Determinants of Delayed Diagnosis and Treatment of Tuberculosis in Cambodia (2020) PDF

Teo et al. Infectious Diseases of Poverty (2020) 9:49 https://doi.org/10.1186/s40249-020-00665-8 RESEARCH ARTICLE Open Access Determinants of delayed diagnosis and tr...

Teo et al. Infectious Diseases of Poverty (2020) 9:49 https://doi.org/10.1186/s40249-020-00665-8 RESEARCH ARTICLE Open Access Determinants of delayed diagnosis and treatment of tuberculosis in Cambodia: a mixed-methods study Alvin Kuo Jing Teo1* , Chetra Ork2, Sothearith Eng2, Ngovlyly Sok2, Sovannary Tuot2, Li Yang Hsu1,3 and Siyan Yi1,2,4,5 Abstract Background: Cambodia is among the 30 countries in the world with the highest burden of tuberculosis (TB), and it is estimated that 40% of people with TB remain undiagnosed. In this study, we aimed to investigate the determinants of delayed diagnosis and treatment of TB in Cambodia. Methods: This mixed-method explanatory sequential study was conducted between February and September 2019 in 12 operational districts in Cambodia. It comprised of a retrospective cohort study of 721 people with TB, followed by a series of in-depth interviews. We assessed factors associated with time to TB diagnosis and treatment initiation using Cox proportional hazards model. Subsequently, we conducted in-depth interviews with 31 people with TB purposively selected based on the time taken to reach TB diagnosis, sex, and residence. Transcripts were coded, and thematic analyses were performed. Results: The median time from the onset of symptoms to TB diagnosis was 49 days (Interquartile range [IQR]: 21–112). We found that longer time to diagnosis was significantly associated with living in rural area (Adjusted hazards ratio [aHR] = 1.25; 95% confidence interval [CI]: 1.06–1.48); TB symptoms—cough (aHR: 1.52; 95% CI: 1.18–1.94), hemoptysis (aHR 1.32; 95% CI: 1.07–1.63), and night sweats (aHR: 1.24; 95% CI: 1.05–1.46); seeking private health care/self-medication (aHR: 1.23; 95% CI: 1.04–1.45); and higher self-stigma (aHR: 1.02; 95% CI: 1.01–1.03). Participants who received education level above the primary level were inversely associated with longer time to diagnosis (aHR: 0.78; 95% CI: 0.62–0.97). The median time from TB diagnosis to the initiation of treatment was two days (IQR: 1–3). The use of smear microscopy for TB diagnosis (aHR: 1.50; 95% CI: 1.16–1.95) was associated with longer time to treatment initiation. Seeking private health care and self-medication before TB diagnosis, lack of perceived risk, threat, susceptibility, and stigma derived qualitatively further explained the quantitative findings. Conclusions: TB diagnostic delay was substantial. Increasing public awareness about TB and consciousness regarding stigma, engaging the private healthcare providers, and tailoring approaches targeting the rural areas could further improve early detection of TB and narrowing the gap of missing cases in Cambodia. Keywords: Tuberculosis, Delayed diagnosis, Health-seeking behavior, Cambodia * Correspondence: [email protected] 1 Saw Swee Hock School of Public Health, National University of Singapore, National University Health System, Singapore, Singapore Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Teo et al. Infectious Diseases of Poverty (2020) 9:49 Page 2 of 12 Background Research Cambodia (NECHR) (NECHR reference: 024/ Tuberculosis (TB) is a leading infectious cause of mor- NECHR) and National University of Singapore (NUS) In- bidity and mortality worldwide, accounting for 10 mil- stitutional Review Board (IRB) (NUS IRB reference: H-19- lion new cases and 1.2 million deaths in 2018. 