Malaria parasitaemia, measures of malnutrition and moderate-severe anaemia were significant co-morbidities in this study population of HIV-exposed and unexposed Ugandan infants living in a rural area. Interestingly, HIV-exposed infants had lower odds of malaria parasitaemia compared to HIV-unexposed infants. This may be due to the higher use of malaria preventative measures, such as bed nets and TS prophylaxis, in the HIV-exposed group. HIV-exposed children also had a lower prevalence of moderate-severe anaemia. Similarly, this could be explained by the lower prevalence of parasitaemia in the HIV-exposed group, as parasitaemia was strongly associated with moderate-severe anaemia. In contrast, HIV-exposed children had a higher prevalence of measures of malnutrition after controlling for other factors measured in this study. Lower household wealth was also independently associated with a higher prevalence of measures of malnutrition. Also of interest was the finding that having a well-constructed house was independently associated with a significantly lower prevalence of malaria parasitaemia and a trend towards a lower risk of anaemia. These findings provide some evidence of the complex interplay between these important co-morbidities and the factors associated with their prevalence in a novel population of HIV-exposed and unexposed Ugandan infants.
The prevalence of malaria parasitaemia in this study population was 20%, which is considerably lower than the 38% prevalence reported in children <5 years of age from a malaria indicator survey conducted in the same region in 2009
. This difference is probably a reflection of the lower risk of malaria and asymptomatic parasitaemia seen in young infants living in highly malaria-endemic areas due to the placental transfer of antibodies
. Indeed, even with the narrow age range of four to six months, older age was significantly associated with a higher prevalence of parasitaemia. Other well-described factors associated with parasitaemia in this study included testing following the rainy season and the protective effect of sleeping under a bed net and the use of TS prophylaxis. Factors associated with parasitaemia in this study that are more difficult to explain include the lower risk seen in females and infants born to older mothers. One of the more interesting findings in this study was the association between living in a well-constructed house and having a lower risk of parasitaemia. Previous studies from Africa have reported that better constructed houses are associated with a lower risk of malaria
[14, 15]. In addition, a study conducted in western Kenya found that modifying housing structure significantly restricted mosquito entry and thus human exposure to malaria vectors
. Other studies have found associations between higher socio-economic status and a lower risk of malaria
[15, 17] but this association was not observed in this study. Findings have been varied in previous studies evaluating the associations between malnutrition and malaria. Some studies suggest a protective effect against malaria for wasted children
 and children with stunting
. Other studies have found either no association between malnutrition and malaria
[20–22] or even an increased risk of malaria among stunted
[5, 23] and underweight children
. In this population of very young HIV negative infants, no association between measures of malnutrition and malaria parasitaemia was observed.
The prevalence of indicators of malnutrition in this population, 10% stunting, 7% underweight and 3% wasted, was somewhat lower than the rates reported for the same area in previous years
 (17%, 8% and 11%, respectively) and could be explained by the high rates of reported breastfeeding in the population included in this study. Within this novel study population, however, HIV-exposed infants born to HIV-infected mothers had significantly higher risk of malnutrition while controlling for other measured factors. It follows that maternal factors such as breastfeeding and maternal nutrition could be responsible for this observed difference. Poor nutritional status of HIV-infected mothers was suggested in a concurrent clinical trial being conducted at the same location in Uganda, where they were observed to have significantly lower than recommended gestational weight gain
. In addition, several other studies conducted in Africa had similar findings of increased prevalence of malnutrition among HIV-exposed but uninfected infants
[24–27]. Not surprisingly, children from the poorest households were also at increased risk of malnutrition.
The prevalence of moderate-severe anaemia in this population of young infants was 12%, nearly identical to those rates reported for the region in a malaria indicator survey conducted in 2009
. Many factors were independently associated with an increased risk of anaemia, yet only increasing age of the infant and parasitaemia retained significance in the final statistical model. Increased risk of anaemia with increasing age of the infant correlates to what is seen at the population level in Uganda
 and could potentially reflect declining nutritional status or the previously discussed increasing risk of malaria parasitaemia in the older infants of this population. What is interesting is that this trend was significant even within the narrow age range (four to six months) of this study population. In addition, there was an observed trend towards decreased risk of anaemia among females, infants in the wealthiest households and those living in a well-constructed house. Wealth and house construction in conjunction with the highly significant increase in odds of anaemia seen with parasitaemia suggest that malaria prevention could have a significant impact on decreasing rates of anaemia in this population. And in fact, previous studies have shown anaemia to be more prevalent among parasitaemic children
[28, 29] as well as describing the link between wealth and malaria and anaemia
The primary limitation of this study was the cross-sectional study design. Associations presented could have been confounded by unmeasured factors and therefore causal inferences cannot be drawn. In addition, the temporal relationship between exposure variables and outcomes of interest cannot be observed. Finally, because this study enrolled participants using convenience sampling and was done in a single geographically defined area, care should be taken in generalizing the results to the other populations.