**An Innovative Public–Private Mix Model for Improving Tuberculosis Care in Vietnam: How Well Are We Doing?**

**Thuong Do Thu 1,\*, Ajay M. V. Kumar 2,3,4 , Gomathi Ramaswamy <sup>5</sup> , Thurain Htun <sup>6</sup> , Hoi Le Van <sup>1</sup> , Luan Vo Nguyen Quang <sup>7</sup> , Thuy Dong Thi Thu <sup>7</sup> , Andrew Codlin <sup>7</sup> , Rachel Forse <sup>7</sup> , Jacob Crewsell <sup>8</sup> , Hoi Nguyen Thanh <sup>9</sup> , Hai Nguyen Viet <sup>1</sup> , Huy Bui Van <sup>1</sup> , Hoa Nguyen Binh 1,3,10 and Nhung Nguyen Viet 1,10**


Received: 14 November 2019; Accepted: 10 February 2020; Published: 14 February 2020

**Abstract:** To improve tuberculosis (TB) care among individuals attending a private tertiary care hospital in Vietnam, an innovative private sector engagement model was implemented from June to December 2018. This included: (i) Active facility-based screening of all adults for TB symptoms (and chest x-ray (CXR) for those with symptoms) by trained and incentivized providers, with on-site diagnostic testing or transport of sputum samples, (ii) a mobile application to reduce dropout in the care cascade and (iii) enhanced follow-up care by community health workers. We conducted a cohort study using project and routine surveillance data for evaluation. Among 52,078 attendees, 368 (0.7%) had symptoms suggestive of TB and abnormalities on CXR. Among them, 299 (81%) were tested and 103 (34.4%) were diagnosed with TB. In addition, 195 individuals with normal CXR were indicated for TB testing by attending clinicians, of whom, seven were diagnosed with TB. Of the 110 TB patients diagnosed, 104 (95%) were initiated on treatment and 97 (93%) had a successful treatment outcome. Given the success of this model, the National TB Programme is considering to scale it up nationwide after undertaking a detailed cost-effectiveness analysis.

**Keywords:** public–private mix model; public–private partnership; missing cases; operational research; SORT IT

#### **1. Introduction**

Tuberculosis (TB) is the leading cause of mortality from a single infectious agent globally, accounting for 1.45 million deaths annually [1]. The global community has pledged to end the TB epidemic by 2030 [2]. While there has been progress, the rate of decline of TB incidence has been modest at ~2% each year [3]. At this rate, we will not be able to realize the goal of ending TB by 2030. To accelerate progress, the Stop TB Partnership recommends that countries should strive to achieve 90-(90)-90 targets (diagnosing 90% of all people with TB including 90% among key populations and treating 90% of them successfully) [4].

One of the major challenges in TB control is "missing cases". Globally, of the 10 million people estimated to have developed TB in 2018, only 7 million were notified [1]. The gap of 3 million includes people who are not diagnosed and treated, and those managed in the private health sector, but not notified to National Tuberculosis Programmes (NTP).

A multi-country study found that in more than 60% of TB patients, the private sector was the first point of contact, yet the proportion of cases notified to NTP was less than 10% [5]. Also, health care provision in the private sector is not standardized and poorly regulated in many countries. These may lead to delays in diagnosis and treatment, improper case management, increased risk of developing drug resistance, disease transmission, and catastrophic health expenditure [6]. Engagement and collaboration with private sector in those countries where a large proportion of care seeking is sought with private providers in critical [2].

Vietnam is one among the 30 high TB burden countries and has strong private health care sector. It is estimated that approximately 50% of TB patients seek initial care in the private sector before visiting the public health system [7,8]. In 2018, only 57% of estimated incident cases were notified to NTP, meaning there is no information about the rest of the patients and a majority of these may be receiving care in the private sector [1]. Mirroring the global picture, previous studies from Vietnam have also reported mismanagement in diagnosis and treatment of TB in the private health sector [9–11]. All these findings underline the need to engage the private health care providers in Vietnam's TB care and prevention efforts.

To engage the private sector, several public–private mix (PPM) models have been implemented by the Vietnam NTP since 2001. One such model involved training of private health care providers and strengthening referral mechanisms between the private sector and NTP. This yielded an increase in overall TB case detection rate of Ho Chi Minh City by 7% [12], but there were several gaps. About 30% of presumptive TB patients referred to NTP for sputum microscopy did not reach the diagnostic facility and nearly 60% of the patients diagnosed were lost to follow-up before starting treatment. Of those started on treatment, only 60% successfully completed it [13].

To address these gaps, a new model was implemented in Haiphong International General Hospital (HIGH) in 2018 as part of the TB REACH-funded Zero TB Vietnam initiative. This model included three unique components: (i) active facility-based screening of all adults for TB symptoms (and chest x-ray (CXR) for those with symptoms) by trained and incentivized providers, with on-site diagnostic testing or transport of sputum samples, (ii) an innovative mobile application to reduce dropout in the care cascade and iii) enhanced follow-up care through engagement of a local network of community health workers (CHWs). However, this model has not yet been systematically evaluated. In this study, we aimed to evaluate the performance of this private sector engagement model by tracking the cascade of tuberculosis care among the individuals attending the HIGH in Vietnam from June to December 2018.The specific objectives were to determine, (i) the number (proportion) with presumptive tuberculosis and among them, the number (proportion) who were investigated for tuberculosis (ii) the number (proportion) diagnosed with tuberculosis and initiated on treatment (iii) the treatment outcomes among those initiated on treatment and (iv) the delays involved at different steps of the care cascade.

#### **2. Materials and Methods**

#### *2.1. Study Design*

This was a cohort study involving analysis of routine surveillance data.

#### *2.2. Setting*

Vietnam is a South-East Asian country with a population of 93.7 million. The country is divided into 63 provinces, which are further divided into districts, communes, and sub-communes. About 34.4% of the population live in urban areas and about 10% people are below the poverty line [14]. Healthcare services are provided by both public and private sectors.

#### 2.2.1. TB Control Program in Vietnam

The public health facilities for TB care and prevention are managed either directly by the NTP or indirectly through the Department of Health (DOH). Under the NTP, TB diagnosis, treatment, and control activities are carried out by the national level unit, 63 TB and lung disease hospitals or provincial TB units, and 707 TB management units (TBMUs) at the district level.

#### 2.2.2. PPM Models

There has been significant progress in the implementation of PPM initiatives for TB care and prevention in Vietnam. There are four types of PPM models. PPM Model 1 entails the referral of presumptive TB patients identified by the private providers to NTP for further evaluation. PPM Model 2 encompasses the diagnosis of TB and referral to the NTP. PPM Model 3 entails the provision of directly-observed treatment. PPM model 4 provides both diagnostic and treatment services similar to a TBMU. Private providers participating in PPM initiatives are trained for TB screening, diagnosis with sputum smear microscopy, and recording and reporting as per NTP guidelines.

#### 2.2.3. PPM Model at HIGH

PPM model at HIGH is similar to the model 4. HIGH was chosen for the intervention because it is the biggest tertiary care private hospital in Haiphong province (considered a poor-performing province by NTP) and there was political commitment from the director of the HIGH to collaborate with the NTP. Three departments from the HIGH (Endocrinology, Otolaryngology, and Respiratory medicine) were targeted for systematic screening. These three departments were chosen because they were expected to account for the majority of pulmonary presumptive TB patients visiting HIGH. The doctors and nurses of these departments were trained on the management of TB cases as per NTP guidelines and were given incentives for TB screening, diagnosis and treatment, and systematic recording and reporting.

A total of 110 USD of fixed allowance per month was provided to the hospital. About 70 USD was provided every month for conducting the monthly review meetings and printing of forms. In addition, performance-based incentives were provided: (i) 0.5 USD given to the nurse for each chest X-ray conducted (ii) 1 USD given to the nurse for each sample transported for Xpert MTB/RIF testing (iii) 2 USD given to the doctor for each patient diagnosed with TB (iv) 3 USD given to the doctor for each TB patient completing treatment. The project was supervised by a focal point (@45 USD per month), PPM coordinator (@130 USD per month) and a PPM supervisor (@130 USD per month). These incentives were on top of the salaries they received. Thus, the total cost incurred on the project for six months was 4100 USD. All costs were incurred in Viet Nam Dong and translated to USD based on the average exchange rate during the implementation period.

All out-patients attending the selected departments (Endocrinology, Otolaryngology, and Respiratory medicine) were screened for symptoms such as cough, hemoptysis, chest pain and dyspnea, fever, fatigue and unexpected weight loss. Individuals who had any of these symptoms or who had a history of contact with TB were requested to undergo chest radiography. Contacts were

defined as people living in the same household with a TB patient for at least two nights per week during the last 6 months. Individuals with symptoms suggestive of TB or exposure to a person with TB and parenchymal abnormalities suggestive of TB on chest X-ray were considered 'presumptive TB' and evaluated further for bacteriologic confirmation. The chest radiographs of the presumptive TB patients were examined by the radiologists at HIGH. Patients with TB symptoms and no abnormalities on chest X-ray were also offered tests for bacteriological confirmation based on the discretion of the attending physician. Spot sputum samples (without induction) were collected from the patients. For those who were not able to produce sputum, bronchoscopy was suggested by the treating physician and the bronchial washings were used for further testing. Patients with lymphadenopathy were referred for histopathological examination for confirmation of TB.

Individuals with a positive symptom screen were counselled by the treating physician about the different diagnostic tests available and their costs and were offered one of more of the following tests in the hospital for bacteriological confirmation: (i) Transcription Concerted Reaction (TRC) Ready 80 test (ii) sputum microscopy (iii) liquid culture and drug susceptibility test for first-line drugs. The TRC Ready 80 test is an automated molecular assay designed to detect mycobacterium tuberculosis (MTB) complex 16S rRNA present in clinical specimens (pulmonary and extra-pulmonary) or culture isolates. This has been described in greater detail elsewhere [15]. Though this test not endorsed by the WHO yet, it has been approved for use in Vietnam by the MOH for diagnosing TB. Chest X-ray screening and diagnostic tests at HIGH were paid by the patient at the following rates: 3 USD for chest X-ray, 3 USD for microscopy, 12 USD for culture, and 36 USD for TRC Ready 80. For patients unable to afford these tests, sputum samples were collected and transported to a nearby NTP facility for TB diagnosis using Xpert MTB/RIF assay, which was offered free of charge. The nurses in HIGH were trained on the procedures of sputum collection in falcon tubes, packaging and transportation in cold chain to the nearby NTP facility (which is located at a distance of 1.5 kms from HIGH). Nurses transported the sputum specimens in-person at the end of the day to the NTP facility.

