Severe Malaria and Academic Achievement

The current study recruited and enrolled children who previously participated in a prospective cohort study conducted at Mulago Hospital, Kampala, Uganda in which children between 18 months and 12 years of age with CM, SMA, or community children (CC) were enrolled⁵  and evaluated for cognitive outcomes over 2 years of follow-up after the severe malaria episode (CM or SMA) or enrollment (CC). Cerebral malaria was defined as: (1) coma (Blantyre Coma Score ≤2 or Glasgow Coma Score ≤8); (2) Plasmodium falciparum on blood smear; and (3) no other known cause of coma. Severe malarial anemia was defined as presence of Plasmodium falciparum on blood smear in children with hemoglobin level ≤ 5 mg/dL. Community control children were recruited from the household or neighborhood of children with CM or SMA within 1 year of age of a recently enrolled child with CM or SMA but were not specifically matched to a child with CM or SMA. Detailed information on exclusion criteria for each group and medical treatment of children with CM or SMA appears in previous publications^5,20  and in the Supplemental Information.

Children from the parent study who had attained ≥5 years of age were eligible for the current study. A sample size of 200 children (100 children with CM or SMA, 100 community children) was estimated to have ≥80% power to detect a 0.4 SD difference in academic achievement scores between children with severe malaria (CM or SMA) and CC.

Malaria incidence during the first 2 years’ postsevere malaria hospital discharge was collected from the parent study database. During the current study, caregivers were asked about malaria episodes during the period after the parent study was completed, and the number of malaria episodes during this period for which the child was given antimalarial medication was recorded. Malaria episodes were reported to be uncomplicated malaria; however, the exact details on admission versus outpatient treatment were not available. The total number of malaria episodes was calculated as the number of episodes recorded during the 2-year follow period plus the parent-reported number of episodes from the end of the parent study to the current study testing date. Malaria incidence was calculated as: (number of malaria episodes/number of years from parent study enrollment to time of testing) × 100, to give incidence per 100 person-years.

The Wide Range Achievement Test, Fourth Edition, an achievement test used to measure academic achievement from ages 5 to 94 years,²¹  was used for the current study. The instrument contains subtests for math computation (ability to count, read number symbols, solve oral problems and perform math calculations), spelling (ability to spell 15 letters and 40 words), reading (ability to identify and name 15 letters and pronounce 42 words), and sentence comprehension (ability to complete a sentence with the appropriate word for the sentence to make sense).²¹  The third edition of this test was adapted for use in Uganda to assess academic achievement among children with HIV,²²  malaria,²³  and deaf children.²⁴  The present test was adapted in a similar way with translation of test instructions to Luganda, the local language in the study setting, and use of words and sentences familiar to Ugandan children. The Green Test form was administered with the reading test first as a prerequisite for the sentence comprehension test. Any child that scored 4 and less on the reading subtest part 2 was not administered the sentence comprehension test, as directed in the administering instructions. This was followed with administration of math computation and spelling subtests.

Baseline demographic and clinical characteristics were compared by disease group by using Fisher’s exact test for categorical measures, 1-way analysis of variance for Gaussian continuous measures, and the Kruskal-Wallis test for non-Gaussian continuous measures. Pairwise comparisons were done by using Fisher’s exact test for categorical variables and Dunn’s test following Kruskal-Wallis for non-Gaussian continuous measures, both with Bonferroni correction. Age-adjusted z-scores for academic achievement domain outcomes were created by using scores of the CC as reference. For each outcome, the z-score was computed as (actual score - mean score for child’s age)/SD, in which the mean score for age were computed by fitting a linear regression model to data for all CC children and the SD was the SD of the actual scores of the CC children. To examine the effect of CM and SMA on academic achievement, children surviving CM and SMA were compared to the CC by using ordinary least-square linear regression, adjusted for age and time from enrollment in the primary study. Comparisons between CM versus CC and SMA versus CC were corrected for multiple comparisons using Dunnett’s test. Adjusted averages are the estimated marginal means from linear regression models.

To examine the association between postdischarge malaria incidence and academic achievement domains among each disease group, stratified multivariable linear regression models were used with square-root–transformed malaria incidence and adjusted for age and time from enrollment in primary study. To test how postdischarge malaria incidence influenced the role of CM or SMA on academic achievement scores, path analysis under the framework of structural equation modeling was employed, in which the models were adjusted for age and time from enrollment in primary study. P values of ≤ 0.05 were considered statistically significant. Statistical analyses were conducted by using Stata 16.1²⁵  and R 4.1.1.²⁶  statistical packages.

