with differential response to vaccination in infancy but with ethical approval to undertake analyses to examine the effect of iron status on infection susceptibility. Details of the study design are described elsewhere37. Infants between the ages of 6 and 18 months living in the Banfora region of Burkina Faso were recruited into a phase 1–2b clinical trial to test the safety, immunogenicity and efficacy of an experimental heterologous viral-vectored prime-boost liver-stage malaria vaccine37. Serum ferritin (Chemiluminescent Microparticle Immunoassay, Abbott Architect), hepcidin (DRG Hepcidin 25 (bioactive) High Sensitive ELISA kit (DRG International)), CRP (Multigent CRP Vario assay, Abbott Architect), hemoglobin (Coulter analyzer, Beckman Coulter) and malaria parasitemia (Giemsa-stained thick and thin blood films) were measured at a single time point. Genotyping of sickle cell trait in the VaccGene study is described below.

Children were recruited from rural villages in Bungoma, Kakamega and Vihiga counties in western Kenya, using a cluster study design as part of the water, sanitation and handwashing benefits RCT (WASH Benefits trial). Details of the study design were published elsewhere38. Samples collected from the environmental enteropathy endline survey were used. Venous blood samples were used to test for malaria parasitemia using RDT (SD Bioline Malaria P. falciparum (HRP2), Alere) and hemoglobin concentrations (HemoCue Hb 301). Serum ferritin and CRP levels were assayed using sandwich ELISA33. Serum hepcidin-25 levels were quantified by using a competitive ELISA kit (PenLabs). Genotyping of sickle hemoglobin types and α-thalassemia was conducted using PCR35,36.

This study used data collected during a nutrition cross-sectional survey of mothers and their children aged 6–59 months in rural Sud Kivu and Kongo Central provinces in the DRC as described elsewhere39. Venous blood samples were used to test for malaria parasitemia using RDT (CareStart Malaria Screen, Access Bio). Serum ferritin, CRP and AGP levels were assayed using sandwich ELISA33. Hemoglobin typing for sickle cell trait was conducted using pyrosequencing, while PCR was used to detect α-thalassemia as described elsewhere39.

This was an ongoing rolling longitudinal study designed to evaluate immunity to malaria in children and is described elsewhere40. Within this cohort, children were followed to 8 years of age with weekly follow-ups and annual cross-sectional surveys during which anthropometry measurements were made and blood samples were collected. Serum ferritin (Chemiluminescent Microparticle Immunoassay, Abbott Architect), hepcidin (DRG Hepcidin 25 (bioactive) High Sensitive ELISA kit (DRG International)), CRP (Multigent CRP Vario assay, Abbott Architect), hemoglobin (Coulter analyzer, Beckman Coulter) and malaria parasitemia (Giemsa-stained thick and thin blood films) were measured from blood samples collected at a single cross-sectional survey based on the availability of plasma samples archived at 80°C. Genotyping of hemoglobin types and α-thalassemia was conducted by PCR35,36 using DNA extracted with the Qiagen DNA Blood Mini kit (Qiagen).

Children were enrolled at delivery into the MOMS Project longitudinal birth cohort at Muheza District Hospital in northeastern Tanzania between 2002 and 2006. Children were assessed for malaria parasitemia every 2 weeks during infancy and monthly thereafter, as well as at the time of any illness. Singleton children without evidence of human immunodeficiency virus in themselves or in their mothers during follow-up were included in studies of iron status and risk of malaria infection as described elsewhere41. Blood samples were collected at 3, 6 and 12 months of age and then once every 6 months in years 2 and 341. In this study, we used data from samples collected at a single time point, each child’s oldest time point, as older children were likely to have experienced more malaria episodes. The presence of P. falciparum parasitemia was determined using Giemsa-stained thick blood smears, while hemoglobin levels were measured using an impedance-based analyzer (Abbott Cell Dyn 1200). Plasma ferritin and CRP levels were assayed using a multiplex bead-based platform (Bio-Rad) and custom assay kits, and hemoglobin was typed by electrophoresis (Helena Laboratories)41. Genotyping for α-thalassemia was conducted as described by Chong et al.36.

