This study set out to examine in children, the effect of malaria with neurological involvement on cognitive ability, academic achievement and behaviour three months after infection. Malaria with neurological involvement was associated with increased behavioural problems and a minimal effect on attention but did not affect academic functioning.
This is the first study to use standardized measures to assess both the behavioural and academic outcomes after malaria with neurological involvement. Recent studies from Uganda show that malaria with neurological involvement is associated with severe behavioural problems like hyperactivity, aggressive behaviour, autism spectrum disorders, internalizing problems (withdrawn/depressed and thought problems) and externalizing problems (aggressive behaviour and oppositional defiant behaviour) [29, 30]. The current study did show an increase in internalizing problems as previously shown  and showed a modest but non-significant increase in externalizing problems similar to those reported by Idro et al  in a clinic-based case-control study. The difference in the externalizing problems results across the studies may be due to the assessment methods used in the two earlier studies compared to the current one. One study  used the CBCL but did not use Ugandan norms to determine behavioural problems, while Idro et al  reviewed hospital records that may not have captured the same information for all patients. In addition, the children in Idro et al  were from a specialist child neurology clinic which treats patients with severe behavioural problems. This may explain why Idro and colleagues  report severe overt behavioural problems while this study, and a previous one  using the CBCL, both report internalizing problems that are covert in nature. The present prospective study is best suited to assess behavioural outcomes after severe malaria than a cross-sectional study despite the small sample size and loss to follow-up.
Both retrospective and prospective studies have shown that severe malaria does affect cognition with rates of impaired cognition ranging between 14 to 26% [4–8, 32]. The prospective studies from Uganda especially show an increasing trend of cognitive deficits from 21% at six months to 26% at 24 months [5, 6]. However it is the assessments at discharge and three months follow-up in these studies that may explain the present study's finding of minimal effect on cognition. At discharge, 36% of the children with cerebral malaria were impaired compared to 11% of the controls, but this rate fell to 19% vs 7% at three months . What is more important is that these differences in impairment were significant at discharge but not at three months. Important to note is these two Ugandan studies used combined cognitive test scores to determine children who were impaired while the present study is assessing individual cognitive skills.
This implies that much as malaria with neurological impairment has been associated with cognitive impairment, there seems to be a gradual development of these deficits, which become more pronounced and severe, as was shown at the 24-months assessments by John et al . This may explain why minimal effect was seen on cognition and no effect on academic achievement in the current study done at three months follow-up. The high frequency of impairment at discharge in the Boivin et al  study could therefore be attributed to the malaise associated with cerebral malaria illness and not necessarily the effect of the disease on the brain; otherwise the frequency of impairment would not drop from 36% at discharge to 19% at three months and then increase to 21% at six and 26% at 24 months, as was observed.
In a re-analysis of cognitive test scores from John et al  and Boivin et al  using a novel method employing a global normalized Z-score [Bergemann TL, Bangirana P, Bruno G, Boivin MJ, Connett JE, John CC: Statistical approaches to assess the effects of disease on neurocognitive function over time, Submitted], more differences were observed in test scores between the malaria group and the community controls than were seen in the earlier two studies. Though significant differences were observed at zero, three, six and 24 months, the effect sizes were smallest at three months with a p value of 0.04 compared to the other time points with very small p values and large effect sizes. This too supports the conclusion that cognitive deficits after severe malaria are not evident at three months after the illness.
The heterogeneous nature of the malaria group may be another reason why minimal group-average effects were seen in the current study. Of the sixty-two children with malaria in this study, nine had cerebral malaria, 34 had seizures, and 19 had impaired consciousness but no coma. The study by Boivin et al , which assessed cognition at three months, included only children with cerebral malaria, the most severe form of malaria. Given the less severe effects of cerebral malaria on cognition seen at three months in Boivin et al  and Bergemann and colleagues [Bergemann TL, Bangirana P, Bruno G, Boivin MJ, Connett JE, John CC: Statistical approaches to assess the effects of disease on neurocognitive function over time, Submitted], compared to the other time points, it was unlikely that the present study, using similar assessments and having more children with the less severe forms of malaria, would show cognitive deficits at the same time point.
Similarly, absence of an effect on the academic achievement scores could also be attributed to the heterogeneous sample and assessment at three months, which is too early to see any effect. This contrasts with studies in which effects on academic performance were seen at two weeks after infection with less severe forms of malaria . Other studies have also observed effects on academic performance after infection with less severe malaria [23, 24]. One difference between this study and the above studies is that they used measures of academic performance developed in the country or specific region while assessment of academic achievement in this study was developed in the west. It is likely that the WRAT-3 in this study is not a sensitive measure of academic achievement in the early period after malaria infection. These contrasting results call for further assessment of school performance after severe malaria in Ugandan children using both local and western tests.
This study adds to the body of knowledge that malaria with neurological involvement does affect behaviour and cognition, highlighting the need for intervention among child survivors. The observation that these deficits may become more evident with time also necessitates interventions to arrest this trend. It has been previously shown that computerized cognitive rehabilitation training does improve cognition and behaviour after severe malaria infection [44, 45].
However, there are several limitations in this study that necessitate caution when interpreting the results. The case-control design used is not sufficient to test causality especially with the limited follow-up time and demographic differences in the cases and controls. The observed effects could also include artefacts of these differences. The heterogeneous sample of the malaria group may affect outcomes since risk factors for cognitive deficits like coma, seizures, neurological signs, hypoglycaemia and malnutrition [3, 5, 6, 32, 46] may be present in some forms of malaria and not others. This is more significant when the numbers of children with these different malaria forms are not similar in the sample. In addition, diagnosis of cerebral malaria did not include retinopathy examination which increases the specificity in cerebral malaria diagnosis .