In rural communities in southern highland Rwanda, G. duodenalis infects two of three predominantly asymptomatic children, is underestimated by conventional microscopy of single stool samples, and contributes to underweight and clinically assessed malnutrition. This suggests an underrated but considerable burden of disease due to G. duodenalis.
The prevalence of G. duodenalis
of 60% in Rwandan children – who mainly were randomly selected from communities – considerably exceeds figures reported from other parts of the world 
. The superior sensitivity of PCR in detecting G. duodenalis
has recently been shown in Danish patients 
. In contrast, PCR and microscopy had similar sensitivities in a Dutch study 
. In our study, light microscopy, which commonly is the only method available in resource-poor areas, failed to detect more than two-thirds of actually present G. duodenalis
infections. Repeated microscopic examinations (which are difficult to implement) or the use of immunoassays could have reduced this discrepancy. In addition, chronically and/or repeatedly infected individuals in highly endemic settings might shed low cyst numbers. Thus, microscopy-based prevalence figures from high-endemicity areas may be underestimations, and analyses comparing individuals categorized by microscopy into Giardia
infected and non-infected may be confounded by a significant number of false negatives. While some false negatives may persist even when applying PCR assays, that risk cannot be determined accurately in the absence of a well defined gold standard.
As to the clinical significance of G. duodenalis
among community children, we herein describe associations of symptoms with the presence of the parasite. G. duodenalis
infection is known to vary widely in clinical manifestation including acute, chronic, and asymptomatic courses 
. Therefore, we did not attempt to a priori
define clinical giardiasis in our study. Nevertheless, no evidence for causation of gastrointestinal symptoms was seen, with the potential exception of abdominal distension. The association with malnutrition, however, was impressive: approximately every third and every sixth child with microscopic infection had underweight and clinically assessed severe malnutrition, respectively, although causation cannot be clarified in a cross-sectional study. Association does not mean causality, and reduced sample sizes in subgroup analyses increases the chance of type 1 statistical errors. Selection bias during recruitment at home, e.g., due to preferential presentation by the parents of children with (a recent history of) sickness cannot completely be excluded. However, recruitment teams were instructed to select children from households randomly. Also, confounding cannot completely be excluded but our results remained stable in multivariate analyses adjusting for socio-economic parameters and for the number of other intestinal pathogens serving as a proxy parameter for polyparasitism (data not shown). Notably in this regard, A. lumbricoides
infections found in one third of the children had no clear-cut effect on malnutrition (data not shown), similar to findings from Brazil and Iran 
. Clinically assessed severe malnutrition in this study reflects the subjective evaluation of children's status which, nevertheless, was supported by comparison with growth percentiles und MUAC. In interpreting our results, it consequently needs to be kept in mind that this parameter is not standardized. Though, the stronger association with G. duodenalis
of clinically assessed severe malnutrition as compared to underweight may also indicate a greater selectivity of the former. Unfortunately, we were unable to analyze the influence of G. duodenalis
on stunting and wasting, i.e., categories based on children's height, because this parameter was inconsistently measured in the field study, had implausible values, and respective Z-scored were consequently omitted. However, our interpretation of malnutrition as a consequence of giardiasis is in line with other studies 
Children with submicroscopic G. duodenalis infection had a prevalence of clinically assessed severe malnutrition, which was intermediate between children without and with microscopic infection. In multivariate analysis, submicroscopic infections still tended to be associated with clinically assessed severe malnutrition but showed no link with underweight. Longitudinal observations and immunological studies may help to understand the relevance of this type of infection and the underlying mechanisms.
assemblage B parasites clearly predominated (86%), which is in accordance with most larger studies (recently reviewed in 
). Interestingly, we detected only one co-infection by both assemblage A and B parasites. By random distribution, a much higher proportion would have been expected possibly reflecting cross-assemblage competition or protection. Unfortunately, we were unable to type all Giardia
infections. Although the typing rate was higher than in other studies (e.g., 
) this is most likely due to the fact that typing relied on the single copy tpi
gene while diagnosis was based on the amplification of the multicopy rDNA. However, this is unlikely to have resulted in a significant bias in the determination of the prevalence of A and B genotypes in the study population. The success of typing correlated with the ct values in the diagnostic PCR, i.e. the content of target DNA in the samples and this was not significantly different for typed samples independent of their A or B status (not shown).