Patients diagnosed with TB disease either through bacteriologic confirmation or clinical diagnosis were initiated on anti-TB treatment at an NTP facility or a private health facility of the patient's choice. Patients diagnosed with drug-susceptible TB were treated with first-line drugs and those with rifampicin resistance were referred to the NTP's provincial TB hospital for further evaluation. While drugs were provided free of charge to the patients at NTP facilities, patients paid out of pocket at private health facilities.

The CHWs were notified immediately after the diagnosis of each TB patient for linkage to care and follow-up from treatment start until completion. CHWs were motivated volunteers identified from each commune and recruited to support TB care and prevention activities in the Zero TB project (under funding support of TB REACH grant). They were mostly women and received formal training (for two days by NTP staff) on screening, counselling, follow-up care and support of TB patients. In the community, CHWs performed household contact tracing and counseling on treatment adherence and infection control. They received performance-based incentives as part of another project and did not receive any specific incentives for the project described in this study. The study's case definitions and treatment outcomes were in accordance with NTP and WHO guidelines.

#### 2.2.4. Recording and Reporting

All patient details were entered in the ACIS (Access to Care Information System, Clinton Health Access Initiative/TechUp, Vietnam) application, a data collection and case management tool for persons with presumptive TB in the community. Dedicated tablets with preinstalled software were procured and provided to the nurses and were trained on its use. Nurses captured data about presumptive TB patients using this tool. This application is bi-directionally connected with the NTP's electronic recording and reporting system, the Vietnam TB Information Management Electronic System (VITIMES), which enabled the electronic referral of case files of persons with suspected or diagnosed TB to NTP facilities.

#### *2.3. Study Population*

All patients aged ≥15 years attending the three outpatient departments of HIGH between 24 June, 2018 and 31 December, 2018 were included.

#### *2.4. Data Variables and Sources*

Case-level covariates were extracted from the ACIS and VITIMES systems. These included TB symptoms, chest X-ray findings, diagnostic test results, diagnoses, treatment initiation dates, and treatment outcomes. We used name, age, and sex to merge the two databases, but removed names and other personal identifiers before analysis to ensure confidentiality of data.

#### *2.5. Analysis and Statistics*

Data was analyzed using Stata software (version 14.0, Statacorp, Texas, TX, USA). We have depicted the cascade of care in the form of a flowchart with dropouts at every stage summarized as frequencies and percentages. To calculate the proportion with presumptive TB, we chose all people attending the three out patient departments (OPD) as the denominator, because information on who among them were screened was not available in the records. TB treatment outcomes were categorized into successful (cured and treatment completed) and unsuccessful (failure, lost to follow-up, died, not evaluated) outcomes. The time delays between screening, undergoing diagnostic test and initiation of treatment were summarized using median and interquartile range (IQR). Factors associated with 'not getting investigated for TB' among presumptive TB patients were assessed using adjusted risk ratios and 95% confidence intervals (CI) calculated using log-binomial regression. We also assessed factors associated with TB diagnosis among patients investigated for TB using the same effect measures.

#### *2.6. Ethics Approval*

Ethics approval was obtained from the Scientific Ethics Committee of the National Lung Hospital, Hanoi, Vietnam (approval number 954/QD-BVPTU) and the Ethics Advisory Group of the International Union Against Tuberculosis and Lung Disease, Paris, France (approval number 22/19). Since this was a review of existing records with no direct interaction with human participants, the need for individual informed consent was waived by the ethics committees. Confidentiality of the patient data was ensured by (i) providing restricted access to patient data only to the research team (ii) using password protection to access the electronic files and (iii) removing all the personal identifiers (such as name, address, phone number) before analysis.

#### **3. Results**

A total of 52,078 adults attended the three OPDs at HIGH during the study period. Of them, 2739 (5.3%) had either symptoms suggestive of TB or were exposed to someone with TB. Of these, 1372 (50%) were male and mean (SD) age was 49 (17) years. The profile of these patients is depicted in Table 1. Cough was the predominant symptom present in 2240 (82%) individuals, followed by chest pain and dyspnea in 1926 (70%), fatigue in 1521 (56%) and fever in 418 (15%). Contact with a person with TB was reported by 34 (1%) patients.


**Table 1.** Socio-demographic and clinical profile of people with tuberculosis (TB) symptoms or contact history attending the Haiphong International General Hospital in Vietnam, from June to December 2018.

HIV = Human immunodeficiency virus infection, TB = Tuberculosis; \* 125 of these patients had hemoptysis.

#### *3.1. Cascade of Care*

The cascade of care among study participants is depicted in Figure 1. Of the 2739 symptomatic individuals or those with contact history, 368 (13.4%) had chest X-ray suggestive of TB and were identified as having presumptive TB (i.e., 368/52,078 [0.7%]). Of these, 299 (81%) underwent at least one of the diagnostic tests for TB and 103 (34.4%) were diagnosed with TB. In addition, 195/2371 (8%) patients with normal chest X-ray also underwent tests for TB, of whom 7 (4%) cases were diagnosed. In total, 110 people were diagnosed with TB. Of the 110 TB patients, 92 (84%) had bacteriologically confirmed TB and the rest were either clinically diagnosed or based on the results of other histopathological investigations. Of them, 104 (95%) were initiated on anti-TB treatment. All, except one, received treatment at an NTP health facility. Two individuals had rifampicin-resistant TB and were referred to the Provincial Lung Hospital and subsequently treated with second-line drugs. The majority of patients had pulmonary TB (n = 98, 88%), while the remaining 12 patients had extra-pulmonary TB (09 pleural TB and 3 lymph node TB).

**Figure 1.** Cascade of tuberculosis care (from screening to treatment outcome) among the patients attending the Haiphong International General Hospital in Vietnam, from June to December 2018. OPD = Outpatient Department; TB = tuberculosis; TRC = transcription concerted reaction. a: successful outcome: cured and treatment completed. B: unsuccessful outcome: death, loss to follow-up, failure, and not evaluated. \*Two individuals had rifampicin resistance and were started on second-line drugs.

#### *3.2. Factors Associated with 'Not Getting Tested for TB'*

Of the 368 individuals with presumptive TB, 69 (19%) did not undergo any diagnostic testing. Patients without cough, without fever, without night sweats, without weight loss, and without TB contact history were less likely to be tested for TB (Table 2).



\* Data missing for one; HIV = human immunodeficiency virus; TB = tuberculosis; CI = confidence interval; aRR = adjusted relative risk.

#### *3.3. Factors Associated with TB Diagnosis*

Of the 494 patients who underwent diagnostic tests, 110 (22%) were diagnosed with TB. Patients with self-reported diabetes and those with weight loss had a significantly higher chance of getting diagnosed with TB (Table 3).


**Table 3.** Factors associated with diagnosis of TB among the patients who were investigated in the Haiphong International General Hospital in Vietnam, from June to December 2018.

\* Data missing for one; HIV = human immunodeficiency virus; TB = tuberculosis; CI = confidence interval; aRR = adjusted relative risk.

#### *3.4. Treatment Outcomes*

The TB treatment outcomes are shown in Table 4. Among 104 patients initiated on treatment, 97 (93%) had successful treatment outcome, while 5 (5%) had unsuccessful outcome and 2 (2%) were still on treatment.

**Table 4.** Treatment outcomes among tuberculosis patients started on treatment in Haiphong International General Hospital in Vietnam, from June to December 2018.


\* Two patients are on second-line treatment and are likely to complete by April 2020.

#### *3.5. Median Delays*

The delays at different steps of the cascade are shown in Table 5. The median (IQR) duration from visiting the HIGH to undergoing TB diagnostic test was 0 (0–1) day and from diagnosis to initiation of treatment was 6 (1–17) days.

**Table 5.** Delays in the TB care cascade among the patients attending the Haiphong International General Hospital in Vietnam, from June to December 2018.


TB = Tuberculosis; IQR = Interquartile Range; HIGH = Haiphong International General Hospital.

#### **4. Discussion**

This is the first report from Vietnam evaluating an innovative PPM model using an information technology based tool for improving tuberculosis care in private health sector. While there are many studies evaluating the specific components of the TB care cascade in the private sector, very few have comprehensively examined all the steps of the cascade in a single study [16]. This is one such effort. Overall, the performance of the model was excellent in plugging the gaps in TB care cascade in the private sector and substantially better than previous PPM models implemented in Vietnam [13]. About 80% of the presumptive TB patients were investigated for TB. Nearly 95% of the cases diagnosed were initiated on treatment, which is significantly better than previous studies from Vietnam and Pakistan, where nearly 60% were lost to follow-up before treatment [13,17]. All the cases were notified to NTP. More than 90% of all TB patients completed the treatment successfully, in line with the global 90-(90)-90 targets. These results were better than those reported from India [18], Pakistan [17], Thailand [19], and Vietnam [9,13] and were on par with outcomes reported from Myanmar [20].

In our view, the success of the model may be attributed to the following aspects. First, unlike earlier PPM initiatives which predominantly used a 'referral model' for investigation of tuberculosis (wherein presumptive TB patients were referred to an NTP facility), the new model offered TB tests on-site or arranged for transportation of sputum samples. This might have reduced the gaps and delays in testing. Second, the use of a mobile application enabled notification of every TB case diagnosed. This alerted the health care system and the last-mile service providers like CHWs to proactively track and provide follow-up care to the patients, thus reducing gaps in treatment initiation and completion. Third, all the providers were trained and performance-based incentives were offered for every successful event in the care cascade. The total costs incurred were modest at 4100 USD for the six-month pilot period (equivalent to ~37 USD per TB case diagnosed). However, we have not undertaken a detailed cost-effectiveness analysis. This should be a topic of future research.

