Correlates of default from anti-tuberculosis treatment: a case study using Kenya's electronic data system

In 2012, the World Health Organization estimated that there were 120,000 new cases and 9500 deaths due to tuberculosis (TB) in Kenya. Almost a quarter of the cases were not detected, and the treatment of 4% of notified cases ended in default.

To identify the determinants of anti-tuberculosis treatment default.

Data from 2012 and 2013 were retrieved from a national case-based electronic data recording system. A comparison was made between new pulmonary TB patients for whom treatment was interrupted vs. those who successfully completed treatment.

A total of 106,824 cases were assessed. Human immunodeficiency virus infection was the single most influential risk factor for default (aOR 2.7). More than 94% of patients received family-based directly observed treatment (DOT) and were more likely to default than patients who received DOT from health care workers (aOR 2.0). Caloric nutritional support was associated with lower default rates (aOR 0.89). Males were more likely to default than females (aOR 1.6). Patients cared for in the private sector were less likely to default than those in the public sector (aOR 0.86).

Understanding the factors contributing to default can guide future program improvements and serve as a proxy to understanding the factors that constrain access to care among undetected cases.