015) approved the study. Informed consent was obtained Cambodia is one of the 30 countries with the world’s from all respondents before study enrolment. highest burden of TB, with an estimated incidence of ac- tive TB of 302 (95% CI: 169–473) per 100 000 popula- Retrospective cohort study tion in 2018. Led by the National Center for Setting Tuberculosis and Leprosy Control (CENAT), the na- We conducted a retrospective cohort study on people with tional TB control program was set up to control and TB in 12 operational districts (OD) in 10 provinces in treat TB, including the introduction and country-wide Cambodia. We selected 100 health centers with a probabil- expansion of directly observed therapy short-course ity proportional to size (by the total population each health (DOTS) since 1994. As in many other high TB- center served) sampling without replacement from the total burden countries, passive case finding (PCF) is the de- number of health centers (n = 143) in the 12 ODs. fault setup for TB case finding at health centers where people with TB symptoms seek care, and the providers can identify the conditions. This strategy is, however, Study population inadequate to detect and measure the burden of undiag- In the selected ODs, trained data collectors recruited partic- nosed TB in the community. Active case finding (ACF), ipants aged 18 and above who were diagnosed with TB a strategy that has gained traction in recent years, was within 1 month of survey implementation, regardless of the found to increase case detection in Cambodia. Yet, history of previous TB treatment, HIV status, and the drug despite increased efforts to find the missing cases, every resistance status. Bacteriological status of participants was fourth person with TB goes undetected in Cambodia , determined by either smear microscopy or GeneXpert similar to the proportion reported globally in 2018. MTB/RIF. For participants who were tested negative, fur- Prolonged delays have been associated with further ther assessment and TB diagnosis were made by clinicians transmission of the infection in the community and based on clinical and laboratory grounds, and chest radio- thus posed a great challenge to TB elimination efforts glo- graphic abnormalities. TB workup and diagnoses were bally [6, 7]. Therefore, understanding the specific determi- done in accordance with the national guideline. We ex- nants of delayed diagnosis can be used as a practical guide cluded those who refused to participate. to enhance outreach programs, increase community en- gagement to reach missing cases, and to improve TB con- Key variables and definition trol strategies. Recent systematic reviews have reported We collected information on socio-demographic charac- empirical evidence associating socio-demographic, clinical, teristics, presence of other known medical conditions health system, and economic factors with delayed diagno- (one or more co-morbidities would constitute a yes ver- sis and treatment of TB [8–10]. In Cambodia, Sundaram sus no), TB symptoms before diagnosis (and the date of and colleagues have reported that strong preference for onset), knowledge and beliefs on TB, TB-related stigma private healthcare services, lack of awareness of TB symp- experiences, and psychological distresses using a paper- toms, and misbeliefs regarding TB (such as TB is a heredi- based questionnaire. We asked if study participants had tary disease and the ability to recover without treatment) sought healthcare prior to TB diagnosis and the facilities may delay seeking TB diagnosis. However, the socio- that they had visited. In this study, private health care fa- cultural and clinical determinants of delayed diagnosis cilities included private pharmacies, private general prac- have yet to be thoroughly examined. Furthermore, other titioners, private hospitals, and traditional healers. Self- individual-level factors such as stigma have been associ- medication referred to the use of medications without ated with health inequalities and a barrier to health care professional advice. Public health facilities referred to , but its impact on health-seeking decisions and TB government hospitals and health centers. We dichoto- diagnosis were inconclusive [13, 14]. In this study, we mized the level of residence urbanization into “urban” aimed to explore the determinants of delayed diagnosis and “rural” based on the Ministry of Planning’s framework and treatment of TB in Cambodia. [16, 17]. Using the WHO TB knowledge, attitude, and practices survey and the General Health Question- Methods naire (GHQ)-12 [19, 20], we measured participants’ TB We applied a mixed-method explanatory sequential study knowledge and the psychological distress they experienced design comprising a retrospective cohort study, followed