There were some other notable findings. First, only 0.7% of patients attending the OPD were identified as 'presumptive TB' patients. This is substantially lower than that reported from other settings like Pakistan (which varied from 2.9% to 7.5%) [21,22]. This difference is likely due to many differences between the settings which include (i) a stricter definition used for 'presumptive TB' (both symptom positive and chest X-ray abnormality) and (ii) the denominator being all patients attending OPD rather than the number screened in our study. It is possible that some of the patients attending the OPD might not have been screened and there was no documentation to find out the exact numbers screened.

Second, about one in five presumptive TB patients did not undergo investigations and people without symptoms were less likely to undergo investigation. This is concordant with the observations by Creswell et al. in Karachi, Pakistan [22]. Patients without symptoms may have low risk perception or may have been accorded lower priority for testing by attending clinicians. Some patients may not have been able to produce a sputum sample. The high costs of the diagnostic tests for which patients had to pay out of pocket may have been another deterrent for uptake of tests. Also, people with symptoms but normal chest X-ray were less likely to be tested. This may be again related to the definition of presumptive TB used in this project, which required an abnormal X-ray in addition to symptoms. This may also be the reason for the high yield of TB (33%) among people with presumptive TB.

Had we tested everyone with symptoms, we might have had lower yield in terms of percentage, but more cases in terms of absolute numbers. Of course, this would have had additional cost implications. One possible way to increase the number of cases detected without too much additional effort will be to use the duration of symptoms to prioritize investigation for TB—like investigating only those with cough of more than or equal to 2 weeks rather than testing everyone with cough of any duration [23]. Unfortunately, we did not have information on duration of symptoms and hence we cannot comment on this issue any further. All these call for revisiting the definition of presumptive TB used in the project.

Third, a standard diagnostic algorithm was not followed in the project. The nature and the number of tests offered to each patient seemed to vary. While we do not know the exact reasons for this variation, we speculate that this was dependent on the ability of individual patients to afford the high costs of the tests and based on the physician's choice of diagnostic test. This aspect needs to be studied further using qualitative research methods. We recommend that all patients undergo a standard diagnostic algorithm, preferably using tests approved globally for use and, if possible, at subsidized costs.

The study had some limitations. First, we relied on routinely collected data and hence errors in documentation cannot be ruled out. However, we estimate that such errors are limited in number and impact given real-time and post hoc data validation mechanisms in the ACIS software and by data management team. Second, there was no documentation about the number of people screened. As a result, we were unable to calculate the 'number needed to screen' to detect an additional TB case. Also, the patients attending only three OPDs were screened. Hence the number of TB cases diagnosed may not reflect the true burden of TB among patients attending HIGH. We may have missed many patients, especially those with extrapulmonary TB because departments such as surgery, gynecology and urology were not involved in the project. We may also have missed many patients because of the strict definition of presumptive TB used in our study. In a national TB prevalence survey from Vietnam, only 10% of all presumptive TB patients fulfilled such strict criteria and accounted for only 27% of all TB patients [24]. Third, the study was conducted in a single hospital thereby limiting its generalizability. For this reason, we have refrained from the assessing the impact of this intervention on case notification at the community level. This kind of impact has been demonstrated by previous studies elsewhere [18,21,25]. Fourth, we did not have data for the pre-intervention period to enable before–after comparisons. There was no systematic recording and reporting of TB-related indicators before the study. The data obtained in this study may act as a baseline for any future evaluations. Finally, the exact reasons for the gaps at each step of the cascade were not investigated in this study. Future research should look into this aspect using qualitative research methods.

In conclusion, the new PPM model in Vietnam performed well with high levels of testing, diagnosis, treatment start and completion among TB patients. Given the success of this model in plugging the gaps in TB care cascade, the NTP in Vietnam is considering to scale-up this model nationwide after undertaking a detailed cost-effectiveness analysis. The lessons learned from this study may be useful to make amendments in the PPM model and optimize project implementation going forward.

**Author Contributions:** Conceptualization and protocol development: T.D.T., A.M.V.K., G.R., T.H., H.L.V., L.V.N.Q., H.N.T., A.C., R.F., T.D.T.T., T.H., H.N.V., H.B.V., H.N.B., and N.N.V. Data Collection: T.D.T., T.D.T.T., and A.C. Data Analysis or interpretation: T.D.T., A.M.V.K., G.R., T.H., and J.C. Writing the first draft: T.D.T., A.M.V.K., and G.R. Critical review of the paper and final approval: T.D.T., A.M.V.K., G.R., H.L.V., L.V.N.Q., H.N.T., A.C., R.F., T.D.T.T., H.N.V., H.B.V., H.N.B., N.N.V., and J.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research study was generously supported by a grant from the Stop TB Partnership's TB REACH initiative, with funding from Global Affairs Canada. The training program, within which this paper was developed, and the open access publication costs were funded by Department for International Development (DFID), UK and La Fondation Veuve Emile Metz-Tesch (Luxembourg). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

**Acknowledgments:** This research was conducted through the Structured Operational Research and Training Initiative (SORT IT), a global partnership led by the Special Program for Research and Training in Tropical Diseases at the World Health Organization (WHO/TDR). The model is based on a course developed jointly by the International Union Against Tuberculosis and Lung Disease (The Union) and Medécins sans Frontières (MSF/Doctors Without Borders). The specific SORT IT program which resulted in this publication was jointly developed and implemented by: The Union South-East Asia Office, New Delhi, India; the Centre for Operational Research, The Union, Paris, France; The Union, Mandalay, Myanmar; The Union, Harare, Zimbabwe; MSF Luxembourg Operational Research (LuxOR); MSF Operational Center Brussels (MSF OCB); Jawaharlal Institute of Postgraduate Medical Education and Research (JIPMER), Puducherry, India; Post Graduate Institute of Medical Education and Research (PGIMER), Chandigarh, India; All India Institute of Medical Sciences (AIIMS), New Delhi, India; Velammal Medical College Hospital and Research Institute, Madurai, India; Society for Education Welfare and Action (SEWA)—Rural, Jhagadia, India; Common Management Unit (AIDS, TB & Malaria), Ministry of National Health Services, Regulations and Coordination, Islamabad, Pakistan; and Kidu Mobile Medical Unit, His Majesty's People's Project and Jigme Dorji Wangchuck National Referral Hospital, Thimphu, Bhutan. The authors would also like to thank the Vietnam National Tuberculosis Control Programme, Friends for International Tuberculosis Relief, Ho Chi Minh City, Vietnam, Haiphong International General Hospital for their support, the Stop TB Partnership's TB REACH for providing grant, and the community health workers for their help with data collection and follow-up of TB patients.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Contact Investigation of Multidrug-Resistant Tuberculosis Patients: A Mixed-Methods Study from Myanmar**

**Aye Mon Phyo 1,\* , Ajay M. V. Kumar 2,3,4 , Kyaw Thu Soe <sup>5</sup> , Khine Wut Yee Kyaw 2,6 , Aung Si Thu <sup>1</sup> , Pyae Phyo Wai <sup>1</sup> , Sandar Aye <sup>1</sup> , Saw Saw <sup>7</sup> , Htet Myet Win Maung <sup>8</sup> and Si Thu Aung <sup>8</sup>**


Received: 20 October 2019; Accepted: 22 November 2019; Published: 26 December 2019

**Abstract:** There is no published evidence on contact investigation among multidrug-resistant tuberculosis (MDR-TB) patients from Myanmar. We describe the cascade of contact investigation conducted in 27 townships of Myanmar from January 2018 to June 2019 and its implementation challenges. This was a mixed-methods study involving quantitative (cohort analysis of programme data) and qualitative components (thematic analysis of interviews of 8 contacts and 13 health care providers). There were 556 MDR-TB patients and 1908 contacts, of whom 1134 (59%) reached the health centres for screening (chest radiography and symptoms). Of the latter, 344 (30%) had presumptive TB and of them, 186 (54%) were investigated (sputum microscopy or Xpert MTB/RIF®). A total of 27 TB patients were diagnosed (six bacteriologically-confirmed including five with rifampicin resistance). The key reasons for not reaching township TB centres included lack of knowledge and lack of risk perception owing to wrong beliefs among contacts, financial constraints related to loss of wages and transportation charges, and inconvenient clinic hours. The reasons for not being investigated included inability to produce sputum, health care providers being unaware of or not agreeing to the investigation protocol, fixed clinic days and times, and charges for investigation. The National Tuberculosis Programme needs to note these findings and take necessary action.

**Keywords:** contacts; contact tracing; contact investigation; MDR-TB

#### **1. Introduction**

Tuberculosis (TB) is one of the top ten leading causes of deaths in the world. In 2017, there were an estimated 10 million TB patients including 558,000 with resistance to rifampicin (RR-TB), of which 82% had multidrug-resistant TB (MDR-TB, defined as resistance to at least rifampicin and isoniazid) [1]. Myanmar is one of the 30 countries classified as having a 'high MDR-TB burden' by the World Health Organization (WHO). There were an estimated 14,000 patients with MDR/RR-TB in 2017 in Myanmar, of whom only 3281 (23%) were reported to be diagnosed by the national TB programme (NTP) [1]. This means that a vast majority of MDR-TB patients remain undiagnosed or not reported to the NTP in Myanmar.

The "End TB strategy" of the WHO emphasizes early diagnosis and prompt treatment of all TB patients, including drug-resistant TB, to break the chain of transmission and prevent further spread of disease in the community [2]. In line with this, the STOP TB partnership proposes 90-(90)-90 targets (diagnosing and treating 90% of all people with TB, including 90% of the key populations at risk of TB, and achieving 90% treatment success for all people diagnosed with TB) [3]. One such key population group at high risk of TB is 'household contacts', living in close contact with source TB patients. Systematic reviews report that the pooled yield of active TB among the contacts is 4.5% [4,5]. A study from New York city reported a yield of active TB of 1% among household contacts investigated [6]. A systematic review among contacts of drug resistant TB patients showed a higher yield of 7.8%, with the majority of secondary cases having the same drug resistance or genotyping pattern as the source case, indicating primary transmission of drug-resistant strains of tuberculosis bacilli [7].

