Background: In 2018, Uganda started only 65% of persons with incident tuberculosis on treatment. Pretreatment loss to follow up is an important contributor to suboptimal treatment coverage. We aimed to describe the patient and health facility-level characteristics associated with pretreatment loss to follow up among patients diagnosed with pulmonary tuberculosis at public health facilities in Uganda.

Methods: At ten public health facilities, laboratory register data was used to identify patients aged ≥ 15 years who had a positive Xpert®MTB/RIF test. Initiation on TB treatment was ascertained using the clinical register. Factors associated with not being initiated on TB treatment within two weeks of diagnosis were examined using a multilevel logistic regression model accounting for clustering by health facility.

Results: From January to June 2018, 510 patients (61.2% male and 31.5% HIV co-infected) were diagnosed with tuberculosis. One hundred (19.6%) were not initiated on TB treatment within 2 weeks of diagnosis. Not having a phone number recorded in the clinic registers (aOR 7.93, 95%CI 3.93-13.05); being HIV-infected (aOR 1.83; 95% CI: 1.09-3.26) and receiving care from a high volume health facility performing more than 12 Xpert tests per day (aOR 4.37, 95%CI 1.69-11.29) and were significantly associated with pretreatment loss to follow up.

Conclusion: In public health facilities in Uganda, we found a high rate of pretreatment loss to follow up especially among TBHIV co-infected patients diagnosed at high volume health facilities. Interventions to improve the efficiency of Xpert® MTB/RIF testing, including monitoring of the TB care cascade should be developed and implemented.