In our study, children infected with assemblage A isolates showed increased proportions of abdominal pain and vomiting while assemblage B infections associated with underweight and clinically assessed severe malnutrition. Much of the available data on the role of assemblages in clinical disease is inconsistent 
. Our data are in line, however, with findings from Western Australia where assemblage A parasites were more likely to be found in symptomatic children with diarrhoea while assemblage B parasites were more prevalent in asymptomatic children 
. Likewise, against high prevalences of assemblage B parasites in Bangladesh, assemblage A isolates were found to be associated with symptomatic disease, i.e., diarrhoea 
. In contrast, abdominal pain in children was associated with assemblage B parasites in Argentina 
. Further investigations in our study population will help elucidating whether - in contrast to possibly endemic assemblage B parasites - the genetically distinct assemblage A parasites may appear epidemically and thus are more correlated with acute abdominal symptoms.
Socio-economic factors had no great influence on the presence of G. duodenalis
in the present study. One reason involved may be a limited selectivity of the questionnaires used in this generally poor agricultural and rural population. Also, we were unable to identify hot spots, let alone their characteristics, potentially underlying the pronounced spatial differences in infection prevalence. In multivariate analysis, only the increases of G. duodenalis
with children's age and with a high number of siblings as well as the reduction with breastfeeding remained statistically significant. Increasing prevalence with increasing number of siblings supports the role of transmission within households. Protection from G. duodenalis
infection by breast milk has previously been reported 
. Considering the high endemicity of G. duodenalis
in the study area it seems plausible that protective antibodies 
are transmitted by the breast-milk of seropositive mothers. Soluble IgA has also been shown to contribute to the clearance of Giardia
spp. in murine studies 
. In addition, lactoferrin and leukocytes in breast milk could be involved in protection 
The association with impaired child growth as observed in the present and other studies 
suggests control of G. duodenalis
infection to be potentially rewarding, e.g. by preventive chemotherapy. Yet, the influence of repeated treatment of G. duodenalis
infection on anthropometric development is unclear 
: evidence for a beneficial impact has been observed in Brazilian children 
but not in Bangladeshi infants 
. Beyond that, G. duodenalis
infection in highly endemic areas has been associated with protection from acute diarrhea 
. Considering the importance of diarrhoea as a leading cause of childhood mortality, such a protective effect might outweigh impaired child growth due to G. duodenalis
infection. To date, there is insufficient data to balance the potential effects of anti-Giardia
treatment against each other. Lastly, food supplementation in areas highly endemic for G. duodenalis
also appears to be complex. G. duodenalis
associated protection from diarrhea has recently been reported to be lost in Tanzanian children with multi-nutrient supplementation 
while vitamin A plus zinc supplementation reduced the incidence of giardiasis in Mexican children 
. Thus, large-scale longitudinal studies are needed to disentangle the role of G. duodenalis
among the interacting factors contributing to malnutrition and diarrhea in high-endemicity regions and to estimate the potential impact of control measures.
In conclusion, our data provide evidence of a very high prevalence of G. duodenalis assemblage B without causing diarrhoea but associated with underweight and clinically assessed severe malnutrition in Rwandan children. The underestimation of G. duodenalis by light microscopy suggests that prevalences and consequences have previously been underrated and asks for the implementation of more sensitive diagnostic, yet simple diagnostic tools. The clarification of the clinical significance of G. duodenalis in high endemicity areas needs to take account of an abundance of submicroscopic infections.