Hence, the WHO recommends 'contact investigation'—systematic investigation of all household contacts of source TB patients for active and latent tuberculosis and institution of appropriate curative and preventive treatment, respectively [8]. This strategy is endorsed by the NTP in Myanmar and it has been recommended that household contacts of MDR-TB patients with TB symptoms are investigated using Xpert MTB/RIF® assay since 2016 [9].

However, the implementation of this policy is poor and aggregate programme data indicate that contact investigation was done in only 30% of all bacteriologically-confirmed TB patients notified [10]. There is no published evidence about contact investigation among MDR-TB patients from Myanmar, as there is no structured, case-based, recording, reporting, and monitoring of this activity.

The International Union Against Tuberculosis and Lung Disease (The Union), an international non-governmental organization, started implementing a community-based MDR-TB care project in selected townships of Myanmar [11]. As part of this project, community volunteers have been trained and incentivized to conduct many activities including 'contact investigation' among MDR-TB patients. This provides an opportunity to assess the extent of implementation of contact investigation, as well as its barriers and possible solutions to address them.

Therefore, we undertook a mixed-methods operational research study with the following objectives: (1) Among the household contacts of MDR-TB patients registered from January 2018 to June 2019, to assess (i) the number and proportion of presumptive TB patients identified, investigated, diagnosed, and treated for TB; (ii) demographic and clinical factors associated with getting or not getting investigated; and (iii) the median duration between the various steps in the cascade. (2) To explore the barriers in implementing contact investigation from the perspective of household contacts and health care providers.

#### **2. Methods**

#### *2.1. Study Design*

This was an explanatory mixed-methods study design involving a quantitative component (a cohort analysis of routinely collected programme data) followed by a qualitative component (descriptive study) involving interviews of providers and contacts [12].

#### *2.2. Study Setting*

#### 2.2.1. General Setting

Myanmar is the second largest country in Southeast Asia, with a population of 52 million people (2014 Census) [13]. About two-thirds of the population lives in rural areas, while the urban populations are concentrated in Yangon and Mandalay regions. Administratively, Myanmar is divided into seven states, seven regions, and one union territory (Nay Pyi Taw), and subdivided into 74 districts with 330 townships [13].

#### 2.2.2. Specific Setting

The study was conducted in 27 selected townships of Mandalay Region, Magway Region, Sagaing Region, and Shan State of Myanmar, implementing the MDR-TB care project with funding support from the Global Fund for AIDS, Tuberculosis, and Malaria. Under this project, the key activities include evening direct observation of treatment for MDR-TB patients, provision of financial incentives to patients, counselling and monitoring of patients for adverse drug effects, health education to family members, and contact investigation. These activities are undertaken by the community volunteers or the project nurses hired and trained for this purpose. Community volunteers are people living in the same ward/village as the patients, but not family members; have reasonable education background (being able to read and write in Myanmar language); have time and interest to learn; and are committed to the care of patients. Peers who have had TB in the past are preferred as volunteers. Community volunteers are supervised by a project focal nurse (one per township) who coordinates with the staff at township level. Every volunteer is assigned a maximum of three MDR-TB patients for providing care. For MDR-TB patients who are not assigned a community volunteer, the project nurse of the respective township conducts the contact investigation, wherever possible.

Both community volunteers and focal nurses receive periodic training on contact investigation and its recording and reporting. The training includes steps to identify household contacts; conduct symptom screening; and refer for investigations, follow-up, and linkage to treatment if required.

#### 2.2.3. Household Contact Investigation

The process of contact investigation is described in Figure 1. First, the volunteer or the project nurse visits the home of MDR-TB patients and educates them about the importance of contact investigation. Then, the contacts are screened for TB symptoms (cough, fever, weight loss, night sweats, or enlarged lymph nodes) and, regardless of symptoms, they are referred to the township TB centre for chest radiography. If the patient is unable to visit, sputum samples are collected and transported. Irrespective of symptoms or chest radiography findings, people who are able to produce a sputum specimen are investigated further using sputum microscopy for acid-fast bacilli (AFB) and Xpert MTB/RIF® assay. People who are positive for AFB and/or positive for TB bacilli on Xpert MTB/RIF® assay are diagnosed as having TB and are started on first-line or second-line TB treatment, depending on the results of rifampicin resistance. Contacts who are unable to produce sputum or those with 'negative sputum results, but shadows suggestive of TB on chest radiography', are referred for further management to the physician, who makes a decision on clinical diagnosis of TB and treatment. In some township TB centres, the facilities for Xpert MTB/RIF® assay and chest radiography are not available. In such situations, contacts are referred to the nearest TB centre or hospital for investigation.

Household contacts are provided a maximum incentive of 7000 MMK (~5 US\$) if they undergo investigation. This is intended to cover the costs of transportation and some investigations like chest radiography, which may not be available free of charge in some townships.

**Figure 1.** Systematic screening and investigation algorithm for household contacts of index MDR-TB patients in the community-based MDR-TB care project in Myanmar, 2018–19. MDR-TB = multidrug resistant tuberculosis; TB = tuberculosis; CXR = chest X-ray; GXP = Xpert MTB/RIF®; TB symptoms = cough, fever, loss of weight, night sweat, and lymph node enlargement.

#### *2.3. Recording*

The information of index MDR-TB patients including the number of household contacts for each patient is captured in an MS Excel database. There is a "contact register" maintained at the township TB centres, which captures all the details of investigation, TB diagnosis, and treatment of contacts. This information is captured electronically in a quality-assured EpiData database by trained data entry operators and validated periodically by the project supervisors.

#### *2.4. Study Population*

#### 2.4.1. Quantitative

All household contacts of index MDR-TB patients newly registered in 27 project townships from January 2018 to June 2019 were included. In line with WHO guidelines, a household contact was defined as "a person who shares the same enclosed living space for one or more nights or for frequent or extended periods during the day with the source patient during the treatment or during the three months before commencement of the current treatment".

#### 2.4.2. Qualitative

The study population includes a purposive sample (maximum variation) of household contacts of MDR-TB patients, community volunteers, and project nurses from selected townships from each region/state. First, we calculated the township-wise Xpert MTB/RIF® testing rates among contacts with presumptive TB. We selected the township with the highest testing coverage and three townships with the lowest testing coverage in such a way that one township was selected from each region/state. In each selected township, two household contacts of MDR-TB patients (one who was investigated and one who was not), two community volunteers, and one project nurse were selected for interviews. In addition, we also interviewed the project supervisor. Thus, a total of 21 interviews were conducted. Participants who were knowledgeable, vocal, and willing to express were purposively selected. The sample size was guided by the saturation of the findings.

#### *2.5. Data Variables, Sources of Data, and Data Collection*

#### 2.5.1. Quantitative

The data were extracted from electronic databases of the project. The variables included symptoms, chest radiography findings, and results of sputum microscopy and Xpert MTB/RIF® assay. In addition, dates of start of treatment among index patients, contact registration, TB investigation, diagnosis, and treatment start among contacts were collected.

#### 2.5.2. Qualitative

Data collection was done between February and March 2019. Interviews were conducted at a time and place convenient to participants using an interview guide by K.T.S. (a medical doctor from the Department of Medical Research), and K.W.Y.K. (an operational research fellow from The Union), who are trained and experienced in qualitative research (Supplementary File S1) [14]. The guide was pilot tested before implementing in the field. Audio recording was done after receiving consent from participants. Verbatim notes were taken during interview. The average duration of interviews was approximately 45 min. After the interview was over, the summary of the interviews was read back to the participants to ensure participant validation.

#### *2.6. Data Analysis*

#### 2.6.1. Quantitative

We analysed using STATA software (version 14.2 STATA Corp., College Station, TX, USA). The demographic and clinical characteristics of the household contacts were summarized using median (inter-quartile range) for continuous variables and frequencies and proportions for categorical variables. The median time between the different stages of the process was calculated.

#### 2.6.2. Operational Definitions

People with either symptoms of TB and/or abnormal shadows on chest radiograph were considered as presumptive TB for this analysis. People who had undergone any of the diagnostic tests (sputum microscopy, Xpert MTB/RIF® assay, or fine needle aspiration cytology) were considered as having been investigated for TB. The date when these investigations were carried out was considered as 'date of investigation'. If a person underwent more than one investigation, the earlier date was considered. For bacteriologically-confirmed TB patients, the date of the positive test was considered as the date of diagnosis, whereas for clinically diagnosed patients, the date of chest radiography was considered as the date of diagnosis.

Factors associated with not being investigated for TB and getting tested with Xpert MTB/RIF® assay were measured using adjusted relative risks (RR) and 95% confidence intervals (CI). We initially tried to perform a log-binomial regression. As we did not obtain convergence, a modified Poisson regression with robust error variance was used. Variables that were significant (*p* value < 0.05) in unadjusted analysis or that were known to be associated with the outcome from published literature were included in the multivariable model.

#### 2.6.3. Qualitative

Transcripts were prepared in Myanmar language on the same day of interview based on the audio recordings and verbatim notes. Manual descriptive thematic analysis was performed by the principal investigator [15]. It was reviewed by a second investigator to reduce bias and subjectivity in interpretation. The decision of coding rules and theme generation was done in consensus among investigators. The analysis was done in Burmese language and only the final result was translated into English. The themes are presented for barriers and solutions with the corresponding quotes. Any difference between the investigators was resolved by discussion. We have adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines and 'Consolidated Criteria for Reporting Qualitative Research (COREQ) in conducting and reporting the study [16,17].

#### *2.7. Ethics Issues*

Ethics approval was obtained from the Ethics Review Committee, Department of Medical Research, Ministry of Health and Sports, Myanmar (Ethics/DMR/2018/159) and the Ethics Advisory Group of The Union, Paris, France (EAG number 48/18). Permission to conduct the study was obtained from the National Tuberculosis Programme, Ministry of Health and Sports, Myanmar. We obtained written informed consent for conducting the interviews and audio recording. A waiver of informed consent was obtained from the ethics committees for quantitative component as this included secondary data analysis.

#### **3. Results**

#### *3.1. Quantitative*

There were 556 MDR-TB patients who had 1908 contacts living with them. Of the latter, 1134 (59%) reached the township TB centre for screening (Figure 2). The median (inter quartile range, IQR) age of contacts was 30 (14–50) years and 664 (59%) were female.

