58 Chapter 3 The nurses in the TB unit will be oriented on how to use the Glucometer as a point-ofcare test by a trained laboratory technician from the health facility’s laboratory using a standardized guide (Supplementary file 5). The Glucometer will initially be calibrated by the health facility’s laboratory technician at baseline, at the end of month 2 and at the end of month 4. This is to ensure the quality of the results produced by the Glucometer is standardized and possible calibration errors are identified and rectified. Additional baseline patient sociodemographic information – educational status, marital status, occupation, smoking and alcohol status which is not routinely collected will be obtained for this study. Staff at the TB clinic will be oriented on the study and one healthcare worker at each of the 12 health facilities will be trained on how to approach and consent new patients into the study. Facilities will be visited monthly with a follow-up call weekly for updates. Statistical Methods and Analysis The baseline characteristics of variables in the patients’ dataset will be presented in a table. The prevalence of DM or impaired glucose will be determined based on the number of patients with DM or impaired glucose at baseline and during the treatment period expressed as a proportion of all the patients treated in the same period. This will be determined overall and by the type of TB disease (drug-sensitive TB (DSTB), rifampicinresistant TB (RRTB), DRTB or extra drug-resistant TB (XDR)). The occurrence of abnormal glucose during treatment will be determined based on the number of patients who had normal values at baseline but developed abnormal values during treatment or at the end of treatment. This will be determined overall, and by type of TB disease with additional analysis to estimate the mean and median time between TB diagnosis and identification of abnormal measurements. A logistic regression (or mixed effect model for repeated data) will be used to predict the occurrence of DM or hyperglycemia. Statistical tests will be significant if p<0.05. Different sub-analysis, comparative and sensitivity analyses will be done to identify possible interactions which may exist between the different patient characteristics (e.g., age, sex, HIV status) and hyperglycemia e.g., testing to ascertain if there is an association between timing of culture conversion and blood glucose. The proposed statistical methods for the different research questions are presented in Table 1. Qualitative analysis in form of analysis of identified codes will be done to identify factors that hinder the care of diabetics receiving TB treatment. Recommendations for improvement will be coded and similar codes will be analyzed and presented.
RkJQdWJsaXNoZXIy MTk4NDMw