Ethical approval was granted by Makerere University School of Medicine Research and Ethics Committee (REC 2014-047) and the Uganda National Council for Science and Technology (SS 4061). Written informed consent and assent were obtained from the parents or caregivers and children respectively.

The child’s identification and caregiver contact number were obtained from the parent study registry. Caregivers were contacted through phone calls and invited to bring their child or children to attend a 1-hour testing. Out of the 718 children originally enrolled in the parent study, 300 children were randomly selected by using a table of random numbers without replacement for recruitment from those had completed 2-year follow-up. The goal was to enroll at least 200 of the 300 children after accounting for loss to follow-up, death, and declined participation in the new study. 229 children (73 with CM, 56 with SMA, and 100 CC) were enrolled (Fig 1). Study testing was conducted on average 67.1 months (range, 19–101 months) after the episode of CM or SMA or previous study enrollment (CC). Children in the CM, SMA, and CC groups were similar in demographic and clinical characteristics, but children with CM or SMA had a significantly higher frequency and incidence of malaria episodes postdischarge than CC and more days of missed school than CC (Table 1).

We compared the current study cohort to the parent study to determine if differences seen in the current study could be because of selection bias from the previous study. As compared to children from the parent study who were not enrolled in the current study, children in the current study had higher socioeconomic status scores (CM, SMA, CC), were older (CM and SMA), had higher home environment scores (CM and SMA), and had higher maternal education levels (CM) than children in the current study but not in the current study (Supplemental Table 3). A higher proportion of children with CM or SMA in the current study had preschool education compared to those not in the current study but in the parent study, whereas among CC the opposite occurred: a lower proportion of those in the current study had preschool education compared to those not in the current study but in the parent study (Supplemental Table 3).

After adjusting for age and time between severe malaria episode and academic achievement assessment, reading scores were lower in children with CM (mean difference [95% confidence interval] −0.15 [−0.27 to −0.03], P = .02) or SMA (−0.15, [−0.28 to −0.02], P = .02) than in CC (Fig 2, Supplemental Table 4). In unadjusted analyses, math computation scores were lower in children with CM than CC, and sentence comprehension scores were lower in children with SMA than in CC, but these differences were not significant after adjustment for age and time from enrollment in the multiple regression analysis (Fig 2, Supplemental Table 4).

In children with CM or SMA, academic achievement was affected by incidence of malaria after discharge. An increase in malaria incidence was associated in children with CM with lower scores in spelling (mean difference [95% confidence interval] −0.88 [−1.59 to −0.17], P = .02) and reading (−0.76 [−1.43 to −0.10], P = .03), and in children with SMA with lower scores in spelling (−1.00 [−1.70 to -0.29, P = .01) (Table 2). Academic achievement scores in CC group did not differ according to malaria incidence.

Pathway analysis was used to further evaluate the contribution of repeated uncomplicated malaria episodes after initial severe malaria hospital discharge to poorer reading scores in children with CM and SMA compared to CC, because malaria episodes could be in the causal pathway of impaired reading scores (Fig 3). Compared to CC, CM, and SMA were each associated with higher incidence of postdischarge malaria (P < .001), and incidence of postdischarge malaria was associated with lower reading z-scores in both groups (β = −0.59, P = .002). In the path analysis model, the indirect path effect through the incidence of postdischarge malaria had a θ = −0.05 for CM versus CC and θ = −0.05 for SMA versus CC, both of which were highly statistically significant (P < .001 for both). Postdischarge malaria incidence was responsible for 30% of the overall difference between CM and CC (P = .04) and 38% of the overall difference between SMA and CC in reading z-scores (both P = .04).

Previous studies have shown that CM and SMA are associated with long-term cognitive impairment, and repeated episodes of uncomplicated malaria are associated with poorer academic achievement.^5,7,9,17,27  In the current study, we show that children with CM or SMA also have poorer academic achievement: 2 to 9 years after the episode of severe malaria, children with CM or SMA had poorer reading scores than community children from the same household or neighborhood. Additionally, we show that repeated episodes of uncomplicated malaria after the episode of severe malaria contribute significantly to poorer reading scores among these children. Academic achievement scores in all areas were lower in children with CM or SMA than CC but achieved statistical significance only in the area of reading.