#### 3.1.1. Cascade of Contact Investigation

Of the 1134 contacts, 344 (30%) had presumptive TB and of them, 186 (54%) were investigated. However, 213 individuals were found to be investigated even though they did not have any symptoms or abnormal chest radiography. Thus, a total of 399 people were investigated for TB and among them, 27 TB patients were diagnosed. Most of the clinically diagnosed cases belonged to the group with 'no symptoms, but positive findings on chest radiography' and nearly half of them were children. There was no TB patient diagnosed in the group without TB symptoms that had normal chest radiography. Barring one patient who died, all the remaining 26 (96%) patients started on the treatment (Figure 2).

The characteristics of TB patients are shown in Table 1. Of the 27 patients, six had bacteriologically-confirmed TB, while the rest were clinically diagnosed. Of the six bacteriologically-confirmed, five had pulmonary TB (all with rifampicin resistance) and one had extrapulmonary TB who was AFB-positive on Fine Needle Aspiration Cytology aspirate. All were new cases, barring one who reported a previous history of TB. A total of 362 contacts underwent sputum microscopy and only one was AFB-positive. Xpert MTB/RIF® assay was conducted among 176 contacts and 5 were diagnosed as TB (this included the one case diagnosed by sputum microscopy).

**Figure 2.** TB investigation, diagnosis, and treatment of household contacts of MDR-TB patients registered in a community based MDR-TB care project in Myanmar, between January 2018 and June 2019. TB = tuberculosis; MDR-TB = multidrug resistant tuberculosis; CXR = chest X-ray. The numbers in the shaded boxes indicate people with presumptive TB defined as those with symptoms (cough, fever, weight loss, and night sweats) and/or abnormal shadows on the chest radiograph.

**Table 1.** Characteristics of TB patients diagnosed among household contacts of MDR-TB patients registered in a community based MDR-TB care project in Myanmar, between January 2018 and June 2019.


TB = tuberculosis; MDR-TB = multidrug resistant tuberculosis.

#### 3.1.2. Factors Associated with Not Being Investigated for TB

Of 344 presumptive TB patients, 158 (46%) were not investigated for TB. In adjusted analysis, failure to do TB investigation was significantly higher among the contacts who were less than 15 years old, those who were registered in health facilities without an Xpert MTB/RIF® machine, and those who were referred when compared with those whose sputum was collected and transported to health facilities by project staff (Table 2).

**Table 2.** Factors associated with not being investigated for TB among household contacts with presumptive TB registered in a community-based MDR-TB care project in Myanmar, between January 2018 and June 2019.


TB = tuberculosis; MDR-TB = multidrug resistant tuberculosis; GXP = Xpert MTB/RIF® machine; CI = confidence interval; RR = relative risk; aRR = adjusted relative risk; *n* = number; NE = not estimated.\* = statistically significant. The variables that were significant in the unadjusted analysis and that were found to be associated in previous studies were included in the adjusted analysis. Fever was not included in the adjusted model owing to collinearity with cough.

### 3.1.3. Factors Associated with Getting Tested for Xpert MTB/RIF®

Of 344 presumptive TB patients, 121 (35%) were tested using Xpert MTB/RIF®. In the adjusted analysis, the Xpert MTB/RIF® testing was significantly lower among contacts aged less than 15 years and significantly higher in health facilities with an Xpert MTB/RIF® machine on-site (Table 3).


**Table 3.** Factors associated with GXP testing among household contacts with presumptive TB registered in a community-based MDR-TB care project in Myanmar, between January 2018 and June 2019.

TB = tuberculosis; MDR-TB = multidrug resistant tuberculosis; GXP = Xpert MTB/RIF® machine; CI = confidence interval; RR = relative risk; aRR = adjusted relative risk, *n* = number; NE = not estimated. \* = statistically significant. The variables that were significant in the unadjusted analysis and that were found to be associated in previous studies were included in the adjusted analysis.

#### 3.1.4. Delays

The median (IQR) duration between treatment start of index case to contact screening at the township TB centre was 81 (28–208) days. Among those investigated, 75% underwent the investigation within a day. The median time to treatment from diagnosis was 8 days—this was 14 days among bacteriologically-confirmed patients, but 4 days among clinically diagnosed patients (Table 4).

**Table 4.** Median duration (days) between different steps in the cascade of contact investigation among household contacts registered in community-based MDR-TB care project in Myanmar, between January 2018 and June 2019.


TB = tuberculosis; MDR-TB = multidrug resistant tuberculosis; IQR = inter quartile range.

#### *3.2. Qualitative*

Implementation barriers in contact investigation were multi-factorial and inter-related with each other. We organized the barriers under two broad themes—household contacts-related barriers and health system-related barriers. The barriers summarized here reflect the perspectives of both the household contacts and health care providers. Overall, the participants from the townships with low testing coverage reported a greater number of barriers and, more predominantly, health system barriers. The verbatim quotes (translated in English) are italicized and placed within double quotes.

3.2.1. Household Contact-Related Barriers

#### Unable to Visit the Clinic

Working people and school-going children were unable to visit the township TB centre for investigation because the clinic times conflicted with the work/school timings.

*"Some contacts were students. So they have to attend school from Monday to Friday. They can't come on these days for taking CXR (Chest X ray)."*

*(Community volunteer-5)*

*"Contacts did not want to go to OPD (Outpatient Department) because they didn't want to absent their jobs."*

*(Project Nurse-3)*

The other barrier was related to distance requiring a long time to travel, which was compounded by personal problems.

*"Some contacts couldn't come because they were very old and they lived far away"*

*(Project Nurse-1)*

*"I feel motion sickness when I travel* . . . *Therefore, I rarely travel"*

*(tested household contact-1)*

#### Inability to Produce Sputum

Some contacts could not produce sputum at all or only an inadequate amount of sputum for investigation.

*"Sayarma (The Nurse) gave the sputum cup to me and told to produce sputum. But I can't produce the sputum."*

*(Non-tested Household contacts-4)*

#### Financial Constraints

Some contacts, especially daily wage labourers, were reported to have financial constraints associated with visiting the township TB centre, as it meant absence from work and loss of daily wages, in addition to transportation charges. Although the project supported their travel allowance, it was a fixed amount and did not cover all the expenses.

*"They could not spend time for investigation. They are daily-wages workers. Therefore, they need to work for their daily income."*

*(Project Nurse-4)*

*"For the contacts who lived far away from township TB centre, there are higher transportation costs. Although project supports this cost, it is not enough for them."*

*(Project Nurse-3)*

#### Beliefs and Attitude

Some contacts refused to do contact investigation because they did not have any signs and symptoms and strongly believed that they do not have the disease. Others did not want to undergo investigation because they were afraid of possible side-effects of the TB drugs in the eventuality that they were diagnosed to have TB. One person mentioned that *God will take care of her illness*, even if it existed.

*"Contacts said that they believed that they have no disease (TB). So they don't want to test."*

*(Community volunteer-8)*

*"I heard TB patients are afraid of the injections and they can't withstand the side-e*ff*ects, so do I." (Not-tested household contacts-3)*

#### 3.2.2. Health System-Related Barriers

Lack of or Inadequate Counselling

Not all MDR-TB patients were assigned a volunteer in the project. So, the contact investigation may not have been done in such patients. The project nurse reported that some volunteers were not able to communicate and counsel effectively and convince the contacts to undergo TB screening.

*"Volunteers could not explain well about the importance of TB screening to contacts"*

*(Project Nurse-2)*

*"No one told me how to produce sputum"*

*(non-tested contact-4)*

The project supervisor reported that some of the contacts had already been investigated by the time volunteer visited the home, and hence were not referred. Such contacts were not recorded in the project database.

#### Do Not Know

Some of the health care providers at the township TB centre were not aware of the contact investigation protocol. So, chest radiography was not provided for asymptomatic contacts.

*"Even if the contacts reached the health facility, health care providers at TB centre did not o*ff*er chest X ray, because they had no signs and symptom of TB"*

*(Project Nurse-3)*

#### Do Not Agree

Some of the health care providers were aware of the protocol followed in the project, but did not agree, because it did not align with the NTP guidelines. While the project protocol advocated for screening using chest radiography in addition to symptom screening, NTP recommends only symptom screening and further investigation is limited to those with symptoms.

*"The TB focal person informed us that if there are no symptoms, we cannot do any investigation"*

*(Project nurse-3)*

#### They Do Not Do: High Workload

It was reported that investigations were not offered to contacts by the staff of the TB centre for various reasons. One was related to the high workload and shortage of human resources in laboratory unit in the township TB centre. Some of the staff at the township TB centre were unable to pay attention to the contact investigation as they were engaged with multiple responsibilities.

*"The laboratory technician position is vacant in TB centre"*

*(Project Nurse-2)*

*"The focal person does not involve fully in TB related activity as he also worked for other public health programmes. He is always busy"*

*(Project Nurse-4)*

#### They Do, But on Fixed Days and Times

Some laboratories had a fixed time to receive sputum specimens from the patients. If the patients arrived outside the times, they were asked to return the next day. Sputum specimens received outside the fixed times were discarded and this meant requesting contacts for additional specimens. In some health facilities, the chest X-ray unit imposed restrictions on the number of chest radiographs that could be taken on a given day (such as a maximum of 10 persons per day). All of the others were asked to come on the next day. This was very inconvenient for the contacts who had travelled from far off places. Similarly, there was a fixed day in a week for doctors to examine presumptive TB patients and make a decision about clinical diagnosis.

*"The laboratory accepts sputum sample between 9 am and 10 am only. Specimens received outside this time are discarded and then it is di*ffi*cult to request for additional specimens from contacts."*

*(Project Nurse-4)*

*"Chest X ray unit opens at 9 am and they allow only 10 persons per day to take chest X ray from TB department. Therefore, when the contact came and if it is beyond their maximum number, this person is asked to return the next day. And, the contact may not return."*

*(Project Nurse-1)*

#### They Do, But They Charge

It was reported that the contacts had to pay to undergo chest radiography in some places.