All children aged 2–6 years old were recruited from ten rural villages in the West Kiang region of The Gambia during the malaria season (July to August 2001)42. We used cross-sectional data collected at the start of the malaria season. All children had a clinical examination, anthropometric measurements and a 3-d course of mebendazole for possible hookworm infection. A blood sample was collected for complete blood count, observations with a malaria slide, measurements of ferritin (Microparticle Enzyme Immunoassay (Abbott Architect)), hepcidin (Hepcidin-25 (human) EIA Kit (Bachem) and ACT (immunoturbidimetry, Cobas Mira Plus Bio-analyzer, Roche) levels and DNA extraction. Children with a temperature >37.5°C had a malaria blood film, appropriate clinical treatment and a blood sample 2 weeks later after recovery from illness. Genotyping of sickle cell was performed on amplified DNA as detailed elsewhere42. The Gambian Bachem hepcidin values were harmonized by converting to the old DRG hepcidin assay values ((0.266×Bachem values)+1.633) and then to the new High Sensitive DRG hepcidin assay values ((1.989× old DRG values)-3.24) as previously validated43.

Children 12–59 months of age were recruited to a cluster survey that aimed to determine the prevalence of inherited hemoglobin disorders in Yaoundé and Douala, Cameroon, as described elsewhere44. Venous blood samples were collected for malaria testing using RDT (SD Bioline Malaria Ag Pf/Pan, Standard Diagnostics) and hemoglobin measurement using a photometer (HemoCue). Plasma ferritin, CRP and AGP levels were assayed by ELISA33. Hemoglobin genotypes were determined by HPLC using an ultra-Resolution Variants Analyzer (Trinity Biotech), while α-thalassemia type was determined by PCR44.

EMaBS is a prospective birth cohort that was originally designed as an RCT to test whether anthelminthic treatment during pregnancy and early childhood was associated with differential response to vaccination or incidence of infections, such as pneumonia, diarrhea or malaria (http://emabs.lshtm.ac.uk/)45. This cohort was part of the VaccGene study as described elsewhere46. Blood samples were collected in Vacutainer tubes containing EDTA at birth and at subsequent birthdays up to 5 years of age. Anthropometry and biomarkers of iron and inflammation were measured in samples from a single annual visit based on the availability of stored samples. Serum ferritin (Chemiluminescent Microparticle Immunoassay, Abbott Architect), CRP (Multigent CRP Vario assay, Abbott Architect), hepcidin (DRG Hepcidin-25 (bioactive) High Sensitive ELISA kit (DRG International)), hemoglobin (Coulter analyzer, Beckman Coulter) and malaria parasitemia (Giemsa-stained thick and thin blood films) were measured. Genotyping of sickle cell trait in the VaccGene study is described below.

Infants born in Chris Hani Baragwanath Hospital living in the Soweto region of Johannesburg, South Africa were recruited from vaccine trials47 coordinated by the Respiratory and Meningeal Pathogens Unit (http://www.rmpru.com/). Mothers of the infants were approached if the infants had received all of their EPI vaccines up to 6 months of age. The infants were sampled prospectively at 12 months after receipt of the measles vaccine at 9 months of age. Single whole-blood samples were collected in vacutainer tubes containing EDTA for measurement of iron and inflammatory markers and DNA extraction. Serum ferritin (Chemiluminescent Microparticle Immunoassay, Abbott Architect), CRP (Multigent CRP Vario assay, Abbott Architect) and hepcidin (DRG Hepcidin-25 (bioactive) High Sensitive ELISA kit (DRG International)) levels were measured. This cohort was part of the VaccGene study, and genotyping is described below. This is a population-based longitudinal study of African Americans aged ≥21 years living in the Jackson, Mississippi metropolitan area in the USA48. This study was designed to evaluate risks of cardiovascular disease as described elsewhere48,49. Serum ferritin (Roche immunoturbidimetric assay), CRP (ELISA) and hemoglobin (Coulter analyzer, Beckman Coulter) levels were measured from blood samples collected at a single clinic visit. Whole blood was used to extract DNA using Puregene reagents (Gentra Systems). Genetic studies were conducted as described elsewhere49, and rs334 genotypes were extracted from exome sequencing datasets as described in Peloso et al.50.

