A Clinical Risk Score for Early Management of TB in Uganda (PredicTB)

PredicTB: Validating a Clinical Risk Score for Early Management of Tuberculosis in Ugandan Primary Health Clinics

Although curative treatment exists, tuberculosis (TB) remains the leading cause of infectious mortality worldwide - often because people seek care for TB symptoms in highly resource-constrained clinics that cannot provide same-day diagnostic testing. The research team has developed an easy-to-use clinical risk score that, if implemented in these settings, might help clinicians identify patients at high risk for TB and thereby start treatment for those patients on the same day. This study will investigate the effectiveness and implementation of this score in four peri-urban clinics in Uganda, providing critical pragmatic data to inform (or halt) the design of a definitive large-scale cluster randomized trial.

An estimated 1.5 million people die of tuberculosis (TB) every year. Many of these are people who seek care in under-resourced clinics (for example, in rural areas or informal settlements) where same-day TB diagnosis is not available. These patients are often unable to return promptly to receive their results and start treatment, resulting in ongoing disease transmission and often death. If TB treatment could be started on the same day as these patients initially seek care, substantial mortality and transmission could be averted. The research team has developed and validated a clinical risk score ("PredicTB") for adult pulmonary TB that could aid in clinical decision-making. This risk score ranges from 1-10, can be calculated by hand in under a minute using readily available clinical data (e.g., age, sex, self-reported HIV status), and has sufficiently high accuracy to inform decisions about same-day empiric treatment initiation while confirmatory test results are pending. Same-day treatment initiation improves patient outcomes for other infectious diseases (for example, sexually transmitted diseases including HIV), and this novel clinical risk score holds similar promise for TB, the leading cause of infectious mortality worldwide. However, before conducting a large-scale cluster randomized trial to evaluate whether this score could improve patient-important outcomes, it is critical to first generate evidence that this score could be effective and be implemented in the most-resource-limited settings for which it is intended. The research team proposes a type 2 hybrid effectiveness-implementation evaluation of the PredicTB clinical risk score in four peri-urban clinics in Uganda, with an additional four clinics serving as a comparison group. The Specific Aims are to evaluate the effectiveness of PredicTB on clinical outcomes including rapid treatment initiation, TB mortality, and loss to care (Aim 1); to evaluate the implementation of PredicTB in terms of reach, adoption, implementation, and maintenance (Aim 2); and the project the long-term impact and cost-effectiveness of PredicTB implementation (Aim 3). The primary outcome is the increase in the proportion of patients with microbiologically confirmed TB who start treatment within seven days of initial presentation. To accomplish these aims, the research team will adopt a highly pragmatic study design in which the research team train clinicians in the use of the PredicTB score and perform quarterly site visits but otherwise minimize contact between study staff and treating clinicians. This will enable the research team to evaluate whether implementation of PredicTB is likely to impact clinical decision-making and patient outcomes under actual field settings. If successful, this evaluation will provide critical data to justify (or halt) the conduct of a large-scale pragmatic clinical trial - not only will it generate preliminary evidence of effectiveness, but it will also inform appropriate implementation. Patients in highly resource-constrained settings are at the greatest risk of suffering the ill effects of TB disease, including long-term morbidity and death. This study represents an important first step toward improving clinical management for these marginalized patients and thus toward reaching global targets for ending the TB epidemic.

This is an easy-to-use clinical risk score designed to improve early management of tuberculosis in highly resource-constrained settings where same-day microbiological testing is unavailable. It consists of readily accessible demographic and clinical data and is scored from 1-10. We will train clinic staff in eight clinics (four study clinics and four comparison clinics) on the Ugandan standard of care for the diagnosis and treatment of TB. In addition, in the four study clinics, we will provide training on the PredicTB score.

(number of participants with microbiologically confirmed TB who initiate treatment within 7 days)/(number of participants with microbiologically confirmed TB) in the post- intervention period minus this same quantity in the pre-intervention period, excluding those confirmed on the basis of culture alone

(number of patient encounters in which PredicTB score is calculated and empiric treatment is initiated [or withheld] as suggested by the score)/(number of patients submitting sputum for evaluation of pulmonary TB) in the post- intervention period in study clinics

(cost of implementing PredicTB - cost of standard of care)/(projected disability-adjusted life years (DALYs) in standard of care - projected DALYs with PredicTB)

(number of participants with microbiologically confirmed TB who die of any cause)/(number of participants with microbiologically confirmed TB) in the post-intervention period minus this same quantity in the pre-intervention period, excluding those confirmed on the basis of culture alone

(number of participants with microbiologically confirmed TB who remain in care at 3 months)/ (number of participants with microbiologically confirmed TB) in the post-intervention period minus this same quantity in the pre-intervention period, excluding those confirmed on the basis of culture alone

(Difference in 7-day treatment initiation in study clinics) - (Difference in 7-day treatment initiation in comparison clinics)

(number of participants initiating TB treatment on the same day)/(number of participants submitting sputum for evaluation of pulmonary TB) in the post-intervention period in study clinics

(number of providers using PredicTB in over 50% of encounters in which sputum is submitted for pulmonary TB diagnosis)/(number of providers seeing >5 patients who submit sputum for diagnosis of pulmonary TB) at month 18 in the study clinics

(number of participants with microbiologically confirmed TB who initiate treatment within 7 days)/(number of participants with microbiologically confirmed TB) in Months 16-18 in the study clinics minus this same quantity in Months 7-9, excluding those confirmed on the basis of culture alone

Modeled changes in mortality at year 5, comparing simulations in which PredicTB is implemented to those in which PredicTB is not implemented, using a Markov state-transition model

Inclusion Criteria: All adult patients submitting sputum for a new diagnosis of pulmonary TB in the four study clinics and four comparison clinics between month -6 and month 18 will have their records abstracted by study staff. Starting in the 13th month after PredicTB implementation (i.e., after the 12-month post-implementation period has ended), study staff will position themselves in the four study clinics for purposes of recruiting and enrolling adult patients submitting sputum for a new diagnosis of pulmonary TB. No exclusions will be made except for age (as above), and we will seek to enroll all consecutive patients until our target sample size (25 participants per clinic, total n = 100) has been reached. Exclusion Criteria: Age < 15 years old