*"Chest X-ray fee is high. Here, it is 1500 MMK and this charge is higher in other township hospitals."*

*(Project Nurse-3)*

#### **4. Discussion**

This is the first study from Myanmar providing information on contact investigation among MDR-TB patients and its implementation challenges. We discuss the magnitude of gaps at each step of the cascade and their reasons below.

One of the main gaps was that nearly four in ten contacts did not reach the health facility for screening. This is higher than that reported from South Africa and similar to Ethiopia [18,19]. The possible reasons included lack of knowledge about the need for contact investigation, lack of risk perception owing to wrong beliefs, financial problems related to loss of wages and high transportation charges not entirely reimbursed by the project, and conflicts of clinic times with work/school times. It is possible that home visits and educating about contact investigation may not have happened in some MDR-TB patients. An interesting observation revealed during key informant interviews was that several contacts had already been investigated for TB by the time the project staff made home visits. Such people were not included in the numerator, but were counted in the denominator, when calculating this indicator, thus marginally overestimating the proportion not reached.

The next gap was at the level of screening and investigating the contacts who had reached the health facility. Only half of the presumptive TB patients received any investigation for bacteriological confirmation. The children were less likely to be tested, mostly because they were unable to produce sputum, and the gastric lavage was not routinely done in our setting, which requires hospitalization. Access to health facilities was another factor. The contacts who lived in townships that had Xpert MTB/RIF® facility were more likely to be tested as it reduced the travel cost and time.

Contacts whose sputum samples were collected at home and transported by project staffs were more likely to be investigated than contacts who had reached the health facility. Because investigation required two sputum samples, as per NTP guidelines, contacts referred to health facilities had to make multiple visits. Sometimes, the nearest health facility (to where the contacts were referred) did not have Xpert MTB/RIF assay services. In such instances, contacts had to be referred to another health facility with Xpert services. All these were reported as inconvenient and may have led to the losses in the cascade. While some TB focal persons at the township TB centre were unaware about the contact investigation protocol, some disagreed with the requirement of screening all contacts with chest radiography. There was also confusion among the providers about the eligibility criteria for prescribing the Xpert MTB/RIF® assay. The other barriers included fixed times and days for receiving sample or patients and demanding charges for investigations.

This study had several strengths and some limitations. First, we included a large sample of contacts covering 27 project townships of four states and regions. Thus, the findings are likely to be representative of the situation in these areas. Second, we used a mixed-methods study design, which helped in understanding the underlying reasons for the gaps in care cascade. Third, we used quality-assured data collected by project staff, which is routinely monitored and validated. Fourth, we achieved saturation in our qualitative interviews. Fifth, we followed the STROBE and COREQ guidelines for reporting the quantitative and qualitative components, respectively [16,17]. One limitation was that we had no information on 40% of household contacts who did not reach the health facility for screening; hence, we do not know if they were similar to those who reached the health facility. The impact of this on overall findings is unclear. Another limitation was that we did not interview the health care providers responsible for providing TB services and include their perspectives. This should be considered in future research.

Despite this limitation, our findings have many implications for programme policy and practice. First, we recommend that chest radiography be used for screening all household contacts regardless of TB symptoms, wherever possible, because the yield of TB was highest in the group with 'no symptoms, but abnormal chest radiograph'. Although most of the cases in this group were clinically diagnosed, there was one case of rifampicin resistance too. We may have diagnosed more cases of TB, had we tested everyone with Xpert MTB/RIF® assay. However, the feasibility of this recommendation needs to be tested before wider scale-up.

Second, we recommend that contacts 'without TB symptoms and normal chest radiograph' should not be investigated any further because there was zero TB in this group. This is also supported by evidence from systematic reviews [19]. A substantial number of patients were unnecessarily investigated and the resources could have been used elsewhere to increase the testing rates among presumptive TB patients.

Third, as shown in our study, the prevalence of drug resistant TB among contacts of MDR-TB patients is high in studies conducted elsewhere [7,20–22]. Hence, Xpert MTB/RIF® test should be the first diagnostic test of choice, as recommended by WHO [23]. There was no additional yield of TB owing to sputum microscopy in our study. Hence, we recommend discontinuing sputum microscopy for contacts of MDR-TB patients and focusing on the Xpert MTB/RIF® assay, as it can reduce the workload at township laboratories. This strategy can also be more convenient for the contacts, because sputum microscopy requires two specimens requiring multiple visits, whereas Xpert MTB/RIF® testing requires only one specimen.

Fourth, refresher training should be conducted periodically for community volunteers to improve their knowledge about contact investigation and counselling skills. The training content can be tailored to resolve the specific myths and beliefs among the contacts.

Fifth, efforts should be made to bring the contact investigation services closer to the community. This includes strengthening of sputum collection and transportation to township TB centres. However, this alone will not obviate the need for visiting health facility, as contacts also have to undergo chest radiography. To address this, we recommend exploring the possibility of using new technologies such as digital chest radiography with automated computer-aided detection of tuberculosis, which can be mounted in a mobile van for greater outreach [24].

In conclusion, we identified the magnitude of gaps in the cascade of contact investigation among MDR-TB patients in Myanmar, as well as reasons for the same. We hope these findings can be shaped into practical recommendations that will inform the NTP in Myanmar.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2414-6366/5/1/3/s1, S1: Interview Guide for Key Informant Interview.

**Author Contributions:** Conceptualization and protocol development: A.M.P., A.M.V.K., K.T.S., K.W.Y.K., A.S.T., P.P.W., S.A., S.S., H.M.W.M., and S.T.A.; Data Collection: A.M.P., K.T.S., and K.W.Y.K.; Data Analysis: A.M.P., A.M.V.K., K.T.S., and K.W.Y.K.; Writing the first draft: A.M.P., A.M.V., and K.T.S.; Critical review of the paper and final approval: A.M.P., A.M.V.K., K.T.S., K.W.Y.K., A.S.T., P.P.W., S.T.A., S.S., H.M.W.M., and S.T.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding. The training programme, within which this paper was developed, was funded by the Department for International Development (DFID), London, UK. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

**Acknowledgments:** This research was conducted through the Structured Operational Research and Training Initiative (SORT IT), a global partnership led by the Special Programme for Research and Training in Tropical Diseases at the World Health Organization (WHO/TDR). The model is based on a course developed jointly by the International Union Against Tuberculosis and Lung Disease (The Union) and Medecins Sans Frontieres (MSF/Doctors Without Borders). The specific SORT IT programme that resulted in this publication was jointly organised and implemented by The Centre for Operational Research, The Union, Paris, France; Department of Medical Research, Ministry of Health and Sports, Yangon; Department of Public Health, Ministry of Health and Sports, Nay Pyi Taw; The Union Country Office, Mandalay, Myanmar; The Union South-East Asia Office, New Delhi, India and London School of Hygiene and Tropical Medicine, London, UK.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Outcomes of Community-Based Systematic Screening of Household Contacts of Patients with Multidrug-Resistant Tuberculosis in Myanmar**

**Nang Thu Thu Kyaw 1,\* , Aung Sithu <sup>1</sup> , Srinath Satyanarayana 2,3 , Ajay M. V. Kumar 2,3,4 , Saw Thein <sup>5</sup> , Aye Myat Thi <sup>1</sup> , Pyae Phyo Wai <sup>1</sup> , Yan Naing Lin <sup>1</sup> , Khine Wut Yee Kyaw <sup>1</sup> , Moe Myint Theingi Tun <sup>1</sup> , Myo Minn Oo <sup>1</sup> , Si Thu Aung <sup>5</sup> and Anthony D. Harries 3,6**


Received: 16 October 2019; Accepted: 13 November 2019; Published: 25 December 2019

**Abstract:** Screening of household contacts of patients with multidrug-resistant tuberculosis (MDR-TB) is a crucial active TB case-finding intervention. Before 2016, this intervention had not been implemented in Myanmar, a country with a high MDR-TB burden. In 2016, a community-based screening of household contacts of MDR-TB patients using a systematic TB-screening algorithm (symptom screening and chest radiography followed by sputum smear microscopy and Xpert-MTB/RIF assays) was implemented in 33 townships in Myanmar. We assessed the implementation of this intervention, how well the screening algorithm was followed, and the yield of active TB. Data collected between April 2016 and March 2017 were analyzed using logistic and log-binomial regression. Of 620 household contacts of 210 MDR-TB patients enrolled for screening, 620 (100%) underwent TB symptom screening and 505 (81%) underwent chest radiography. Of 240 (39%) symptomatic household contacts, 71 (30%) were not further screened according to the algorithm. Children aged <15 years were less likely to follow the algorithm. Twenty-four contacts were diagnosed with active TB, including two rifampicin- resistant cases (yield of active TB = 3.9%, 95% CI: 2.3%–6.5%). The highest yield was found among children aged <5 years (10.0%, 95% CI: 3.6%–24.7%). Household contact screening should be strengthened, continued, and scaled up for all MDR-TB patients in Myanmar.

**Keywords:** multidrug-resistant tuberculosis; household contact; screening; TB diagnosis; yield; operations research

#### **1. Introduction**

Myanmar is one of the 30 high tuberculosis (TB) and multidrug-resistant TB (MDR-TB) burden countries in the world. In 2017, of the estimated 14,000 MDR-TB cases in Myanmar, 3281 were diagnosed and 2666 were enrolled for treatment, indicating a significant gap in case detection and treatment [1]. Similarly, of the estimated 191,000 TB cases, only 132,025 were notified and treated. To reduce the TB and MDR-TB burden, it is essential to diagnose TB and MDR-TB early and provide quality assured treatment [2,3]. Early diagnosis and treatment reduce morbidity, mortality, and transmission of TB and MDR-TB in the community.

Close contacts of active TB and MDR-TB patients are at high risk of TB infection and disease. A systematic review reported a pooled yield of 3.4% active TB among close contacts of active TB, with the incidence being highest during the first year of exposure [4]. Another systematic review reported that the prevalence of active TB among household contacts of drug-resistant TB was as high as 7.8% [5]. Hence, there is a strong recommendation to screen all household contacts of MDR-TB patients for active TB, and if they are diagnosed with active TB, to initiate them on treatment as soon as possible [6].