n Coulter) levels were measured from blood samples collected at a single clinic visit. Whole blood was used to extract DNA using Puregene reagents (Gentra Systems). Genetic studies were conducted as described elsewhere49, and rs334 genotypes were extracted from exome sequencing datasets as described in Peloso et al.50. Children aged 10–16 years were recruited from the Nairobi Urban Health and Demographic Surveillance System51 as part of studies investigating the relationship between sickle cell trait, malaria and blood pressure52,53. Nairobi is located at a high altitude (1,800 m above sea level), and there is no evidence of malaria transmission there54. Population-wide censuses are conducted four times a year within the study area51. Using census data, we selected all school children aged 10–16 years who had a continuous record of residence since birth and had therefore had minimal exposure to malaria. To increase our efficiency in recruiting participants with sickle cell trait, we limited our recruitment to those who identified themselves as genetically descended from ethnic groups whose ancestral residence was in regions endemic for malaria (for example, Luhya, Luo, Teso, Mijikenda). The frequency of sickle cell trait is much higher in these ethnic groups55. We measured ferritin, CRP (ILab systems immunoturbidimetric assay, Instrumentation Laboratory) and hemoglobin (Coulter analyzer, Beckman Coulter) levels from stored blood samples. Sickle cell trait was typed by PCR35,36 using DNA extracted with the Qiagen DNA Blood Mini kit (Qiagen).

Infants born in Chris Hani Baragwanath Hospital living in the Soweto region of Johannesburg, South Africa were recruited from vaccine trials47 coordinated by the Respiratory and Meningeal Pathogens Unit (http://www.rmpru.com/). Mothers of the infants were approached if the infants had received all of their EPI vaccines up to 6 months of age. The infants were sampled prospectively at 12 months after receipt of the measles vaccine at 9 months of age. Single whole-blood samples were collected in vacutainer tubes containing EDTA for measurement of iron and inflammatory markers and DNA extraction. Serum ferritin (Chemiluminescent Microparticle Immunoassay, Abbott Architect), CRP (Multigent CRP Vario assay, Abbott Architect) and hepcidin (DRG Hepcidin-25 (bioactive) High Sensitive ELISA kit (DRG International)) levels were measured. This cohort was part of the VaccGene study, and genotyping is described below.

This is a population-based longitudinal study of African Americans aged ≥21 years living in the Jackson, Mississippi metropolitan area in the USA48. This study was designed to evaluate risks of cardiovascular disease as described elsewhere48,49. Serum ferritin (Roche immunoturbidimetric assay), CRP (ELISA) and hemoglobin (Coulter analyzer, Beckman Coulter) levels were measured from blood samples collected at a single clinic visit. Whole blood was used to extract DNA using Puregene reagents (Gentra Systems). Genetic studies were conducted as described elsewhere49, and rs334 genotypes were extracted from exome sequencing datasets as described in Peloso et al.50.

Children aged 10–16 years were recruited from the Nairobi Urban Health and Demographic Surveillance System51 as part of studies investigating the relationship between sickle cell trait, malaria and blood pressure52,53. Nairobi is located at a high altitude (1,800 m above sea level), and there is no evidence of malaria transmission there54. Population-wide censuses are conducted four times a year within the study area51. Using census data, we selected all school children aged 10–16 years who had a continuous record of residence since birth and had therefore had minimal exposure to malaria. To increase our efficiency in recruiting participants with sickle cell trait, we limited our recruitment to those who identified themselves as genetically descended from ethnic groups whose ancestral residence was in regions endemic for malaria (for example, Luhya, Luo, Teso, Mijikenda). The frequency of sickle cell trait is much higher in these ethnic groups55. We measured ferritin, CRP (ILab systems immunoturbidimetric assay, Instrumentation Laboratory) and hemoglobin (Coulter analyzer, Beckman Coulter) levels from stored blood samples. Sickle cell trait was typed by PCR35,36 using DNA extracted with the Qiagen DNA Blood Mini kit (Qiagen).

