Reliability of the Luganda version of the Child Behaviour Checklist in measuring behavioural problems after cerebral malaria

Mental and behavioural problems in children and adolescents are common with 10-20% of the world's children estimated to have one or more such problems [1]. These high rates of behavioural problems in children and adolescents are attributed to several factors including poverty, armed conflict, infectious diseases like malaria and HIV/AIDS, unfavourable family environments and substance abuse [1–5]. In some cases of exposure to war trauma, the percentage of those with symptoms of clinical importance can range from 50 to 90% [2].

In Uganda, malaria accounts for 30% of paediatric admissions with approximately 6500 admissions annually at Mulago Hospital, the National Referral Hospital [6]. Cerebral malaria accounts for 8.2% of these malaria cases with a mortality of 17% [6]. Despite its low prevalence, cerebral malaria is one of the major causes of neurodevelopmental difficulties [7, 8] with several studies documenting cognitive and neurologic deficits in survivors, with 14% to 26% having cognitive deficits [9–11]. There is however little evidence of behavioural problems resulting from cerebral malaria with only one study describing the behavioural problems [12]. The other studies that have assessed the behavioural outcomes of cerebral malaria are either cases studies [13, 14] with limited ability to generalise, or do not describe the behavioural functions affected [15, 16]. Documentation of the behavioural problems after cerebral malaria can highlight what problems to assess for in survivors and to help develop appropriate interventions for affected children. Despite the lack of documentation of behavioural problems, the burden of cerebral malaria on children's development is evident with estimates putting the number of children with cognitive and behavioural problems after cerebral malaria at over 200,000 per year [11, 16].

In order to deal with the increasing burden of mental and behavioural problems, WHO has emphasised the need to manage and treat patients in primary health centres [1]. In low and middle-income countries where there is a shortage of trained mental health professionals [3, 4, 17], there may be difficulties in correctly assessing, diagnosing and treating mental and behavioural problems. In Uganda for example, 36% of physically ill adult patients seeking treatment at primary health care facilities have a current major depressive episode [18]. Due to the large patient load and health workers not suspecting mental illness as the underlying problem [19], patients may only be treated for the presenting problem (e.g., backaches, fevers), leaving the psychiatric problem causing these complaints untreated.

One way to overcome the difficulty in patient management is to use psychiatric rating scales that are validated in the target population and summarise the patients' complaints into a probable psychiatric diagnosis. These scales can be completed by a child's caregiver as a screening tool and can also track treatment progress. These rating scales also ensure appropriate investigations and treatment, thus saving time and limited resources in suspected psychiatric cases. In non-clinical settings, ratings scales for children have been used in assessing the functioning of immigrant children [20].

One such childhood rating scale is the Child Behaviour Checklist (CBCL), a widely used psychiatric rating scale for children and adolescents [21]. Research with the CBCL has demonstrated its sound reliability and validity for the scale in different cultures [22–24], and cross-cultural norms have been established [20]. A recent validation of the school age CBCL in 30 societies showed that its eight syndrome scales have a good fit when tested separately in these societies and that they are a reasonable tool for conceptualising children's emotional and behavioral difficulties in those societies [22].

Despite its sound reliability and validity in a number of cultures, the CBCL is yet to be validated in Uganda. We present a study carried out in Ugandan children to examine the reliability of the Luganda version of the CBCL and to document the behavioural problems after cerebral malaria as measured by the CBCL.

This study was carried out to examine the reliability of the Luganda version of the CBCL and document behavioural problems in a sample of Ugandan children with a history of cerebral malaria. Test-retest reliabilities and internal consistencies for the three main scales (Internalising Problems, Externalising Problems and Total Problems) ranged from 0.64 to 0.83. Other studies of the CBCL have produced reliability coefficients higher than what we present here. For example in Mexican children, the test-retest reliability of the CBCL's scales ranged from 0.69 to 0.86 and internal consistency from 0.69 to 0.96 [29]. In China the test-retest reliability of the Internalising, Externalising, Attention and Total Problems was between 0.79 to 0.84 [30]. Larger studies are needed to confirm the reliability of the CBCL in Ugandan children as the current reliabilities are lower than those reported from similar studies elsewhere, possibly owing to the long test-retest interval (9 weeks) and small sample (N = 22).