before TB diagnosis, respectively (Supplementary Mate- by in-depth interviews (IDIs) with people with TB in rials). Study participants who scored above the median Cambodia. The National Ethics Committee for Health were regarded as having good TB knowledge. Teo et al. Infectious Diseases of Poverty (2020) 9:49 Page 3 of 12 We measured TB stigma using validated scales devel- assessed the determinants of diagnostic and treatment delay oped by Van Rie and colleagues [22, 23]. Self-stigma (12 using time-to-event analyses. The time to events —TB items) and perceived stigma by the community (11 diagnosis and initiation of TB treatment—were measured items) were measured on a Likert scale of 0 to 3 in days. In this study, events were regarded to have oc- (0 = “strongly disagree” and 3 = “strongly agree”). The curred when a TB diagnosis was made, or TB treatment scales measured thoughts and feelings of people with TB was initiated. As all the participants in this study were diag- (self-stigma) and the perceptions of people with TB re- nosed with TB and initiated on treatment, no data were garding how the community feels about people with TB censored. However, participants (n = 100) who started on (perceived stigma by the community). Based on the treatment on the same day (time to event = 0) as they were four-point Likert scale, the minimum score was 0, and diagnosed were not included in the risk set. Log-rank test the maximum possible score was 36 and 33, respectively. and univariate Cox proportional hazard regression were Summary stigma scores were standardized to 50 with used to estimate the statistical significance of categorical higher scores indicating a higher level of stigma. and continuous variables, respectively. Epidemiologically The main outcomes of interest in this study were diag- relevant variables and other exposure variables with P-value nostic and treatment delay. We defined diagnostic delay ≤ 0.1 in univariate analyses were included in the Cox pro- as the duration (time in days) between the onset of portional hazard model. We checked model fit using symptoms first recognized and self-reported by study Akaike Information Criteria and assessed the assump- participants and the final diagnosis of TB. Treatment tion of proportional hazards using Schoenfeld global test of delay was defined as the duration (time in days) between residuals , and no violations were observed (diagnostic TB diagnosis and treatment initiation. delay model: P = 0.428, treatment delay model: P = 0.840). We evaluated the fit of the final models using Cox-Snell re- Data collection procedures and quality assurance siduals. Hazard ratios (HR) were reported with 95% confi- Trained data collectors conducted face-to-face inter- dence interval (CI), and a 2-tailed P-value < 0.05 was views at the health centers with study participants using considered significant. All data were analyzed using STATA a paper-based questionnaire. We pre-tested the ques- 14 (StataCorp LP, Texas, United States of America). tionnaire with eight people with TB and TB survivors (data excluded from the main study). Data collectors In-depth interviews were trained to use prompts such as significant cultural Study population and public holidays in Cambodia to aid participants in Study participants for the qualitative IDIs were purpos- recalling the most accurate date of onset of TB symp- ively selected using maximum variation sampling from toms possible. Participants were reimbursed with United the cohort recruited in the survey described above. To States dollar (USD) 5 at the end of the interview. We ob- ensure equal representation of perspectives from persons tained information on the final diagnosis of TB and who experienced diagnostic delay and those who were treatment initiation date from the facilities where TB diagnosed early in the course of their disease, we se- diagnosis was made. To ensure the quality of the data, lected participants from the both extremes of the time the study team conducted several supervisory field trips, to diagnosis spectrum— < 15 days (short delay) and > and teleconference meetings with the field team were 100 days (long delay)—and stratified them according to held fortnightly. Before