The International Union against TB and Lung Disease (The Union) has been implementing a community-based MDR-TB care (CBMDR-TBC) project in Myanmar to support the National Tuberculosis Programme's (NTP) programmatic management of DR-TB since 2015. Due to the high presumed prevalence of TB among household contacts of MDR-TB patients, a systematic screening algorithm including a combination of screening methods (symptoms and chest radiography) and diagnostic tests (sputum smear microscopy and Xpert MTB/RIF assay) to screen for active TB and MDR-TB was incorporated as a key component of the CBMDR-TBC project. In early 2016, community volunteers and focal nurses were trained under the CBMDR-TBC project to implement this screening for all household contacts of index MDR-TB patients in project townships, and the project started systematic data collection (which included dedicated recording and reporting systems) of this activity in March 2016.

To date, there has been no published report from Myanmar describing the process of screening household contacts of MDR-TB patients, how well the screening algorithm was followed, and the yield of active TB among those screened. Therefore, in this study we assessed: (a) the proportion of household contacts who were screened for active TB using the systematic screening algorithm (the proportion who were screened using symptoms and chest radiography and those with TB symptoms who were investigated for active TB using sputum smear microscopy and/or Xpert MTB/RIF assay), (b) the socio-demographic characteristics associated with screening of TB according to the algorithm, (c) the yield of active TB, and (d) socio-demographic characteristics associated with the diagnosis of active TB during one year of the implementation of the project.

#### **2. Materials and Methods**

#### *2.1. Study Design*

This was an analysis of routinely collected program data.

#### *2.2. Setting*

#### 2.2.1. Country Setting

Myanmar is a lower middle-income country with a population of 51 million. Geographically, the country is divided into 15 states/regions, which are further administratively divided into 412 townships. The country bears a high burden of MDR-TB along with its neighboring countries such as China, India, Bangladesh, and Thailand. The Programmatic Management of DR-TB was initiated in 2011 as part of the NTP's National Strategic Plan (2011–2015) to control TB in Myanmar [7]. Systematic contact tracing for all household contacts of MDR-TB patients is one of the key activities of the NTP's National Strategic Plan (2016–2020) [8]. The NTP recommends active screening of all household contacts of index MDR-TB patients and upfront use of Xpert MTB/RIF for investigating those with presumptive TB [9]. In order to facilitate Xpert MTB/RIF testing, the NTP has rolled out Xpert MTB/RIF machines in Myanmar since 2012, and by 2016 there were 65 Xpert MTB/RIF functional machines in the country.

#### 2.2.2. Project Description and the Implementation of Systematic Screening

The CBMDR-TBC project was started in 2015 in collaboration with The Union and NTP to support the treatment initiation and adherence among MDR-TB patients in 33 townships across four states/regions in the upper part of Myanmar. Details of the CBMDR-TBC project have been described elsewhere [10]. Briefly, under the project, each township has a focal nurse who visits the index MDR-TB patients' house monthly to monitor treatment adherence and side effects and provide health education and psychosocial support. The project also assigns a community volunteer for each patient who visits the patient's house daily in the evening to provide directly observed treatment. Focal nurses supervise the volunteers, and the project managers of the CBMDR-TBC project in turn supervise the focal nurses.

In 2016, a systematic screening of the household contacts of MDR-TB patients was incorporated into the project. The focal nurses and community volunteers were trained to implement the TB screening for all household contacts of index MDR-TB patients and the systematic data collection. A household contact is defined as "a person who shares the same enclosed living space for one or more nights or for frequent or extended periods during the day with the index case during the treatment or during the three months before the commencement of the current treatment".

The trained focal nurse facilitates the screening of household contacts for TB once the MDR-TB patients are diagnosed and every six months thereafter using a screening and investigation algorithm as described in Figure 1. During the home visit to the MDR-TB patients, the focal nurse screens each household member using a symptom-based questionnaire and refers these members for a chest radiograph. Those with symptoms and/or abnormal chest radiograph submit one early morning sputum sample for Xpert MTB/RIF and two samples, one spot and one early morning sputum, for smear microscopy for acid-fast bacilli (AFB). The focal nurses are trained to instruct contacts on how to expectorate sputum according to the guidelines. The instruction includes: First, rinse the mouth with clean water; second, take a deep breath in and out for three times; third, take one deep breath and cough forcefully; and finally, spit the sputum into the sputum container provided. Those with no symptoms are closely monitored unless the chest radiograph is abnormal, at which point the patient is referred to a TB specialist for further assessment. The contacts who are positive on smear microscopy for AFB and/or Xpert MTB/RIF assay are diagnosed with active TB. Those with negative results on smear microscopy or the Xpert MTB/RIF assay but have an abnormal chest radiograph are referred to the TB specialist for clinical evaluation and a decision on whether there is a clinical diagnosis of active TB. Each household contact is line listed and given a unique contact registration number. The nurse records the results of this screening process in the MDR-TB household contact screening register (Annex S1).

The focal nurses are responsible for screening the household contacts of MDR-TB patients as soon as the index MDR-TB patients are diagnosed. However, not all household contacts of index MDR-TB patients are screened immediately for several reasons, and therefore there could be a considerable delay between the diagnosis of an index case and the initial screening of contacts. In addition to 6-monthly systematic screening, the volunteers and nurses also check whether any TB-related symptoms have developed in household contacts during their regular home visits. If a household contact reports any TB-related symptom before the scheduled screening appointment, the focal nurse facilitates the evaluation of such patients for active TB in line with the algorithm. In addition, the focal nurses and volunteers provide TB health education and support for household infection control measures. They receive periodic training on systematic screening and investigation of active TB in household contacts conducted by the project managers using a standardized training package. The training includes steps to identify household contacts, the conduct of symptom screening, the referral of persons for diagnostic investigations for active TB and drug-resistant TB, follow up, linkage to treatment if required, education, and support for infection control measures.

**Figure 1.** Systematic screening and investigation algorithm for household contacts of index multidrug-resistant tuberculosis (MDR-TB) patients in the community-based MDR-TB care project in Myanmar. TB = tuberculosis; CXR = chest radiography; AFB = acid-fast bacilli; NTP = National TB Programme; DST = drug sensitivity testing.

Every month, the information from the MDR-TB household contact screening register is entered into an electronic database (developed using EpiInfo version 7.2 software) by the project's data entry operators, and the data are checked and validated by the monitoring and evaluation officer of the project.

#### *2.3. Study Sites and Population*

The study population includes all household contacts of index MDR-TB patients enrolled for TB screening in 33 townships of the CBMDR-TBC project in Upper Myanmar between April 2016 and March 2017. The index MDR-TB patients of the contacts included in this study were MDR-TB patients who were newly initiated on treatment between April 2016 and March 2017 as well as those who were already on treatment before April 2016.

#### *2.4. Sources of Data, Data Variables, and Data Collection*

We used secondary data routinely collected in the electronic database. Data variables of household contacts of index MDR-TB patients included: Contact registration number; registration date for contact screening; age; sex; history of previous TB; HIV status; history of diabetes mellitus; clinical information on symptoms such as cough, fever, weight loss, hemoptysis, lymph node enlargement, and night sweats; and results of diagnostic investigations such as sputum smear microscopy, Xpert MTB/RIF assay, and chest radiography and the treatment registration number of their index case. These data were extracted from the electronic database.

#### *2.5. Analysis and Statistics*

The demographic and clinical characteristics of the household contacts were described using numbers (proportions) and medians (interquartile ranges). We assessed the proportion of the household contacts with TB symptoms and of those, the proportion who underwent further sputum evaluation (smear AFB and/or Xpert test MTB/RIF). We used binomial logit models to study the association between measured demographic and clinical characteristics and the odds of further sputum evaluation according to the screening algorithm.

The yield/proportion of TB was calculated by dividing the number of TB or MDR-TB cases diagnosed by the number of household contacts screened for TB. We also calculated the yield of active TB across various measured demographic and clinical characteristics. The prevalence ratios of active TB across various measured demographic and clinical characteristics were estimated using binomial log models. STATA software (version 12.1, copyright 1985–2011 StataCorp LP, College Station, TX, USA) was used for all analysis. The 95% confidence intervals (CIs) for proportions, odds ratios, and prevalence ratios were adjusted for clustering at the township and household level using cluster robust standard error estimates.

#### *2.6. Ethics*

Ethics approval was received from the Myanmar Ethics Review Committee, Department of Medical Research, Ministry of Health and Sports, Myanmar (Approval number: Ethics/DMR/2017/084) and the Ethics Advisory Group of International Union Against Tuberculosis and Lung Disease, Paris, France (EAG number: 120/16). Permission to conduct the study was granted from the National Tuberculosis Programme, Ministry of Health and Sports, Myanmar.

#### **3. Results**

There were 620 household contacts of 210 index MDR-TB patients who were enrolled for systematic screening for active TB. Of those enrolled, all were screened for symptoms and 505 (81%) also underwent chest radiography. There were 240 (39%) contacts who had one or more TB symptoms and were eligible for sputum smear microscopy and Xpert MTB/RIF testing. Of those eligible, 169 (70%) underwent sputum smear microscopy and/or an Xpert MTB/RIF assay. The remaining 71 (30%) contacts did not undergo either of these tests, though some were evaluated clinically by the TB specialist (Figure 2). As a result of all these investigations and clinical evaluations, 24 contacts (3.9%, 95% CI: 2.3%–6.5%) were diagnosed with active TB (seven were bacteriologically confirmed, including two with Rifampicin-resistant TB, and 17 were clinically diagnosed). The number of household contacts screened to diagnose one case of active TB was 26 (95% CI: 15–44).

**Figure 2.** Number of household contacts of MDR-TB patients who underwent TB screening and investigations under the community-based MDR-TB Care Project in 33 townships in Myanmar, April 2016–March 2017. TB = tuberculosis; CXR = chest radiography; AFB = acid-fast bacilli.

#### *3.1. Characteristics Associated with Following the Systematic Screening Algorithm among Symptomatic Contacts*

The demographic and clinical characteristics of contacts with TB symptoms (*n* = 240) who underwent further evaluation by sputum tests (*n* = 169) versus those who did not undergo further evaluation by sputum tests (*n* = 71) are presented in Table 1. The age of the contact was the only characteristic that was statistically associated with whether contacts underwent further evaluation by sputum examination or not. Children aged less than 15 years were less likely to have had a sputum examination (either smear for AFB or Xpert test MTB/RIF), and contacts older than 49 years were more likely to have had a sputum examination when compared to contacts in the 15–49 year age group.