Exceedingly low test-retest and internal reliabilities were observed for Thought Problems which contains items dealing the with child's covert behaviour (eg Hears things, Has strange ideas). Parents may find it difficult to be consistent in rating these covert behaviours which are not easily observed compared to the overt behaviours. In this study, low reliabilities (below 0.4) were mostly found in the covert behaviours while high reliabilities (above 0.7) were mainly in the overt behaviours.

Depressive symptoms and Thought Problems (a combination of depressive, obsessive-compulsive, hallucinatory and sleep problems) were commonly reported in the current study, similar to an earlier study among Ghanaian adults that found higher scores of anxiety and depressive symptoms among those with a history of malaria compared to controls who had never had the illness [31]. However the Ghanaian study was carried out in adults who had uncomplicated malaria while the current study is in children who had complicated malaria. The similarity in findings may thus be a coincidence subject to further investigation. Cerebral malaria has been associated with increased behavioural problems in African children [12, 15, 16]. It is therefore likely that the frequency of behavioural problems reported here are higher than in the general population of Ugandan children of similar age with no history of cerebral malaria or other central nervous system infection or disorder.

Ugandan children had higher scores than Group 3 children for most of the behavioural problems except attention problems (both Attention Problems in the Syndrome Scale and ADH Problems in the DSM Scale). This observation of more behavioural problems is not a surprise as prior studies have associated cerebral malaria with behavioural problems in children [13–16]. Consistent with the low frequency of attention problems, Cohen's d for both attention problems was below zero indicating that the Group 3 children had higher scores on attention problems than the Ugandan children. However caution is needed when interpreting our findings due to lack of Ugandan norms for the present study and the limited sample size. Lack of Ugandan norms makes it difficult to confidently conclude whether the observed problems are due to cerebral malaria, environmental characteristics, problems with the translation or a combination of factors.

A high frequency of attention deficits was earlier observed in these children [11] which is contrary to the present findings. This could be attributed to the different methods of assessing attention in the studies. John and colleagues [11] used a computerised measure of attention based on the child's ability to respond to the target stimuli [32] while the present study used parents' endorsement of behaviour depicting attention problems. The computerised method is a measure of sustained attention measuring the child's reaction time in milliseconds and the ability to discriminate between a target and non target [32]. Sustained attention is best measured by computerised tests [33] as used in the earlier study documenting attention problems [11]. The CBCL on the other hand gives a broad description of the child's behaviour and may not accurately measure sustained attention like the computerised tests. When the CBCL attention scores of children in the current study who were earlier categorised as having attention impairment [11] were compared to those not impaired, there was no significant difference in the scores (data not shown) which may suggest that these two tests may not measure the same kind of attention. We cannot fully explain this finding of lower attention problems measured by the CBCL, further studies are needed before conclusive statements can be made.

This under reporting of attention problems by the Ugandan parents may partly explain why the Internalising and Externalising Scales had a higher frequency of children with problems of clinical importance compared to the Total Problems Scale. The Ugandan parents consistently rated highly the behaviours making up the Internalising and Externalising Scales than the Group 3 parents but this was reversed for the attention problems.

This study was not able to evaluate the CBCL's concurrent validity since no other measures of child behaviour were administered due to time constraints. In addition, our sample was not recruited from a psychiatric setting where their clinical diagnoses could be compared with the CBCL scores to evaluate its sensitivity and specificity. A further limitation of our study was the limited sample size of 64 children which gave a small number of 22 for the test-retest reliability calculation. With the nine week interval between the baseline and post-intervention testing, it can be argued that important changes in behaviour can take place. This we believe was not the case in this study as comparison of the two test scores of the control group showed no significant changes [26].

The Luganda version of the CBCL has moderate reliability and can be used in behavioural assessment. Depressive and thought problems are likely behavioural outcomes of cerebral malaria. Future studies are needed to document these problems and their course, develop country norms for the CBCL, evaluate its validity in a clinical sample so as to determine its sensitivity and specificity and provide a broader range of responses.