data entry, the questionnaires sex (men/women) and residence (urban/rural). We were checked for completeness. Four members of the re- approached 37 individuals, and six people refused to search team entered data in the questionnaires into participate due to illness and work priorities. In total, 31 KoBoToolbox (Harvard Humanitarian Initiative, Cam- people with TB were interviewed. bridge, Massachusetts, USA). Regular checks on the entered data were conducted. Inconsistencies were dis- Data collection cussed, and discrepancies were resolved by revisiting the We developed a semi-structured interview guide that original questionnaires. Data were exported into Micro- comprised of elements related to participants’ illness ex- soft Excel (Microsoft Office Professional Plus 2016, periences, diagnosis, and treatment-seeking behavior. Microsoft Corp., Redmond, Washington, USA) for data Questions and probes were focused on 1) knowledge cleaning before analyses. and perception of TB as a disease; 2) experiences with seeking diagnosis and treatment for TB, with clinical Statistical analyses and/or traditional approaches; and 3) barriers and facili- First, we described and presented the data using fre- tators experienced in seeking diagnosis and treatment. quencies and percentages for categorical variables and The semi-structured interview guide was pilot tested median with interquartile range (IQR) or mean with and revised accordingly. The IDIs were conducted in standard deviation (SD) for continuous variables. We four ODs in September 2019 by CO, SE, and NS, who Teo et al. Infectious Diseases of Poverty (2020) 9:49 Page 4 of 12 were trained in qualitative research. We matched the in- 0.878 and for the perceived stigma by the community terviewers and interviewees by sex to minimize potential scale was 0.877. biases in participants’ responses. Most of the interviews The median time from onset of symptoms to TB diag- were conducted 1∶1 in a private room at the health nosis was 49 days (IQR: 21–112). In univariate analysis, a centers nearest to participants’ homes. We also con- longer time to TB diagnosis was significantly associated ducted some interviews at participants’ homes, and on with TB symptoms (cough, hemoptysis, fever, weight some occasions, family members were present. All the loss, and night sweats), education level, seeking private interviews were conducted in Khmer, and each interview health care/self-medication, poor TB knowledge, experi- lasted between 30 to 60 min. Interviews were audio- encing psychological distress after falling sick (higher recorded, and field notes were taken during the inter- GHQ-12 scores), higher self-stigma, and TB diagnosis views. Participants’ received USD 10 in return for their informed by smear microscopy (Table 2). In the multi- time and effort. variate model adjusted for age, gender, and residence, a longer time to TB diagnosis remained significantly asso- ciated with rural residence (Adjusted hazards ratio Data analyses [aHR] = 1.25; 95% CI: 1.06–1.48, P = 0.01); TB symp- CO, SE, and NS transcribed all the IDIs verbatim. Tran- toms—cough (aHR = 1.52; 95% CI: 1.18–1.94, P = 0.001), scripts in Khmer were then translated to English for ana- hemoptysis (aHR = 1.32; 95% CI: 1.07–1.63, P = 0.01), lyses. The translated English transcripts were reviewed and night sweats (aHR = 1.24; 95% CI: 1.05–1.46, P = against the original version for verification by SE. Annota- 0.01); seeking private health care/self-medication (aHR = tion and analysis of complete transcripts were conducted 1.23; 95% CI: 1.04–1.45, P = 0.01); and higher self-stigma using NVivo (Version 10, QSR International, Burlington, (aHR = 1.02; 95% CI: 1.01–1.03, P = 0.003). Participants Massachusetts, USA). Textual references to the topics of who received education level above the primary level interest were retrieved and categorized using thematic were inversely associated with longer time to diagnosis analysis. AKJT and SE independently read and coded the (aHR = 0.78; 95% CI: 0.62–0.97, P = 0.03). transcripts characterized by 1) those with shorter delay to The median time from TB diagnosis to the initiation of diagnosis and 2) those with a longer delay to