Provision of research methods support for neglected tropical diseases’ research projects at eastern and southern African centre of international parasite control, Kenya medical research institute, Nairobi

The availability of a large variety of elementary and complex statistical methods and the quick pace of change and development in statistics makes most researchers' statistical knowledge insufficient for them to be independent. It is unrealistic to expect researchers to solve the statistically-related challenges that arise in the course of their investigations on their own. Consequently, support from RM specialists is expected to improve the quality of research outputs by minimizing studies with unsound and unreliable outputs. Reported in this dissertation are the research methodology tasks undertaken at ESACIPAC (KEMRI) during internship period of one year. The tasks included participation in research by offering consultancy services, data management, statistical analysis and reporting. The consultancies were diverse as they ranged from technical support in design of projects to reporting of research findings. Reported in this document are consultations involving statistical inputs in drafting sections of a proposal, sample size computations for studies, statistical support in reporting of findings and computation of confidence intervals. Data management was conducted for the study; ―Adverse effects associated with mass drug administration of praziquantel and albendazole …‖ Double data entry was done in Excel. The spreadsheets were constructed with measures for minimizing data entry errors while maximizing processing efficiency, that is, frozen top rows, drop down lists and validation checks. On data entry completion, auditing was done, and then comparisons using Epi Info. Lastly, data were exported to SPSS and coded further in preparation for analysis. Data were analyzed for the study; ―Factors contributing to re-infection with S. mansoni among primary school children …‖ The sample was described using frequencies, rates and proportions for categorical variables while appropriate measures of central tendency and dispersion were used for continous variables. To identify factors predictive of re-infection, crude associations were explored using bivariate analyses and then logistic regression model constructed. Adjusted odds ratios were computed for the significant factors. The final model showed a reasonable fit when assessed for goodness-of-fit using Hosmer-Lemeshow test (p=0.270). Finally, reporting of the findings was done as a collaborative effort between the investigator and the research methods intern.