This was a prospective study with a sample of 259 children (171 with severe anaemia and 88 community controls) aged 6 - 42 months conducted between August 2016 and June 2017 at Lira Regional Referral Hospital in Northern Uganda. Participants with SA were in-patients of an implementation research study on management and outcomes of severe anaemia in Ugandan children where SA was defined as Hb < 5g/dL (27, 28).
Parents of children with SA were informed about the need for healthy control children in the study age range to generate a control group. The non-febrile healthy community children (CC) were siblings from their nuclear or extended family, within the household compound area or neighbours (playmate) residing in the same locality (<1 km) of the enrolled children with SA who had been volunteered by the parents after invitation to participate in the study. The utilization of siblings or neighbours was to minimize differences between groups in relation to family background, household environment, socio-environment/ neighbourhood, culture and linguistics, number of siblings in the household, number of meals per day.
The CC were aged between 6 and 42 months, and healthy. They were examined at the time of enrolment to ensure that they did not have clinical pallor on clinical examination of the palms, nail-beds, tongue and conjunctivae (as a method of SA assessment) or a history of hospitalization for SA 6 months prior to enrolment. Assessment of SA using clinical signs such as pallor is a commonly used method (especially in our settings with poor or absent diagnostic facilities) for diagnosis of severe anaemia in children with sensitivity and specificity ranging from 53% to 96% (23, 29). The children in the SA and CC groups were otherwise identical and sampled at the same time to minimise any biases.
Clinical and Demographic assessment
Social economic status (SES) and demographic characteristics were obtained using a questionnaire of material possessions assessing housing quality, cooking resources, water accessibility and the presence of key amenities (radio, shoes for subject, mobile phone, poultry) in which lower SES scores have been associated with worse cognitive functioning in healthy Ugandan paediatric population under 5 years of age (30). Nutritional status was obtained by comparing anthropometric indicators (weight-for-age (WAZ), height for age (HAZ) and weight for height (WHZ) using the WHO anthro survey analyser software for children under 5 years of age to generate the standardized z-scores (31). We followed internationally recognised cut-offs to consider children whose HAZ, WAZ, or WHZ fall more than two SDs below the international mean to be stunted, underweight or wasted, respectively (32).
Behavioural assessment
Behavioural assessment was done using the Bayley Scales of Infant and Toddler Development, 3rd edition (Bayley-III). It is one of the most commonly adapted comprehensive psychometric assessment tools used in research, in clinical practice, and to evaluate interventions as it assesses several developmental domains as a measure of early global development among very young children (24, 25, 33). The interviews with the caregivers were conducted in a quiet room with minimal distractions at the hospital. For uniformity and language concerns, trained assessors with Bachelor’s degrees in Psychology and fluent in Langi (the local dialect of the study setting) administered the test to the child’s primary caregiver. The first author (Health psychologist) supervised the administration of interview questionnaire and reviewed the record forms at the end of each week for completeness to ensure data quality. The assessors were periodically given review trainings by a psychology graduate independent of the study and with expertise in Bayley-III. For consistency and to ensure data quality, the team (two interviewers and first author) met regularly to review, feedback and discuss the data.
Assessments were conducted 14 days post discharge for the caregivers of the children with SA and at enrolment for the CCs or when appropriate for the caregiver to return to the hospital for assessment. We interviewed the primary caregiver of each child using the social–emotional and adaptive behaviour scales of the Bayley–III (24, 25). Bayley-III has not been validated for our setting however has been adapted for appropriate use among Ugandan children in previous studies and used for rural populations similar to our study sample and setting (34-36).
Majority of the caregivers were mothers, familiar with the child and could provide meaningful, accurate and complete response ratings of their child’s personal, adaptive and social skills necessary for daily living. The social-emotional scale assesses emotional and social development as well as sensory processing that influences a child’s emotional responses based on the Greenspan Social-Emotional Growth Chart (37). The scale provides a general indication of a child’s level of social-emotional development and presence or absence of sensory processing difficulties (38). The scale assesses the child’s functional, social and emotional milestones namely; self-regulation and interest in the world, relationship engagement, emotional engagement in an interactive and purposeful manner, communication with interactive emotional gestures, problem solving through interactive emotional gestures, communicating intentions and feelings using symbols and ideas, using symbols to express intentions, wishes or feelings more than basic needs, creating logical bridges between ideas and emotions (24, 37, 38).
Adaptive behaviour is a collection of skills (conceptual, social, and practical) for effective functioning that concern the way individuals meet their personal needs while meeting their demands in their environment (39, 40). The adaptive behaviour scale is derived from items for children 0-5 years of the Parent/Primary Caregiver Form of the Adaptive Behaviour Assessment Scale – Second Edition - ABAS-II (41). The scale assesses ten areas categorized in three broader domains: (1) conceptual (communication, functional academics, and self-direction); (2) social (social and leisure); and (3) practical (self-care, home or school living, community use, health and safety) (39, 42). A summation of the ten sub-scales composite scores was obtained to generate an overall adaptive behaviour score also known as the General Adaptive Composite (GAC) score.
Statistical methods
Data were entered into Filemaker 11.0v3 (FileMaker Inc. US) database, and exported into IBM SPSS 23 (IBM Corp., Armonk, N.Y., USA) for statistical analysis. For this study, raw scores for each scale were converted into an age and sex-specific standardized z-score, based on the scores of healthy community children (CC, n=88). The z-scores were computed as (actual score – mean score for a child’s sex and age)/SD, where the mean score for a child’s sex and age and SD were computed by fitting a linear regression model to data for all CC children (18). Z-scores have a mean of 0 and SD 1 in the CC reference population. Multiple linear regression was used to compare z-scores on all the scales between the two groups after adjusting for weight-for-age z-score, social economic status, mother’s education, father’s education and father’s employment. We adjusted for multiple testing for the adaptive subscales using the Hommel’s procedure (43) and p<0.05 was statistically significant.
Ethics
Approvals for this study were obtained from Makerere University School of Medicine Research Ethics Committee (REC Ref: 2015-045), Uganda National Council for Science and Technology (Ref: HS 2017) and the Lira Regional Referral Hospital administration. Participation in the study was voluntary and the caregivers of the study participants who took part in the study provided written informed consent.