Background The global prevalence of X-linked glucose-6-phosphate dehydrogenase (G6PD) defi ciency is thought to be
a result of selection by malaria, but epidemiological studies have yielded confusing results. We investigated the
relationships between G6PD defi ciency and both malaria and non-malarial illnesses among children in Kenya.
Methods We did this study in Kilifi County, Kenya, where the G6PD c.202T allele is the only signifi cant cause of G6PD
defi ciency. We tested the associations between G6PD defi ciency and severe and complicated Plasmodium falciparum
malaria through a case-control study of 2220 case and 3940 control children. Cases were children aged younger than
14 years, who visited the high dependency ward of Kilifi County Hospital with severe malaria between March 1, 1998,
and Feb 28, 2010. Controls were children aged between 3–12 months who were born within the same study area
between August 2006, and September 2010. We assessed the association between G6PD defi ciency and both
uncomplicated malaria and other common diseases of childhood in a cohort study of 752 children aged younger than
10 years. Participants of this study were recruited from a representative sample of households within the Ngerenya and
Chonyi areas of Kilifi County between Aug 1, 1998, and July 31, 2001. The primary outcome measure for the casecontrol
study was the odds ratio for hospital admission with severe malaria (computed by logistic regression) while for
the cohort study it was the incidence rate ratio for uncomplicated malaria and non-malaria illnesses (computed by
Poisson regression), by G6PD defi ciency category.
Findings 2863 (73%) children in the control group versus 1643 (74%) in the case group had the G6PD normal
genotype, 639 (16%) versus 306 (14%) were girls heterozygous for G6PD c.202T, and 438 (11%) versus 271 (12%)
children were either homozygous girls or hemizygous boys. Compared with boys and girls without G6PD defi ciency,
we found signifi cant protection from severe malaria (odds ratio [OR] 0·82, 95% CI 0·70–0·97; p=0·020) among
G6PD c.202T heterozygous girls but no evidence for protection among G6PD c.202T hemizygous boys and
homozygous girls (OR 1·18, 0·99–1·40; p=0·056). Median follow-up for the mild disease cohort study was
2·24 years (IQR 2·22–2·85). G6PD c.202T had no eff ect on other common diseases of childhood in heterozygous
girls (incidence rate ratio 0·98, 95% CI 0·86–1·11; p=0·82) or homozygous girls or hemizygous boys (0·93,
0·82–1·04; p=0·25), with the sole exception of a marginally signifi cant increase in the incidence of helminth
infections among heterozygous girls.
Interpretation Heterozygous girls might be the driving force for the positive selection of G6PD defi ciency alleles.
Further studies are needed to defi nitively establish the mechanisms by which G6PD defi ciency confers an advantage
against malaria in heterozygous individuals. Such studies could lead to the development of new treatments.