Project Lead(s): Lazarus Odeny
Issue
Although tuberculosis (TB) treatment is free in Kenya, 64% of TB patients do not seek treatment because they are not aware they are infected.
In addition, only 9% of those seeking care are diagnosed at their first health facility visit, with the vast majority being diagnosed after several visits to hospitals and non-health facilities.
Solution
Implemented in Kenya, this study evaluated the effectiveness of a mobile phone-based application to predict the probability of a patient having TB and to advise on the next steps, using proven algorithms.
The feasibility of this tool was tested by augmenting current practices with the mobile app as a TB screening aid and comparing agreement between the mobile app and the clinical diagnosis.
Outcome
The results showed use of the mobile app provided a more accurate diagnosis of TB than when the app was not used.
In a population of 1,045 HIV-infected participants, the study found the agreement between the clinicians and mobile app on TB status was generally poor.
Clinicians tended to rule out TB more often than the mobile app results (85% versus 61%), while the algorithm used by the mobile app recommended further testing more frequently than clinicians (39% against 5%).
Overall, the mobile app identified two more cases than the clinicians out of the six infected participants, with the diagnosis being confirmed by GeneXpert.
The app has the potential to be administered by non-specialists, thereby improving TB case detection and reducing TB morbidity and mortality.
A manuscript detailing the work of the project will be prepared for publication.