**Table 1.** Characteristics of symptomatic household contacts of MDR-TB patients, and their association with following the systematic screening algorithm under the community-based MDR-TB Care Project in 33 townships in Myanmar, April 2016–March 2017.

OR = odds ratio; CI = confidence interval; Ref = reference group; \* CIs are adjusted for clustering at household level as well as township level. † Column percentage; § Row percentage of total contact number. NA = not applicable.

#### *3.2. Characteristics Associated with Diagnosed with Active TB among Registered Contacts*

The demographic and clinical characteristics of 610 household contacts screened for active TB and the yield/prevalence of active TB in association with these characteristics are shown in Table 2. Overall 58% of the contacts were female, the median age of all contacts (IQR) was 31 (16–46) years and 40 (7%) contacts were children aged less than 5 years. Seventeen (3%) contacts had a previous history of TB, 6 (1%) had positive HIV status, and <1% of the contacts had a history of diabetes mellitus. Children aged less than 5 years had a significantly higher yield of TB when compared to contacts in the adult age groups. Since a small number of contacts were diagnosed with TB (*n* = 24), we did not perform a multivariable analysis to calculate the adjusted prevalence ratios.

**Table 2.** Demographic and clinical characteristics of household contacts of MDR-TB patients and the yield of TB among household contacts under the community-based MDR-TB Care Project in 33 townships in Myanmar, April 2016–March 2017.



**Table 2.** *Cont.*

PR = prevalence ratio; CI = confidence interval; Ref = reference group; \* CIs are adjusted for clustering at household level as well as township level. † Column percentage. § Row percentage of total contact number. NA = not applicable as PR cannot be calculated as there is zero prevalence in one of the two groups.

#### **4. Discussion**

This is the first study describing and evaluating the process of the systematic screening and investigation of household contacts of index MDR-TB patients in Myanmar. The study identified major gaps in the implementation of the screening as per the contact investigation algorithm. About 20% of all contacts enrolled were not screened by chest radiography. A third of the contacts with TB symptoms were not investigated by any sputum examination, and only one-fifth of contacts with TB symptoms were investigated by both sputum smear microscopy and the Xpert MTB/RIF assay. About 4% of contacts were diagnosed as having active TB disease. Children under 5 years of age who were contacts were more likely to be diagnosed with active TB. Since the study used routinely collected project data, we strongly believe that the findings can inform the national program in scaling up MDR-TB household contact screening in Myanmar.

There are a few limitations to the study. First, we did not have information on the total number of contacts of 210 index cases (the denominator) of which 620 were enrolled. This was due to a gap in our recording system that may have led to the focal nurses enrolling only those who they were able to meet and perform the symptom screening. Therefore, the gap between the number of contacts eligible and the number screened is likely to be higher than shown in our study. Second, as this study was cross-sectional in design, the results only provide an estimate of the prevalence of TB cases among contacts at a certain time period. Since the household contacts are more likely to develop TB anytime following exposure to the index case, a longitudinal study that provides information on both prevalent and incident cases would have provided much better estimates of the actual yield of TB among contacts. Third, due to the cross-sectional nature of the study and also since genotyping of contact's mycobacterial specimens was not done, we are unable to assess the temporal relationship between the exposure to the index patients and development of TB in the contacts, and therefore we cannot make any inferences about whether the TB disease diagnosed among contacts is due to TB transmission within the households. Fourth, about one-third of contacts did not undergo diagnostic evaluation according to the screening algorithm, and therefore the yield of MDR-TB cases among household contacts of MDR-TB patients in our study is an underestimate of the true yield. Finally, the study was based on routinely collected program data, and therefore there could be some errors in recording and reporting. We did not estimate the magnitude of these errors. However, we believe that due to the supervision and monitoring protocols in place, these errors are likely to be minimal and random, and therefore these errors are unlikely to have a major influence on the study results.

Despite these limitations, the study has some key findings to inform the program and future research. About 80% of contacts as per the screening algorithm underwent chest radiography. Anecdotally we were informed by the field-level health workers that this required substantial resources, time, and effort from them as well as the household contacts. Therefore, the large proportion of contacts who underwent chest radiography in our study may not be sustainable or replicable in routine practice. Therefore, whether chest radiography is required for all household contacts irrespective of the presence of symptoms is a subject matter for further exploration and future study. In this future study, we suggest that different screening and investigation algorithms are compared and tested for their efficacy and cost-effectiveness in detecting active TB among household contacts of MDR-TB patients [11].

One-third of household contacts with TB symptoms did not undergo further sputum evaluation. A study from South Africa also showed that less than half of the symptomatic household contacts of MDR-TB underwent further TB diagnosis evaluation [12]. Similarly, many other studies have reported high drop-out rates during TB contact investigation [13]. There are possible patient-level and health system-level barriers that prevent the systematic contact investigation algorithm from being followed. A study conducted in Vietnam reported that contacts and patients' knowledge, attitude, and practices regarding TB influenced continued engagement in the TB investigation process [14]. Another study from Uganda reported that stigma about TB, the constraint on time and space in clinics for counselling, mistrust of health-center staff by patients and contacts, and high travel costs for health staff to conduct contact screening and for contacts to travel to health facilities were barriers to implement TB contact screening [15].

In addition, not all contacts who were tested by sputum smear examination were tested by Xpert MTB/RIF. Studies have shown that Xpert MTB/RIF can detect up to 59% additional TB cases when compared to sputum smear microscopy [16–18]. It can also detect RR-TB as well as reduce the turnaround time from sample collection to diagnosis and treatment [19]. Inadequate access or lack of access to Xpert MTB/RIF machines was one of the main barriers for Xpert MTB/RIF testing for all eligible patients. During the study period, there were only 65 Xpert MTB/RIF machines in the country while there were 330 townships with an MDR-TB center. Many townships did not have Xpert MTB/RIF machines, and some of the townships were far from those that had a functioning machine for referral. The national program has a plan to increase the number of machines in the country (85 machines by the end of 2018), and the NTP's drug-resistant TB guidelines (February 2017) also recommend screening of household contacts of MDR-TB using Xpert MTB/RIF [9]. This could substantially reduce the barriers for Xpert MTB/RIF testing and increase the number of TB cases detected among the household contacts. Other patient- and provider-level barriers for accessing Xpert MTB/RIF should be explored in this context.

The prevalence of TB among household contacts in our study is similar to other studies conducted in high TB and MDR-TB burden countries [4,20–22]. However, we believe that due to several gaps in implementation and the limitations mentioned above, the prevalence in our setting is likely to be higher than what we observed in this study. In order to obtain more accurate estimates of the burden of TB among household contacts, the gaps and limitations identified in our study must be addressed. This includes close and active surveillance for 24 months for early detection of active disease in those who may be infected [23]. In addition, there is a need to support index patients to improve infection control measures at the household level, such as simple measures to improve cross-ventilation so that further transmission can be minimized [20,23,24].

We found that child contacts younger than five years had the highest risk of being diagnosed with TB. Although some studies and systematic reviews have reported that the yield among children is comparable to that seen in adult contacts [5,25], some studies have shown that there is a high prevalence of TB in children among contacts, as seen in our study [26,27]. This can be explained by the fact that young children are more likely to stay at home, which can increase exposure time especially if the index cases are their first-degree relatives [28]. Hence, it would be worthwhile to consider chemoprophylaxis in children after active TB is excluded to prevent the development of TB or MDR-TB [29–31]. Currently, we do not have national guidelines on how to manage TB infection in child contacts of MDR-TB, and there is no consensus on the preventive regimen for contacts of MDR-TB. Therefore, there is a need to develop guidelines to manage childhood contacts of patients with MDR-TB and to provide preventive therapy in Myanmar. In the meantime, as per the existing strategy, the program should maintain active surveillance of all contacts so as to detect and treat cases early [32].

### *Policy and Practice Implications*

The project needs to (1) strengthen the listing of all household contacts in the contact register and continue to record the results of the screening process in a systematic manner; (2) evaluate the efficacy and effectiveness of different contact screening algorithms and identify the most cost-effective and convenient algorithm that can be used in this setting; and (3) identify and address individual and system-level barriers for sputum smear examination and Xpert MTB/RIF testing.

#### **5. Conclusions**

The yield of TB (~4%) from screening household contacts of index MDR-TB patients was similar to what has been reported from other parts of the world. However, there were major gaps in screening according to the algorithm, and sputum smear microscopy and Xpert MTB/RIF testing were not done in all of the eligible contacts. The project should strengthen the systematic screening and investigation of TB in household contacts of MDR-TB patients, and the NTP should scale up the contact screening for all MDR-TB patients countrywide in order to achieve early detection and treatment of TB and MDR-TB.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2414-6366/5/1/2/s1, Annex S1: Household contact screening register used in the community-based MDR-TB Care Project in 33 townships.

**Author Contributions:** Conceptualization and protocol development: N.T.T.K., A.S., S.S., A.M.V.K., S.T., A.M.T., K.W.Y.K., S.T.A., A.D.H.; Data Collection: N.T.T.K., A.S.T., S.T., A.M.T., P.P.W., Y.N.L., K.W.Y.K., M.M.T.T., M.M.O.; Data Analysis: N.T.T.K., S.S., A.M.V.K., Y.N.L., K.W.Y.K., M.M.O., A.D.H.; Writing the first draft: N.T.T.K., S.S., A.D.H.; Critical review of the paper and final approval: N.T.T.K., A.S., S.S., A.M.V.K., S.T., A.M.T., P.P.W., Y.N.L., K.W.Y.K., M.M.T.T., M.M.O., S.T.A., A.D.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We thank all the clinical, administrative and program staff from the National Tuberculosis Programme, Department of Public Health and The Union office in Myanmar for their dedication in caring for patients and community and their contribution in collecting and providing data. We thank the Department for International Development (DFD), UK, for funding the Global Operational Research Fellowship Programme in which first author and co-authors (N.T.T.K., K.W.Y.K. M.M.O.) work as operational research fellows.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