We measured hepcidin levels in children with severe malaria admitted to Kilifi County Hospital56. Sixty-two samples were randomly selected. Hepcidin levels were measured using the Hepcidin-25 (human) EIA Bachem kit and harmonized to DRG hepcidin values43. Severe malaria was diagnosed as P. falciparum parasites in the blood film plus clinical features of severe malaria, including hemoglobin levels <50gl-1, a hematocrit level of <15% (for severe malarial anemia) or a Blantyre coma score of 3 (for cerebral malaria).

Sickle cell trait (rs334) single-nucleotide polymorphisms (SNPs) were directly genotyped in the VaccGene populations (in Uganda, Burkina Faso and South Africa) using the HumanOmni 2.5M-8 (‘octo’) BeadChip array version 1.1 (Illumina) (n =648) and the Illumina Multi-Ethnic Global Array (n =197) performed by the genotyping core facilities at the Wellcome Trust Sanger Institute. Genomic DNA underwent whole-genome amplification and fragmentation before hybridization to locus-specific oligonucleotides bound to silica beads with a 3-μm diameter. Fragments were extended by single-base extension to interrogate the variant by incorporating a labeled nucleotide, enabling a two-color detection (Illumina, 2013, https://emea.illumina.com/content/dam/illumina-marketing/documents/products/brochures/datasheet_omni_whole-genome_arrays.pdf). Genotypes were called from intensities using two clustering algorithms (Illuminus and GenCall) in GenomeStudio version 2.0.5 (Illumina) incorporating data from proprietary predetermined genotypes. Details of genotyping and quality control are described elsewhere46. The variant rs334 was retained in all datasets following stringent quality control processes46.

ID was defined according to WHO recommendation as ferritin levels <12µgl-1 or <30µgl-1 in the presence of inflammation (defined as CRP>5 mgl-1, ACT>0.6gl-1 or AGP>1 gl-1) in children <5 years or<15µgl-1 in children ≦5 years15. Anemia was defined as hemoglobin levels<110gl-1 in children aged<5 years or hemoglobin levels<115gl-1 in children ≦5 years17. In the JHS and Nairobi,ID was defined as ferritin levels<30µgl-1. In the JHS, anemia was defined as hemoglobin levels<120gl-1 in women or<130gl-1 in men16,17. IDA was defined as the presence of ID and anemia17. Malaria parasitemia was defined as a blood slide positive for asexual P. falciparum parasites. Underweight was defined as weight-for-age z score< -2 using WHO Growth Standards57.

As ferritin is an acute-phase reactant and correlates positively with inflammatory markers58,59, we further defined ID after regression correction for the effect of inflammation on ferritin levels as proposed by the Biomarkers Reflecting Inflammation and Nutritional Determinants of Anemia project19,60. This approach predicts what the ferritin level would have been in the absence of inflammation or infection and then applies the corrected values to estimate the prevalence of ID. The regression-correction approach followed a three-step process. In the first step, internal reference values for inflammatory markers (CRP or ACT) were defined as the tenth percentile. CRP levels were measured in all studies except in The Gambia, where ACT levels were measured. In addition to CRP, AGP was measured in Malawi, Ghana, DRC and Cameroon; however, because CRP was positively correlated with AGP, and also for consistency, we only corrected for CRP in these studies. To avoid overcorrection for very low levels of inflammatory markers, only participants with CRP or ACT values above the tenth percentile had their ferritin values subtracted from observed values in equation (1)below19. In the second step, univariable linear regression models were applied to each study, with ferritin as the dependent variable, to estimate regression coefficients for the crude association between inflammatory marker and ferritin (β).In the third step, the regression coefficients estimated in step 2 were used to calculate adjusted ferritin values using equation (1). Ferritin and inflammatory markers were applied in the equations after ln transformation. (1)Ferritinadjusted = Ferritinunadjusted− β(CRP or CTobs−CRP or ACTref) ‘Obs’ is the observed value, and ‘ref’ is the reference value.

