The development of a screening tool to evaluate infants who are HIV positive

HIV/AIDS continues to be one of the greatest health challenges which South Africa faces. The epidemic in children is closely linked to that in women, the prevalence of which continues to grow according to antenatal statistics from the South African Department of Health (DOH). HIV is known to invade the central nervous system at the time of infection, and causes widespead damage. In children, this leads to a well-researched condition known as HIV encephalopathy, which affects all areas of neurodevelopment. The effects of timely initiation of antiretroviral therapy on alleviating the impact of encephalopathy have been well described.

Neurodevelopmental delay is a stage four disease indicator according to the World Health Organisation (WHO), and therefore is a criterion for initiation of Highly Active Antiretroviral Therapy (HAART). HAART is often only administered according to the virologic and immunologic status of a child, as standardised neurodevelopmental assessment tools are not widely available in South African clinics. When HAART initiation is dependent on immunologic status, it is often too late to prevent encephalopathy. To date, the only means of prevention of this condition is early initiation of HAART, which has not been widely available in South Africa. Stringent guidelines for the commencement of this therapy according to the WHO, and the South African Department of Health (DOH) have had to be followed, leading to late initiation of HAART, and widespread central nervous system encephalopathy.

Studies which have been carried out in South African clinics have demonstrated the high prevalence of this condition. Once there is evidence of encephalopathy, children should be referred for assessments in all facets of development, and where necessary, for rehabilitation. A standardised developmental screening tool which is suitable for use in a developing country is therefore necessary in order to screen for neurodevelopmental delays to allow for further assessment and referral to rehabilitation services, as well as providing an additional assessment criterion for initiation of HAART. Paediatric HIV clinics in developing countries are understaffed, and children may be seen by junior staff or screened by nurses due to the high numbers of clinic attendees. This often results in neurodevelopment being inadequately assessed and children are therefore not referred for intervention services. A standardised screening tool, could be administered by clinic staff in order to ensure correct and timely referral of children for further assessment and intervention is therefore necessary. This is of importance both locally and internationally where a screening tool, which has been developed specifically for this purpose, does not exist.

The aim of this study was therefore to evaluate the agreement between the Bayley-III Screening Test and the Bayley Scales of Infant Development (3rd version) in a population of HIV positive infants in order to evaluate its appropriateness for use in South Africa. The Bayley Scales of Infant Development have long been considered the ‘gold standard’ in infant developmental assessment, which is why this tool was chosen to evaluate the Bayley-III Screening Test against. The developmental scores in each facet (cognitive, motor or language) were evaluated to determine which should be included in an assessment tool for this population. Further objectives for the study were to adapt the screening tool to the needs of the population, or to develop a new screening tool shouldthe Bayley-III Screening Test not prove suitable for use in this population.

In order to meet the aims and objectives, a cross-sectional study was conducted where 112 HIV positive infants between the ages of six and eighteen months were assessed using the Bayley-III Screening Test and the Bayley Scales of Infant Development (3rd version) (BSID III). The infants were stratified into four age groups namely 6-8 months, 9-12 months, 13-16 months, and 17-18 months. Children were recruited from Harriet Shezi Children’s Clinic at Chris Hani Baragwanath Hospital in Soweto.

The agreement between the Bayley-III Screening Test and the Bayley Scales of Infant Development (3rd version) was analysed using Kappa, for the overall group, and for each age group. Overall agreement between the tools was as follows: K=0.58 for the Cognitive facet, K=0.82 for the Expressive Communication facet, K=0.76 for the Receptive Communication facet, K=0.44 for the Fine Motor facet and K=0.57 for the Gross Motor facet. These values indicate that the Bayley-III Screening Test is therefore not acceptable for clinical use, as excellent agreement (k≥0.75) in all facets would be necessary for this purpose.

A new screening tool therefore had to be developed. The infant’s developmental scores from the BSID III were analysed to determine which facets of development were most severely affected, and therefore which facets should be included in a new screening tool. Gross motor function was demonstrated to be the area which was most severely affected, followed by cognitive function. A gross motor screening tool would therefore be suitable for use in this population, as no equipment would be necessary. Gross motor development is the most universally similar aspect of development, which is not completely dependent on cultural or socioeconomic factors which often have an influence on language and cognitive development.

Item selection from the BSID III was undertaken to determine which items should be included in a brief screening tool. In each of the four age groups, item selection occurred as follows: Two items which discriminated the At-Risk, from Emerging and Competent groups (less than 20% in the At-Risk group, and 100% in the other groups) were selected. Two items, which discriminated between children in the ‘Emerging’ and ‘Competent’ categories on the BSID III were selected (0-5% of children who were At-Risk obtained credit, 30-50% of the Emerging group obtained credit, and 100% of the Competent group obtained credit). Lastly, two items were selected which discriminated the Competent group from the other two groups (100% or as high as possible in the Competent group, and 0% in the other groups).

The new gross motor screening tool was assembled using the selected items, scoring was allocated, and it was tested against the scores obtained on the Gross Motor facet of the BSID III for the initial 112 infants. Agreement between the tools was analysed using Kappa, and refinements were made according to the discrepancies. This was done three times, until the Kappa value revealed excellent agreement between the tools (k = 0.87). A panel of experts was then invited to examine the new gross motor screening tool, and to comment on it, and further adjustments were made accordingly.

Preliminary concurrent validity testing of the new gross motor screening tool was then carried out against the Gross Motor facet of the BSID III on 60 children, who were recruited from the Harriet Shezi Children’s Clinic at Chris Hani Baragwanath Hospital in Soweto. Statistical analysis revealed that the agreement between the BSID III and the new screening tool was excellent (k=0.85). The diagnostic properties of the new gross motor screening tool were as follows: sensitivity 97.4%, specificity 85.7%, positive predictive value 92.7%, and negative predictive value 94.7%. These values indicate that the statistical properties of the tool are excellent, and the tool will not be predisposed to underreferrals or over-referrals. Preliminary reliability testing was carried out on 15 children for test-retest/intrarater reliability and 15 children for interrater reliability. Interrater, test-retest and intrarater reliability were excellent (r=1, r=0.98, r=0.98 respectively). Further testing of reliablity and validity should be undertaken in order to establish these properties, and standardisation should also be carried out on healthy children. Given the need for an assessment tool of this nature in South Africa and other developing countries, and the statistical properties thus far, the tool may be used clinically for the purposes for which it was developed.