The Feasibility and Acceptability of a Post-tuberculosis Lung Disease Diagnostic Algorithm in Uganda.

Background Tuberculosis (TB) remains a major cause of death globally, with around 1.5 million deaths in 2020. These figures account only for deaths during the active disease phase, not those occurring after microbiologic cure. People who complete pulmonary TB treatment are about four times more likely to die early compared to the general population. Studies show up to 50% of TB survivors report health issues consistent with post-TB lung disease (PTLD). In Malawi, 61% of TB patients still had symptoms at treatment completion, and 28% had irreversible lung damage after five years. Pulmonary TB is now seen as a risk factor for chronic lung disease and poor quality of life. Between 60-90% of TB survivors face limited exercise ability, respiratory symptoms, and reduced quality of life. PTLD is marked by abnormal spirometry (airflow obstruction or low forced vital capacity) and recurrent symptoms often misdiagnosed as TB relapse. This leads to unnecessary TB treatment and drug-related side effects, especially with false-positive Xpert tests in the absence of smear or culture confirmation. PTLD is frequently missed in low-income, TB-endemic settings due to limited resources. Chest imaging shows lung cavities, bronchiectasis, and fibrosis. Spirometry reveals obstructive, restrictive, or mixed patterns, but these tests are not routinely available at primary health centers. Newer tests to distinguish PTLD from TB relapse have not been widely adopted, and culture-the gold standard-is not easily accessible. PTLD receives little attention in global TB guidelines, leading to limited research and diagnostic tools in low- and middle-income countries. Patients at risk of PTLD could be identified using pre-treatment chest X-rays, history, and physical exams, but these approaches are underused. The burden of PTLD varies across sub-Saharan Africa depending on local settings and case definitions. For example, a study in Tanzania found that 90% of people treated for TB in the past five years showed signs of PTLD. Despite a proposed definition from a recent post-TB symposium, no standard PTLD diagnostic criteria exist for low-resource settings. A study in three African countries showed inconsistent PTLD diagnosis, including in Ethiopia, where providers failed to diagnose PTLD in eligible patients. This highlights a critical gap in PTLD diagnosis. More research is needed to create a practical diagnostic tool and understand the barriers to clinical diagnosis. The overall aim of this trial is to build on an existing cohort and design a diagnostic algorithm for PTLD suitable for TB-endemic, low-income settings. PTLD presents a serious but often ignored challenge in TB-endemic areas. People finishing TB treatment should be assessed for overall health, including lung function and nutrition. However, low-income settings lack access to advanced tests like spirometry, relying only on symptom checks. TB programs often overlook PTLD, and health workers at primary facilities may have limited awareness. This trial focuses on improving early PTLD diagnosis at primary health centers, which can impact care for other chronic diseases. While training programs have improved diagnosis of non-communicable diseases, few focus on PTLD. A simple screening algorithm is needed for use at the primary care level. This trial aims to address both personal and system-level barriers to diagnosis. In Uganda, TB treatment is offered at primary health facilities, so early PTLD diagnosis depends on front-line clinicians. PTLD diagnosis is especially difficult at level-three health centers (HC3s) due to limited knowledge, no access to spirometry or imaging, and lack of clear guidelines.

Innovation Innovation A pilot cluster randomized clinical trial in post-TB morbidity is a step closer to improving post-TB care. Importantly, the selected health facilities will focus on post-TB diagnosis and care which has not been part of routine care. Despite the recommendation by the first PTLD symposium, this study aim presents an opportunity to solve this health challenge using locally generated data that can be applied to other low-income settings eliminating a one-size-fits-all approach.

Approach The main idea is that adding a clinical diagnostic algorithm to TB clinicians' workflow will support timely PTLD diagnosis and referrals for advanced care.

Design: This will be a prospective two-arm cluster randomized clinical trial (RCT) using the diagnostic algorithm designed in Aim 1 as the intervention. The primary outcome will be the duration of appropriate referral from level three health centers (HC3s) to a post-TB clinic at Mbarara Regional Referral Hospital (MRRH). The aim is to test the feasibility of implementing a clinical diagnostic algorithm for PTLD aiding clinicians at primary healthcare facilities (HC3s) in making accurate diagnoses of PTLD and facilitating timely referrals for effective treatment.

Study Setting: This study will be conducted at HC3s within Mbarara district Study Participants: The study will involve HC3s selected from the health facilities used for the qualitative interviews in Aim 2. TB clinicians at each health facility will screen individuals who have completed TB treatment for potential PTLD.