ficients estimated in step 2 were used to calculate adjusted ferritin values using equation (1). Ferritin and inflammatory markers were applied in the equations after ln transformation. (1)Ferritinadjusted = Ferritinunadjusted− β(CRP or CTobs−CRP or ACTref) ‘Obs’ is the observed value, and ‘ref’ is the reference value. We then defined ID using the regression-corrected unlogged ferritin value (that is, adjusted for the effects of inflammation) using the same thresholds that were applied to the uncorrected ferritin levels in the WHO recommendations (that is, ferritin levels <12µgl-1 in children <5 years or <15µgl-1 in children aged ≥5 years15).

Mendelian randomization is an instrumental variable analysis that reduces biases from confounding and reverse causation by using genetic variants to proxy the exposure (for example, malaria) and estimate a causal effect of that exposure on the outcome (for example, ID). To perform Mendelian randomization analyses, valid instrumental variables are required. The instrumental variables must be associated with the exposure of interest, and the effect size can be obtained from association studies. We therefore performed a systematic review and meta-analysis of published studies in African populations to determine the overall effect of common genetic polymorphisms that are associated with uncomplicated malaria. These polymorphisms included those influencing sickle cell trait,α-thalassemia and G6PD. We performed the search in the PubMed database for papers published before 20 March 2020. For sickle cell trait, the search terms included (‘malaria’ (title/abstract) or ‘malaria/blood’ (MAJR) or ‘malaria/ genetics’ (Medical Subject Headings (MeSH) terms) or ‘Malaria’ (MeSH) or ‘Malaria, Falciparum’ (MeSH) or ‘Plasmodium falciparum’ (MeSH)) and (‘HbAS’ (title/abstract) or ‘sickle cell trait’ (title/abstract) or ‘sickle cell trait/genetics’ (MeSH terms) or ‘sickle cell trait/blood’ (MeSH terms) or ‘Hemoglobin, Sickle’ (MeSH) or ‘Sickle Cell Trait’ (MeSH) and ‘Africa’ (MeSH terms)), yielding 594 articles. For α-thalassemia, the search terms included (‘malaria’ (title/abstract) or ‘malaria/ blood’ (MAJR) or ‘malaria/genetics’ (MeSH terms) or ‘Malaria’ (MeSH) or ‘Malaria, Falciparum’ (MeSH) or ‘Plasmodium falciparum’ (MeSH)) and (‘alpha-thalassemia’ (title/abstract) or ‘alpha-thalassemia/genetics’ (MeSH terms) or ‘alpha-thalassemia/ blood’ (MeSH terms) or ‘α-thalassemia’ (MeSH)) and Africa’ (MeSH terms) and yielded 65 articles.

ria/genetics’ (MeSH terms) or ‘Malaria’ (MeSH) or ‘Malaria, Falciparum’ (MeSH) or ‘Plasmodium falciparum’ (MeSH)) and (‘alpha-thalassemia’ (title/abstract) or ‘alpha-thalassemia/genetics’ (MeSH terms) or ‘alpha-thalassemia/ blood’ (MeSH terms) or ‘α-thalassemia’ (MeSH)) and Africa’ (MeSH terms) and yielded 65 articles. The search terms for G6PD included (‘malaria’ (title/abstract) or ‘malaria/blood’ (MAJR) or ‘malaria/genetics’ (MeSH terms) or ‘Malaria’ (MeSH) or ‘Malaria, Falciparum’ (MeSH) or ‘Plasmodium falciparum’ (MeSH)) and (‘G6PD’ (title/abstract) or ‘glucose-6-phosphate-dehydrogenase’ (title/abstract) or ‘glucose-6-phosphate-dehydrogenase/genetics’ (MeSH terms) or ‘glucose-6-phosphate-dehydrogenase/blood’ (MeSH terms)) and ‘Africa’ (MeSH terms) and yielded 187 articles. We restricted our analysis to studies conducted in Africa, as our outcome of interest (ID) was measured in African populations. We also focused on studies reporting IRR, as children are repeatedly infected with malaria; and, therefore, an IRR provides a better estimate of the true malaria risk reduction attributable to genetic polymorphisms. Studies included in the meta-analysis are shown in Supplementary Table 2.