diagnosis. treatment was two days (IQR 1–3). In univariate analysis We derived initial themes based on the semi-structured (Table 2), a longer time to TB treatment initiation was sig- interview guide in developing a codebook of structural nificantly associated with TB symptoms (cough, and weight codes. We added emergent codes accordingly to the code- loss) and TB diagnosis informed by smear microscopy. In book. No new concepts relevant to the objective of this the multivariate model adjusted for age, gender, and resi- study were identified after interviewing 31 individuals, and dence, a longer time to treatment initiation remained sig- data saturation were reached. Conclusions were drawn nificantly associated only with the use of smear microscopy through the interpretation of the derived themes and the for TB diagnosis (aHR = 1.50; 95% CI: 1.16–1.95, P = 0.002). triangulation of qualitative data with quantitative findings. In-depth interviews Results Among the 31 IDIs, half were classified as having short Retrospective cohort study delays and were living in an urban setting. The median In total, 721 people with TB participated in this study age was 56 years (IQR: 45–68). We interviewed 18 men (Table 1). The median age was 61 years (IQR: 52–71), and and 13 women (Table 3). An overview of the results is 36.9% were living in an urban setting. The cohort com- presented in Table 4. prised of men (53.1%), individuals who were married (77.8%), and individuals who had at least primary educa- Barriers to TB care-seeking tion (84.5%). Forty percent of the participants were diag- Participants were not aware and felt that they were not nosed with bacteriologically confirmed TB, and at risk for TB and therefore delaying care-seeking. The GeneXpert MTB/RIF test informed most of the TB diag- lack of awareness and the perceived risk was com- nosis. The median distance from the place of residence to pounded by the absence of symptoms or features per- the nearest public health facility was four kilometers (IQR: ceived to be present among people with TB, such as 2–6). Sixty-six percent of the study participants sought severe cough and appearing underweight. For some par- private health care and/or self-medicated before TB diag- ticipants, the long-distance or inability to travel for ser- nosis. The median knowledge score reported by study par- vices was a barrier to care-seeking. ticipants was 9. The mean scores for self-stigma and perceived stigma by the community were 25.1 (SD = 6.9) “I doubted I would have TB. I only had a cough for and 26.0 (SD = 7.2), respectively, and they were normally several days and was exhausted.” (IDI17, 64yo, male, distributed. The Cronbach’s alphas for the self-stigma was urban, long delay) Teo et al. Infectious Diseases of Poverty (2020) 9:49 Page 5 of 12 Table 1 Participants characteristics and time from onset of Table 1 Participants characteristics and time from onset of symptoms to TB diagnosis symptoms to TB diagnosis (Continued) Variables Median IQR Variables Median IQR Distance from home to health facility, in kilometer 4 2–6 No 330 45.8 Age, in years 61 52–71 Chills Frequency % Yes 126 17.5 Total participants 721 100.0 No 595 82.5 Residence Weight loss Urban 266 36.9 Yes 482 66.9 Rural 455 63.1 No 239 33.1 Sex Night sweats Male 383 53.1 Yes 352 48.8 Female 338 46.9 No 369 51.2 Education levela Health care seeking prior to TB diagnosisc Primary and lower 605 84.5 Public health facilities 244 33.8 Above primary 111 15.5 Private health facilities/self-medication 477 66.2 Marital status TB knowledged Never married 31 4.3 Poor 179 24.8 Currently married 561 77.8 Good 542 75.2 Widowed/divorced 129 17.9 Perception about the seriousness of TB as a diseasea Ever smoked Very serious 275 38.3 Ever (current and past smokers) 216 30.0 Not very serious 444 61.7 Never 505 70.0 Self-perceived risk of getting TBa Current smoker Yes, at-risk 382 56.8 Current smoker 132 18.4 No, not at-risk 291 43.2 Not current smoker 587 81.6 Total General Health Questionnaire-12 scoree Current alcohol use a,b ≤3 364 50.5 Non-drinkers 509 71.1 >3 357 49.5 Drinkers 207 28.9 SD Standard deviation, IQR Interquartile range, TB Tuberculosis a Exclude missing values Presence