Six-year trend and risk factors of unsuccessful pulmonary tuberculosis treatment outcomes in Thai Community Hospital

Tuberculosis (TB) is a major cause of morbidity and mortality globally. Despite efforts to eliminate TB in Thailand, the incidence rate has declined slowly. This study aimed to identify the incidence and risk factors of unsuccessful pulmonary TB treatment (failed, died and loss-to- follow up) in a community hospital in Chachoengsao Province, Thailand from 1st January 2013 to 31st December 2019. A total of 487 patients were eligible for the study. The incidence of unsuccessful treatment was 21.67/100 population person year. Risk factors of unsuccessful pulmonary TB treatment were unemployment (adjusted hazard ratio (AHR) 3.12, 95%CI 1.41–6.86), HIV co-infection (AHR 2.85, 95%CI 1.25–6.46), previous history of TB (AHR 2.00, 95%CI 1.04–3.81), positive sputum AFB at the end of the intensive phase (AHR 5.66, 95%CI 2.33–13.74), and sputum AFB was not performed at the end of the intensive phase (AHR 18.40, 95%CI 9.85–34.35). This study can be utilized to improve prevention and intervention of TB treatment by strengthening public health system on treatment quality especially TB patient monitoring tools or methods easy for accessing to patients in communities.


Introduction
Tuberculosis (TB) is a major public health problem worldwide [1,2]. Most cases were in South-East Asia, Africa, and West Pacific [1]. Despite the decline of TB incidence in recent years, the progression was unable to achieve the goal of World Health Organization (WHO) End TB Strategy by 2020 [2].
In Thailand, community hospitals are frontiers in combating TB in rural areas. Community hospitals are public hospitals that emphasize primary and secondary health care [3]. Thus, most of patients' health care in rural areas of Thailand relies on these hospitals. Challenges in TB treatment in these settings included patient's poor TB perception and knowledge [4,5], low socioeconomic status [4], and case management by non-physician healthcare providers [6].
Studies of risk factors of unsuccessful pulmonary TB treatment were scarce in community hospitals. This study aimed to identify incidence and risk factors of unsuccessful pulmonary TB treatment outcomes in Thai community hospital setting.

Study population and setting
This study included pulmonary TB patients receiving treatment at outpatient department (OPD) of a community hospital in Chachoengsao Province, Thailand from 1st January 2013 to 31st December 2019. Patients were followed from the date they initiated on TB treatment. This study included only adult patients (age ≥ 18 years at the start of TB treatment) andexcluded patients with extrapulmonary TB without pulmonary TB involvement, latent TB, died from causes other than TB and TBrelated illness during TB treatment and transferred to other treatment units. The exclusion of extrapulmonary TB without co-existing pulmonary TB was due to limited cases. The studied hospital was F1 grade, defined by Health Administration Division as a large-sized community hospital with 120 beds with both general practitioner physicians or family physicians and specialists from at least one major specialty.

Study design
Retrospective cohort study was conducted from 1st January 2013 to 31st December 2019 to identify incidence and risk factors for unsuccessful pulmonary TB treatment.

Data collection
Data stored in tuberculosis treatment card and online registration platform from 2013 to 2019 were retrieved. To counter file lost and incompleteness and verifying number of files against cases notified, data were retrieved from both treatment cards and online platform which were identical. The data were registered into electronic platform and confirmed by double-checking on both sources, patients' OPD cards and records of chest radiograph. They included baseline characteristics (gender, age, occupation, history of imprisonment and history of TB contact), medical history (co-morbid illnesses, human immunodeficiency virus (HIV) infection, history of previous TB, history of Bacillus Calmette-Guérin (BCG) vaccination), courses of TB treatment (date of treatment initiation and completion of current TB episode, sputum acid-fast bacilli (AFB) at the start of treatment and end of intensive phase, chest radiographs when treatment was initiated, case status, and hepatotoxic side effect following the treatment) and treatment outcomes (cured, completed, failed, died, and loss-to-follow up).

Operational definitions
Pulmonary TB is a case of TB involving lung parenchyma [7,8]. Treatment outcomes can be classified as cured, completed, failed, died, loss-to-follow up, and transferred [8]. Patients are cured when their sputum AFB smear or culture is negative at the end of treatment if their sputum AFB is positive at the start of treatment. Patients are registered completed when their treatment ended without sputum AFB smear or culture results in the last month of treatment result, but the latest sputum smear is negative. Treatment failure is considered when patients' sputum AFB at the 5th month or later is positive. Patients are classified 'die' when they die before treatment completed regardless of causes. Loss-to-follow up is defined as interrupted treatment for at least 2 consecutive months. Patients with cure and complete treatments were grouped as successful treatment. Patients who failed, died and loss-to-follow up were classified as unsuccessful treatment. Definitions of cases and hepatotoxicity were described in Additional file 1.

Statistical analysis
This study used STATA 15.1 (Stata Corporation, College Station, TX, USA) for analyzing incidence and risk factors of unsuccessful pulmonary TB treatment (failed, died and loss-to-follow up). Incidence was calculated using person-time function. The incidence rate was per 100 person-year (PY). Risk factors were calculated by Cox proportional hazard model. Univariate cox proportional hazard model was used for analyze risk factors of unsuccessful treatment outcomes. Factors which were significant or had p-value < 0.20 in univariate analysis or significant in previous studies were recruited for multivariate cox proportional hazard model. Factors with p-value ≤ 0.05 were statistically significant.

