Date of Award

January 2018

Document Type


Degree Name

Medical Doctor (MD)



First Advisor

Jeremy I. Schwartz


Essential Medicines (EM) for non-communicable diseases (NCD) are often unavailable to patients in Uganda. We sought to determine the gap between prescribed and dispensed medicines for the treatment of NCD - specifically, diabetes mellitus (DM) and cardiovascular disease (CVD) - in Ugandan public healthcare facilities. We conducted a cross-sectional study in which systematic sampling was used to collect data from patients attending outpatient NCD clinics at 20 national, regional and general higher level public healthcare facilities throughout the country. Convenience sampling was used to obtain data from prescribers and dispensers of NCD medicines at the same facilities. Data were collected using structured questionnaires. The primary outcome was the percentage of prescribed doses that were not dispensed. We analyzed data from 637 patient participants, 24 prescribers and 28 dispensers. 454 (71.3%) patient participants were female, 107 (16.8%) had no formal education, 259 (40.7%) were unemployed and 391 (62.3%) earned below the poverty line. The mean number of prescribed medicines per encounter was 2.3 (SD 1.1), of which only 1.4 (SD 0.9) were dispensed. The most commonly prescribed medicines for DM were metformin 291 (48.6%) and glibenclamide 161 (26.9%) while for CVD they were nifedipine 188 (21.7%), bendroflumethiazide 134 (15.4%). Of the 1467 prescriptions analyzed, 1157 (78.9%) were generic names and 1412 (96.3%) were on the Essential Medicines List (EML), which is less than the World Health Organization (WHO) standard of 100% for prescription of generic names and medicines on the EML. 1151 (78.5%) prescriptions were adequately written by prescribers, and 802 (89.6%) dispensed medicines were adequately labelled. Overall, there were 82,591 total prescribed doses and 35,290.5 dispensed doses, meaning that 57.3% of prescribed doses were not dispensed. The percentage of non-dispensed doses varied significantly by participant travel time (p=0.04), facility (p<0.001) and facility type (p=0.001). Additionally, the proportion of non-dispensed doses varied with the disease (CVD 67.5% v. DM 46.2%, p<0.001). 535 (84.0%) patient participants did not receive all their prescribed doses. Of the 442 patient participants who reported not receiving all of their medicines, 317 (71.7%) reported this being due to low pharmacy stock. Upon follow-up, 90 (35.6%) of 253 patients contacted had not obtained their non-dispensed doses elsewhere. There is a marked discrepancy between prescribed and dispensed doses for medicines that treat NCD at public healthcare facilities in Uganda. The discrepancy differs by disease and appears to be driven by availability of medicines. Most patients do not close this gap by purchasing medicines at private pharmacies, suggesting that under-treatment of these chronic conditions is widespread among patients already linked to care. As noted above, this study found a number of indicators of irrational drug use including prescription of brand name medicines, prescription of medicines not included on the EML and inadequately labeling of prescriptions and dispensed medicines. Unavailability of medicines in facilities has been previously documented but this study describes the unavailability of medicines to patients. This offers a different statistical target for policy makers when determining policy to increase availability of NCD medicines to patients.


This thesis is restricted to Yale network users only. It will be made publicly available on 06/25/2100