of other known medical conditions b Drinkers reported frequency of alcohol use that ranges from once a month Yes 523 72.5 or less to 4 times or more per week. Non-drinkers refer to teetotalers c Participants were asked if they have sought health care prior to TB diagnosis No 198 27.5 and if so, the facilities that they have visited. Private health facilities referred to pharmacy, private general practitioner, and traditional healer. Self- Type of TB medication referred to the use of medications without professional advice. Bacteriologically confirmed TB 284 39.4 Public health facilities referred to government hospitals and health centers d Evaluated based on the answers from 8 questions regarding the Clinician diagnosed TB 437 60.6 characteristics, symptoms of TB, route of transmission, prevention, and treatment of TB with a total score of 13 (median = 9). Respondents were Type of diagnostic tests performed regarded as having poor TB knowledge if they scored the median and below GeneXpert 605 83.9 and good TB knowledge if they scored above the median e Evaluated based on the total score of the 6 negative items. Scoring method: Smear microscopy 116 16.1 0–0–1–1, with 0 = “less than usual”, 0 = “no more than usual”, 1 = “rather more than usual”, or 1 = “much more than usual” Cough Yes 638 88.5 No 83 11.5 Participants with short delay reported the lack of per- ceived threat of TB to their health and competing prior- Hemoptysis ities such as work and family commitments as barriers to Yes 124 17.2 TB care-seeking. Several participants with longer delay to No 597 82.8 diagnosis indicated care seeking at private health facilities Fever or self-medication when symptoms surfaced, many of Yes 391 54.2 which occurred frequently and repetitively prior to TB diagnosis. Table 2 Cox proportional hazard models by time (days) from onset of symptoms to TB diagnosis and from TB diagnosis to treatment initiation among people with TB in the study Characteristics Time from onset of symptoms to TB diagnosis (days) Time from TB diagnosis to treatment initiation (days) Crude HR 95% CI P-value Adjusted HR 95% CI P-value Crude HR 95% CI P-value Adjusted HR 95% CI P-value Residence Urban 1.00 1.00 1.00 1.00 Rural 1.46 1.25–1.70 < 0.001 1.25 1.06–1.48 0.010 0.92 0.78–1.08 0.296 0.95 0.80–1.13 0.562 Sex Male 1.00 1.00 1.00 1.00 Teo et al. Infectious Diseases of Poverty Female 1.02 0.88–1.18 0.760 0.91 0.77–1.06 0.217 1.05 0.90–1.23 0.517 1.06 0.90–1.24 0.504 Age, in years 1.00 1.00–1.01 0.657 1.00 0.99–1.00 0.259 1.00 0.99–1.00 0.642 1.00 0.99–1.01 0.604 Distance from home to the nearest 0.98 0.96–1.00 0.041 0.98 0.96–1.00 0.115 1.01 0.99–1.03 0.427 public health facility, in kilometers Education level (2020) 9:49 Primary and lower 1.00 1.00 1.00 Above primary 0.70 0.57–0.86 0.001 0.78 0.62–0.97 0.026 1.03 0.82–1.30 0.775 Cough Yes 1.00 1.00 1.00 1.00 No 1.56 1.24–1.97 < 0.001 1.52 1.18–1.94 0.001 0.74 0.58–0.94 0.015 0.80 0.62–1.03 0.078 Hemoptysis Yes 1.00 1.00 1.00 No 1.25 1.02–1.51 0.026 1.32 1.07–1.63 0.010 0.92 0.74–1.15 0.455 Fever Yes 1.00 1.00 1.00 No 1.28 1.10–1.48 0.001 1.14 0.97–1.34 0.113 0.88 0.75–1.04 0.125 Weight loss Yes 1.00 1.00 1.00 1.00 No 1.21 1.03–1.42 0.018 0.97 0.81–1.16 0.719 0.83 0.70–0.98 0.027 0.86 0.73–1.02 0.085 Night sweats Yes 1.00 1.00 1.00 No 1.42 1.22–1.65 < 0.001 1.24 1.05–1.46 0.010 0.89 0.76–1.05 0.164 a Health care seeking prior to TB diagnosis Public health facilities 1.00 1.00 1.00 Private health facilities/self-medication 1.27 1.09–1.48 0.003 1.23 1.04–1.45 0.014 0.91 0.77–1.07 0.265 TB knowledgeb Page 6 of 12 Table 2 Cox proportional hazard models by time (days) from onset of symptoms to TB diagnosis and from TB diagnosis to treatment initiation among people with TB in the study (Continued) Characteristics Time from onset of symptoms to TB diagnosis (days) Time from TB diagnosis to treatment initiation (days) Crude HR 95% CI P-value Adjusted HR 95% CI P-value Crude HR 95% CI P-value Adjusted HR 95% CI P-value Poor 1.00 1.00 1.00 Good 0.82 0.69–0.97 0.019 0.84 0.70–1.01 0.058 0.89 0.74–1.06 0.199 Total General Health Questionnaire-12 scorec ≤3 1.00 1.00 1.00 1.00 >3 1.26 1.08–1.46 0.002 1.14 0.97–1.13 0.113 0.87 0.74–1.02 0.089 0.88 