52 Demographic influences on test performance may not be universal: considerations from a cross-country comparison of South Africa and Zimbabwe

Abstract

Objective:

Appropriate adjustments to normative data for neuropsychological (NP) tests are imperative for their equitable use in brain health practices. Age and education are known to be strong predictors of test performance. In settings where validated tests are not available, common practice has been to adapt and apply them in similar fashion as settings where they were developed. However, demographic adjustments cannot be assumed de facto to be universal in their strength and domain associations. For example, South Africa (SA) and Zimbabwe are neighboring countries with some similarities in their demographic makeup, but with vastly different sociopolitical trajectories- Zimbabwe was colonially occupied until 1980 and SA was oppressed under Apartheid until 1994- which have impacted access to and quality of education by severely limiting educational opportunities for native citizens. The present study explored whether the direction and strength of relationships between age and education on NP test performance were similar or not between SA and Zimbabwe adults living with and without HIV.

Participants and Methods:

Data was extracted from two IRB-approved studies in SA and Zimbabwe with similar inclusion and exclusion criteria. The SA sample (n=214) was comprised of 56% females, 48% HIV-positive adults, mean age of 34 years, and a nine-year range in education (3-14 years). The Zimbabwe sample (n=212) was comprised of 68% females, 67% HIV-positive adults, mean age of 36 years, and a thirteen-year range in education (7-20 years). Participants completed NeuroScreen, a tablet-based battery of 12 brief NP tests adapted for indigenous SA and Zimbabwe languages. The two study samples were analyzed separately. Zero order correlations between each of the tests and age and gender were conducted to determine the influence of the demographic variables. Relationships with moderate correlations (r>0.3) in both samples were further analyzed using univariate ANOVA to examine the main effects and interactions of age and education

Results:

Overall, there was a similar pattern of results across samples, with nine tests showing no-to-low associative relationships with age and education respectively. Moderate, significant relationships were found between age, education and three tests of processing speed (Visual Discrimination A, Visual discrimination B, and Number Speed) in both samples. Age and education had different effects on Visual discrimination A across samples with a significant main effect for age but not education in SA [F(40,83)=3.060, p<0.01], whilst Zimbabwe had a significant main effect for education but not age [F(10,87)=4.541, p<0.01]. Visual Discrimination B and Number Speed showed significant main effects for both variables in both samples. However, there was a significant interaction for both tests in Zimbabwe only.

Conclusions:

The current study is novel in its exploration of country-specific relationships between NP test performance and demographic factors in settings where assessment science is emergent. Results demonstrate the presence of differential relationships between demographic variables on test performance which raises questions about the source of these differences. One important potential source is the socio-cultural context of each country and the intersection of demographic factors in these contexts. Further research is required to explore these considerations.