All statistical analyses were conducted using Stata 13.0 (StataCorp). All measurements were taken from distinct samples. We conducted both cohort-specific and pooled analyses. When appropriate, we computed percentages, geometric means and deciles. We used two-tailed Student’s t-tests to test for differences in means of loge-transformed hepcidin concentrations between children with and without malaria parasitemia or between those with severe malaria and those without parasitemia. When appropriate, we fitted adjusted logistic regression models to determine the effect of sickle cell trait on ID. The cohort-specific ORs were meta-analyzed assuming fixed effects because there was little evidence of heterogeneity between studies. We used meta-analysis to determine the overall effect size and also to identify potential heterogeneity that may have been introduced by different populations and laboratory facilities. A P value <0.05 was considered statistically significant. All P values reflect two-tailed tests. Individuals with sickle cell disease (HbSS) were not included in the analyses, as numbers were few and there is little evidence suggesting that HbSS protects against uncomplicated malaria. We used the online (http://cnsgenomics.com/shiny/mRnd/) Mendelian randomization power calculator to calculate sample size. The sample size of 7,453 had power above 80% given the observed OR of the outcome variable per s.d. of the exposure variable of 2.65, 2% variation of clinical malaria that is explained by sickle cell trait61,62, 27% prevalence of ID and a type I error rate of 0.05.

ndomization power calculator to calculate sample size. The sample size of 7,453 had power above 80% given the observed OR of the outcome variable per s.d. of the exposure variable of 2.65, 2% variation of clinical malaria that is explained by sickle cell trait61,62, 27% prevalence of ID and a type I error rate of 0.05. To investigate whether uncomplicated malaria is causally associated with ID, a two-sample Mendelian randomization63 was conducted using the ‘mrrobust’ software package64 in Stata 13.0. This is a Wald ratio involving two estimates: the SNP –outcome effect divided by the SNP –exposure effect, in this case, the HbAS (rs334) –ID effect divided by the HbAS (rs334) –malaria effect. The first estimate (SNP –outcome) for sickle cell trait (HbAS, rs334) on ID was from the above-described community-based cohorts in which the overall log odds estimate was determined using the meta-analyzed cohort-specific estimates (Fig. 1c). The second estimate (SNP–exposure) was from the meta-analyzed overall log IRR of sickle cell trait on uncomplicated malaria (Extended Data Fig. 4). A causal log odds estimate equation (2), which is the ratio of the first estimate and the second estimate, was computed63. The causal OR was obtained by exponentiating the causal log odds estimate and interpreted as change in ID per unit increase in the log incidence rate of malaria. We further calculated the effect on ID of reducing malaria incidence by half. This was obtained by multiplying the causal log odds by the natural logarithm of 0.5 and then exponentiating the result20. (2)Causal log odds=log odds of iron deficiency in HbASlog incidence rate ratio of malaria in HbAS

ence rate of malaria. We further calculated the effect on ID of reducing malaria incidence by half. This was obtained by multiplying the causal log odds by the natural logarithm of 0.5 and then exponentiating the result20. (2)Causal log odds=log odds of iron deficiency in HbASlog incidence rate ratio of malaria in HbAS To determine whether sickle cell trait influences ID independently of malaria, we conducted sensitivity or zero relevance point analyses in negative controls, that is, populations that were not exposed to malaria14. To do this, we repeated our analyses in two separate populations that were not exposed to malaria including (1) an African American adult population (n=3,207) and (2) life-long adolescent residents of Nairobi (n=611), where there is no evidence of malaria transmission54. As this analysis suggested that an effect was absent in populations that were not exposed to malaria, this allowed us to interpret the effect among children exposed to malaria.