0.75–1.04 0.125 Teo et al. Infectious Diseases of Poverty d Self-stigma 1.02 1.01–1.04 < 0.001 1.02 1.01–1.03 0.003 1.00 0.99–1.01 0.512 Type of diagnostic test GeneXpert 1.00 1.00 1.00 1.00 Smear microscopy 0.82 0.67–1.00 0.048 0.91 0.73–1.14 0.419 1.53 1.19–1.97 0.001 1.50 1.16–1.95 0.002 (2020) 9:49 HR Hazards ratio, CI confidence interval, ACF Active case finding, PCF Passive case finding, TB Tuberculosis a Participants were asked if they have sought health care prior to TB diagnosis and the facilities that they have visited. Private health facilities referred to pharmacy, private general practitioner, and traditional healer. Self-medication referred to the use of medications without professional advice. Public health facilities referred to government hospitals and health centers b Evaluated based on the answers from 8 questions regarding the characteristics, symptoms of TB, route of transmission, prevention, and treatment of TB with a total score of 13 (median = 9). Respondents were regarded as having poor TB knowledge if they scored the median and below and good TB knowledge if they scored above the median c Evaluated based on the total score of the 6 negative items. Scoring method: 0–0–1–1, with 0 = “less than usual”, 0 = “no more than usual”, 1 = “rather more than usual”, or 1 = “much more than usual” d Evaluated based on the answers from 12 questions, measured on a Likert scale (0 to 3), with 0 being strongly disagree and 3 being strongly agree. Minimum score is 0 and the maximum possible score is 36. Summary stigma scores from were standardized to 50 with higher scores representing higher level of self-stigma Page 7 of 12 Teo et al. Infectious Diseases of Poverty (2020) 9:49 Page 8 of 12 Table 3 Characteristics of in-depth interviews participants Facilitators to TB care-seeking Frequency % Participants from both groups expressed that encourage- Age, in years (median, IQR) 56 (45–68) ment from family members and other TB survivors was Sex an impetus to care-seeking, especially when participants were symptomatic and ill. Participants also reported that Male 18 58.1 success stories from TB survivors encouraged them to Female 13 41.9 seek care. Residence Urban 16 51.6 “My mother saw me coughing, and she said that I Rural 15 48.4 should go for [TB] screening.” (IDI12, 33yo, male, Time to TB diagnosis urban, long delay) Short delaya 16 51.6 Participants with shorter delays to diagnosis consistently Long delayb 15 48.4 reported that fear of infecting others and the ill effects of IQR Interquartile range, TB Tuberculosis the conditions on wellbeing, work, and livelihoods a Time from onset of TB symptoms to TB diagnosis: < 15 days b Time from onset of TB symptoms to TB diagnosis: > 100 days prompted TB care-seeking. Many were worried that as they grew weaker, they were unable to go to work and the financial impact that would ensue. Further- more, the economic woes worsened when significant “I didn’t go, I just didn’t care. Until it got really ser- out-of-pocket payment was required for medical con- ious, then I go [to seek care]” (IDI11, 76yo, female, sultations and medications in the private sector, rural, short delay) therefore impelled care-seeking in public health facil- ities where TB services are available for free. Partici- “I wanted to go as well. If I have a lot of time, I pants who experienced longer delay to diagnosis just wanted to go to the hospital to get treatment explained that they only sought care when TB symp- to be cured of that disease, and no longer sick. toms became intolerable. But I really didn’t have time, to be honest.” (IDI3, 37yo, male, urban, short delay) “We don’t have enough money because when I got sick, I spent 2 thousand to 3 thousand dollars [on “I never went to hospital. When I was [had] multiple consultations and treatments]. I couldn’t go cough, I was buying [bought] medicines from to work. I stayed at home.” (IDI8, 32yo, female, pharmacy.” (IDI26, 68yo, male, rural, long delay) rural, short delay) Table 4 Comparison of themes emerged from qualitative interviews between people with shorter delays and people with longer delays to TB diagnosis Themes Similarities Short delay Long delay Barriers to TB Lack of perceived