Ethical consideration
This study was approved by Institutional Review Board Royal Thai Army Medical Department. The study number was R054h/62_Exp. Written informed consents were obtained from all participants when they started TB treatment at the treatment site.

Results
From 519 TB patients, 3 were dropped out due to incomplete treatment initiation or completion dates. Nine were excluded due to extrapulmonary TB without pulmonary TB involvement. Eight had latent TB. Four died or lossto-follow up before initiation of treatment. Six died from causes other than TB and TB-related illness. Two were transferred to other treatment units. Consequently, 487 patients were enabled for this study.

Treatment follow-up results and outcomes
Following the initiation of TB treatment, patients affected by hepatotoxic side-effect of TB treatment were 10.84%. At the end of the intensive phase, 6.34% of the patients had positive sputum AFB. Two (0.41%) patients were transferred out. Total successful treatment cases were 88.09% which can be classified as 55.03% cured and 33.06% completed. The results were shown in Table 2. Unsuccessful treatment outcome has been declining from 21.43% in 2013 to 9.38% in 2019. Additional file 2: Figure  S1 displayed trend of unsuccessful treatment outcomes.

Incidence of unsuccessful treatment outcome
Overall 58 cases were identified after the total follows up time of 267.68 PY which resulted in a cumulative incidence of 21.67 per 100 PY. Incidence rates were high among patients age  Table 3

Discussion
This study addressed incidence and risk factors of unsuccessful pulmonary TB treatment in a community hospital during 2013-2019 which synchronous with WHO End TB 2020 milestone [9]. This study showed decreasing incidence of unsuccessful treatment outcomes. The success rate of this study was similar to the national success rate reported by WHO [1]. By 2019, the successful treatment rate of this study was at 90.63% which was higher than the overall treatment success rate of Thailand [9]. At this percentage, the successful treatment outcome has exceeded the target expected in 2022 by End TB strategy [9]. This could be attributed to the introduction of effective directly observed treatment, short-course (DOTS) and holistic care to patients [10,11].
It was observed that in 2013, there were few TB cases, but had highest incidence of unsuccessful treatment due to founding phase of TB clinic. The situation improved in the following years from effective patient monitoring. In this setting, involving health-care volunteers, trained individuals to oversee communities' health welfare, were employed for DOTS at home, as well as health education for communities. Video-call DOTS were employed in patients who were not convenient in visiting hospitals or living far from the communities. Health-care providers (usually TB clinic nurses) would video-call to these patients in specific time of the day to observe them taking drugs.
This study showed that unemployment was the risk of unsuccessful TB treatment as in previous studies [12,13]. Unemployment had roles in delayed and interrupted treatment [14,15]. Patients with unemployment were more likely to have low socioeconomic status and thus, had difficulty accessing TB treatment [16]. Unemployment also linked to homelessness which related to unsatisfied TB treatment outcomes [14,17].
This study demonstrated HIV co-infection as the risk factor of poor TB treatment outcome as explained in various studies [18][19][20]. This study also showed that incidence of unsuccessful treatment was high among HIV co-infected patients which could be resulted from high mortality rates as 20.00% of TB-HIV patients died, slightly higher than national mortality rate [1,8,21]. Thailand aimed on acquired immunodeficiency syndrome (AIDS) elimination by 2030. The plan included easy access to health care services, HIV screening, antiretroviral therapy (ART) and retaining patients in treatment system. The outcome was to make 90% of HIV patients realize they carry the disease, received ART and able to suppress viral load. The outcomes were satisfying     [22]. Sputum AFB follows up, especially at the end of intensive phase, are important predictor of treatment outcome [23,24]. This study used follow up of sputum AFB at the end of the intensive phase to identify risk of unsuccessful treatment. Positive sputum AFB during follow up could indicate various factors such as patients' poor compliance and drug-resistance TB [23,24]. This study also found that patients who did not received sputum AFB follow up, which contradicted to Thai TB treatment guideline [8], were at risk as well. It could be hypothesized that patients who did not receive proper follow up were either having poor treatment compliance, receiving out-of-track management or missing cases followup by health care providers. Out-of-track management and cases missing can be improved by recruiting public health volunteers. Knowledge implementation about case detection and patient monitoring for these volunteers is required to serve health care to communities and reduce workloads on health care providers.
In previous studies, previous treatment of TB was associated with unsuccessful treatment outcomes [6,25,26]. Unfavorable previous treatment outcomes are predictive for development of drug-resistance TB in the current treatment [27,28]. Patients who were loss-to-followed up in previous treatment episodes were more likely to be loss-to-followed up as well in the current treatment (29).
This study highlighted possibility of effective measures to increase yield in the treatment success rate by enforcing networks of the public health care system on TB treatment quality, especially patients monitoring. Active TB case monitoring by local health sectors is the crucial point for encouraging patients to follow and maintain in the treatment system. Developments of methods, tools or technology for active case monitoring and easy to access patients in the communities are requirements which will bring convenience for both patients and health-care workers.

Limitations
Limitations included registration of tobacco and alcohol consumption in only yes/no form instead of consumption amount and pack-years used Second, the sample size of this study was small due to limited number of patients. Third, this study did not include information on antiretroviral therapy, which would be important predictors of poor TB treatment outcomes.