In previous prospective studies conducted in Uganda we found that children with CM or SMA had lower scores on overall cognition than community controls 12 months after the severe malaria episode,^5,28  but academic achievement was not affected in child survivors of malaria with neurologic involvement 3 months after recovery from the illness.¹⁹  The previous studies prompted the current study on long-term academic achievement, which shows that children with CM or SMA have poorer academic achievement in reading, an average of 6 years after the severe malaria episode (range 2–9 years). Studies in Sri Lanka found that repeated malaria infections among school-aged children impaired their cognitive performance and had adverse effects on children’s academic performance in mathematics and language compared to children who did not have repeated malaria episodes.^17,27  The current study confirms that repeated malaria episodes, which are particularly common in children with previous severe malaria,²⁹  contribute substantially to the worsened performance of children in reading, because pathway analysis showed that postdischarge malaria incidence was responsible for 30% and 38% overall difference in reading scores between CM or SMA, respectively, and CC. The exact causes of cognitive and academic impairment after severe and uncomplicated malaria are still not well defined, but it is likely that worse academic achievement in children with repeated malaria after CM or SMA is because of a combination of factors such as repeated or more chronic anemia that occur with uncomplicated malaria and the effects of missed school days, because children with CM or SMA missed more school days than children with CC. Malaria incidence would likely have been even higher in this cohort if they had not been enrolled in the parent study, because insecticide-treated bed nets and malaria education were provided to all participants in the study.

We found that children who had CM or SMA had significantly lower academic achievement scores in the critical area of reading. Children who perform poorly in reading may experience low self-esteem because they recognize that they can’t read or spell as well as their peers.³⁰  Additionally, these children may be targets for bullying which can lead to dropping out of school and poorer economic attainment.^31,32  Thus, these children may not realize their full socioeconomic potential.^31,33  Poor reading not only hinders an individual’s interaction and participation in society but also increases their economic vulnerability in the future.^33,34  Severe malaria and repeated episodes of malaria have substantial socioeconomic and behavioral costs²⁹  that may lead to school absenteeism, grade repetition or inability to complete the school term or year after school fees have been paid. Our study findings that severe malaria affects long-term academic performance highlight the need for malaria prevention and remedial learning for children surviving severe malaria.

The study cohort differed from the original cohort in that children with severe malaria in the current study had higher socioeconomic status, a better home environment, and more paternal, maternal, and preschool education than those not in the current study but in the original cohorts. All of these differences would bias toward less differences between the children with CM or SMA and the community children, suggesting that differences in the full cohort would be even larger. Parental report of malaria episodes may not be completely accurate, because the diagnosis could not be confirmed. Since febrile episodes due to any cause are often treated as malaria, this would likely bias findings toward the null hypothesis. The study results emphasize the importance of health education regarding malaria prevention to children and their parents or caregivers and encourage seeking prompt medical care for all forms of malaria. It will also be important for the parents or caregivers to engage and provide academic support with the child’s teachers and follow through with what the child has missed at school during their illness. Areas for future research in children with severe malaria include evaluating whether activities like word boxes, an instructional method delivered by teachers, may enhance the child’s reading ability.³⁵  Comorbid conditions and sibling influence, which were not evaluated in this study, could also affect academic achievement, and should be assessed in future studies. Another study limitation is that children in the current study were previously enrolled in a study of the long-term cognitive effects of CM or SMA, so the study sample may not be representative of all children with CM or SMA.

Future studies should also evaluate whether postdischarge malaria chemoprevention in children with severe malaria leads to better long-term academic outcomes. A recent study in Uganda and Kenya found substantial reduction in postdischarge hospital readmission or death in children with severe anemia (most often with concurrent malaria) given 3 months of malaria chemoprevention with dihydroartemisinin-piperaquine.³⁶  Malaria chemoprevention prescribed at the time of discharge has the potential to prevent future malaria infections and school absenteeism and improve attention, school attendance, and academic achievement.^37–39  Previous studies in Sri Lanka showed that malaria prophylaxis in community children in highly malaria endemic areas resulted in improved cognition, but studies of community chemoprevention in other countries have showed variable success.⁴⁰  The current study suggests malaria chemoprevention may be more effective for improvement of academic outcomes if targeted among children with severe malaria, which is a high-risk population.

Children with CM and SMA have worse long-term academic achievement than community children without severe malaria, particularly in the area of reading. Postdischarge malaria episodes contribute significantly to this association. Postdischarge malaria chemoprevention should be assessed as an intervention to improve long-term academic achievement in children with severe malaria.

We thank the children and caregivers who participated in this study and the study team for their dedicated effort in collecting the data.

CC

community children

CM

cerebral malaria

SMA

severe malarial anemia