Participants: TB clinicians will screen people who completed TB treatment for PTLD.

Procedures: Study Procedures: Four HC3s will be recruited from a catchment area within 10 km radius of MRRH. Facility in-charges identified in Aim 2 will be contacted and requested to participate in the clinical trial. Upon their agreement, the in-charges will sign an informed consent form, following which the facilities will be randomly allocated to either the intervention or control group in a 1:1 ratio. A comprehensive training manual focusing on PTLD aligned with consensus guidelines will be developed for all clinicians at the selected HC3s. This training in form of continuing medical education (CME) will emphasize the significance of PTLD screening in TB survivors presenting with respiratory symptoms after completing TB treatment. The intervention will be the diagnostic algorithm that the investigators developed in Aim 1 as a visual aid whereas the control group will rely their clinical judgment and knowledge gained from the CME. Nurses or clinicians attending to TB patients at each HC3 will be requested to screen for PTLD in all TB survivors declared cured. This includes individuals returning to the TB department of each health facility with symptoms after TB cure. Those with a high suspicion for PTLD based on the diagnostic algorithm (intervention) or clinical judgment (control) will be referred to Mbarara Regional Referral Hospital for further diagnostics and care. A one-page triplicate referral form will be used to track referrals and ensure successful referral completion. The study team will conduct zoom conferences every fortnight with each of the four HC3 teams and to identify registration or referral gaps. No monetary incentives will be provided, but monthly internet and airtime will be provided to the contact health workers involved in the trial. Transport reimbursement and a snack will be provided to patients referred to MRRH. The study will be conducted following the guidelines of good practice in pilot interventional studies. Data Collection: Baseline and endpoint questionnaires will be completed by TB clinicians or any other designated personnel (such as a nurse) in collaboration with our research team. In addition to the routine programmatic TB outcomes documentation, data on patients screened for PTLD will be captured using a standardized data collection tool captured in REDCap. This will include participants demographics and TB treatment history including their prior eCBSS numbers. Facility demographic characteristics will be recorded. The data collection tools will be streamlined to ensure efficient data capture minimizing the potential for questionnaire fatigue. To prevent participants from being recorded at different HC3s due to their close proximity, the investigators will prioritize the recording of eCBSS numbers.

Outcomes: The primary outcome is the time from first visit with PTLD symptoms at HC3 to arrival at MRRH. With no current PTLD screening program, a 50% screening rate will indicate implementation success.

Secondary outcomes:

• Proportion of TB survivors screened for PTLD
• Number of successful referrals to MRRH
• PTLD confirmation rate via imaging/spirometry

Feasibility of a full RCT will be based on:

1. ≥50% facility acceptance rate
2. Completeness of REDCap and algorithm questionnaire data
3. Clinician feedback at study end Data Analysis: Baseline characteristics will be summarized separately for both the intervention and control groups. Feasibility and process evaluation data such as practice recruitment rate and adherence to the intervention will be summarized and presented as percentages. Differences in (1) the primary care interval and (2) screening rate between the two treatment groups will be compared. An exploratory analysis using multivariate regression will be performed to identify factors associated with the ability to implement the intervention

Sample Size: Since this is a feasibility study, study is not powered to test an efficacy hypothesis. However, this study has maximum acceptable confidence interval for an anticipated intra-cluster correlation coefficient of 0.05, at a set cluster size of 20 which is typical for pilot cluster trials, with a coefficient of variation 0.1 and anticipated proportion of 0.5 (50%) for our primary outcome. Therefore FOUR (4) clusters are needed to estimate this proportion (50%) with a lower 0.9 confidence (10% margin of error) limit of 0.27 and upper confidence interval limit 0.73 Key Deliverables: The feasibility of a diagnostic algorithm will be presented in a larger cluster randomized clinical trial to scale up for timely PTLD diagnosis.

Key Deliverable: Demonstrate feasibility of a full-scale cluster RCT to improve PTLD diagnosis.

Challenges and Alternatives: There is a possibility of questionnaire fatigue given the existing programmatic TB questionnaires. However, strategies will be implemented to minimize missing data: Given the average patient load at HC3s TB units (an average 15 patients in a month), our study procedures and short electronic questionnaires are not expected to pose a significant workload. Secondly, monthly airtime and internet will be provided to both groups and an additional incentive such as a mobile phone loan for completing REDCap questionnaires. If the clinical algorithm demonstrated poor diagnostic accuracy, behavioral change will be assessed based on the impact of PTLD-focused CME using the consensus definition as the primary intervention compared to current practice.