Bayesian Meta-analysis of the effectiveness of the Implementation Science evidence in improving health outcome for adolescent patients accessing ART in sub Saharan Africa

Isaac Fwemba^1*Samuel Iddi²Lehana Thabane³Alfred Yawson¹Jacques L Tamuzi^4,5*Peter Nyasulu⁵Samuel Bosomprah¹

• ¹Department of Biostatistics, School of Public Health, College of Health Sciences, University of Ghana, Ghana, Greater Accra, Ghana
• ²Department of Epidemiology and Biostatistics, School of Public Health, University of Zambia, Lusaka, Zambia
• ³Department of Health Research Methods, Evidence & Impact, Faculty of Health Sciences, McMaster University, Hamilton, Ontario, Canada
• ⁴Stellenbosch University, Stellenbosch, South Africa
• ⁵Division of Epidemiology and Biostatistics, Department of Global Health, Faculty of Medicine and Health Sciences, Stellenbosch University, Stellenbosch, South Africa

Background: Implementation Research (IR) studies and clinical trials have yielded conflicting results regarding improving treatment outcomes especially for adolescent patients. We performed a Bayesian random effect meta-analysis to evaluate the effectiveness of current IR interventions on improving retention in care and reducing viral suppression in HIV infected adolescents. Method: A comprehensive search from 1st January, 2000 through to 31 st December, 2020. A Bayesian random-effects meta-analysis was undertaken for historical evidence using adolescent interventions from other parts of the world, excluding SSA, and adult historical evidence using adult-derived interventions within SSA. Bayesian Copas random effect model was used to control publication bias and study variations. Power priors were used to add in the analysis in a weighted manner of the information contained in historical data. Bayesian metaanalysis is suited for this study since it allows us to directly include past assessments from several intervention studies into the pooled intervention data.The pooled results from the twelve (12) studies covering eight countries in Africa with a population of 19,223 adolescent patients showed significantly superior retention effects in adolescent specialized interventions compared to the standard of care settings; OR=3.87 (95% Crl: 0.94,10.82). When 100% of data from 8 observational studies covering adolescent treatment outcomes were added to the analysis, the resultant OR was; OR=3.02 (95%Crl: 1.01, 6.92). However, when 100% of the historical data from adult RCTs was used, the retention effect reduced to OR 1.24 (95% Crl: 1.03, 1.48). Whether adolescent historical data was used or adult RCT, the associated posterior probability of benefiting from the intervention remained almost 1. There was no difference between the standard of care and specialized adolescent care in terms of virological suppression effect; OR=1.27 (95% Crl: 0.57, 2.32). However, the adolescent intervention recorded a superior overall retention rate of 59.7% compared to 52.1% in the standard of care.The current adolescent interventions are effective in improving adolescent retention rates in HIV care. Evidence from adult interventions resulted in a reduced retention effect, suggesting that adult promising interventions may improve adolescent treatment outcomes but may need to be modified.