Individual study site ethical approvals were obtained for the Kilifi, Kenya study (by the Scientific Ethics Review Unit of the Kenya Medical Research Institute (KEMRI/SERU/CGMR-C/046/3257/2983)), the Entebbe, Uganda study (locally by the Uganda Virus Research Institute (GC/127/12/07/32) and the Uganda National Council for Science and Technology (MV625) and in the UK by the London School of Hygiene and Tropical Medicine (A340) and the Oxford Tropical Research (OTR) (39-12, 42-14 and 37-15) Ethics Committees), the Banfora, Burkina Faso study (by Ministere de la Recherche Scientifique et de l’Innovation in Burkina Faso (2014-12-151) and the OTR Ethics Committees(4112)), the Soweto, South Africa study (by the University of Witwatersrand Human Research (M130714) and the OTR Ethics Committees (1042-13 and 42-14)) and the West Kiang, The Gambia study (by the Gambian Government, Medical Research Council Ethics Committee (874/830)). For the additional seven study sites, individual study data transfer agreements were signed with the responsible study and/or institution’s principal investigator and the KEMRI-Wellcome Trust Research Programme study principal investigator, S.H.A. These studies had ethical approval to share de-identified data for further secondary analyses presented in this study. Informed written consent was obtained from all children’s parents or guardians.

Overall represents a fixed-effect meta-analysis of study-specific incidence rate ratio (IRR) by genetic polymorphism. Error bars indicate 95% confidence intervals. n shows the number of individuals included in the analysis. Few studies included G6PD homozygous females and numbers were small (Table S2). Het, heterozygous; Hom, homozygous. Overall represents a fixed-effect meta-analysis of cohort-specific odds ratios. Error bars indicate 95 % confidence intervals. n shows the number of individuals included in the analysis. Numbers for IDA are fewer compared to those for ID since not all children had hemoglobin concentrations measured. ID was defined using ferritin levels adjusted for the effects of inflammation using a regression-correction approach as developed by BRINDA. Overall represents a fixed-effect meta-analysis of cohort-specific odds ratios. Error bars indicate 95 % confidence intervals. n shows the number of individuals included in the analysis. Overall represents a fixed-effect meta-analysis of study-specific incidence rate ratio (IRR). Error bars indicate 95 % confidence intervals. n shows the number of individuals included in the analysis.

ID was defined using ferritin levels adjusted for the effects of inflammation using a regression-correction approach as developed by BRINDA. Overall represents a fixed-effect meta-analysis of cohort-specific odds ratios. Error bars indicate 95 % confidence intervals. n shows the number of individuals included in the analysis. Overall represents a fixed-effect meta-analysis of study-specific incidence rate ratio (IRR). Error bars indicate 95 % confidence intervals. n shows the number of individuals included in the analysis. a, Individuals carrying normal beta hemoglobin gene (HbAA) are not protected from malaria. Malaria up-regulates production of hepcidin through inflammatory and non-inflammatory pathways and by increasing the prevalence of other infections. Hepcidin in turn blocks iron absorption.b,Sickle cell trait (HbAS) partially protects individuals from malaria infection, therefore inflammation is reduced leading to reduced hepcidin stimulation and increased iron absorption. Error bars indicate 95 % confidence intervals. n indicates biologically independent samples. Horizontal dotted line indicates the threshold of hepcidin above which iron absorption is inhibited (5.5 μg/L). Inflammation was defined as CRP >5 mg/L, ACT >O.6g/L or AGP >1 g/L. Malaria was defined as a blood slide positive for asexual P. falciparum parasites. Hepcidin was measured in the Burkina Faso, Western Kenya, Uganda, The Gambia, and Kilifi, Kenya cohorts.