risk and susceptibility Lack of perceived threat of TB Sought private health care/self-medicated care-seeking to TB to their own health prior to TB diagnosis Unable to travel to health facilities due Competing priorities – work, to intolerable symptoms or logistical livelihood, and family responsibilities constraints Facilitators to TB Encouragement from family members, Fear of infecting others Intolerable symptoms care-seeking and other TB survivors Easy access to TB diagnosis provided Perceived illness would affect health, by non-governmental organizations wellbeing, work, and livelihood Reasons for seeking Better access (facilities are closer Perceived illness to be less serious Perceived to provide better care and private healthcare/ to home) service, and the medicines are more self-medicate effective Reasons for seeking Financially affordable Perceived to provide better and Experienced good service in the past public healthcare safer care TB stigma Felt ashamed/embarrassed because Perceived stigma and discrimination of TB against people with TB TB Tuberculosis Teo et al. Infectious Diseases of Poverty (2020) 9:49 Page 9 of 12 TB stigma recent meta-analysis on TB delayed diagnosis conducted Another theme that emerged from both groups was per- among lower-middle-income countries. We found ceived stigma against TB. Those with shorter delays to significant correlations between TB knowledge and par- diagnosis indicated that they felt ashamed and embar- ticipants’ perception of the seriousness of TB (P < 0.001) rassed because of TB. and their self-perceived risk of contracting TB (P = 0.01) (Supplementary Table 1). This relationship was further “I was ashamed. I felt embarrassed because I had explained by IDI participants as barriers to TB care- TB.” (IDI1, 50yo, male, urban, short delay) seeking. Participants who had lower education levels might have limited access to information about TB and Many participants with longer delay to diagnosis re- therefore be less aware of the consequences of TB to ported both self-perceived and experience of discrimin- themselves and their close contacts. ation perpetrated by people around them. Several Our results showed that respondents who reported ex- accounts of participants distanced by members of their periencing onset symptoms of hemoptysis and cough community after they were diagnosed were reported. were less likely to delay their diagnosis. In conformity with two systematic reviews [8, 32], hemoptysis was per- “I always think like that. I thought that this disease ceived as more serious than cough, and people with is similar to HIV. People in my village discriminate these symptoms were more likely to seek care immedi- against these diseases.” (IDI23, 58yo, male, rural, ately. IDI participants who experienced longer delay to long delay) diagnosis also identified that care-seeking at public health facilities was only impelled when symptoms be- “As I was sick, that one knows that I was sick, they came unbearable. We found contrasting results for the didn’t dare to talk to us. They walked out while we onset symptoms of cough. Most studies reported a posi- walked in; so, it’s the effect, they know we like this tive relationship between cough and delayed diagnosis as [living with TB], they discriminated us.” (IDI12, it overlaps with the manifestations of other respiratory 33yo, male, urban, long delay) infections and smoking [36–38]. Unfortunately, we did not collect data regarding concurrent respiratory infec- Discussion tions during the onset of cough. Also, we did not find a In this study, we found that the median time from the significant correlation between smoking and cough in onset of symptoms to TB diagnosis among people with this study (P = 0.64). TB in Cambodia was 49 days, and it is comparable to re- Comparable to the finding from several systematic re- cent studies conducted in other high TB-burden coun- views [8, 32, 39], we found respondents who sought pri- tries in Asia [27, 28]. We found respondents living in vate health care and/or had self-medicated before TB rural areas of the country were more likely to experience diagnosis were 23% more likely to experience a longer a delay in diagnosis. This is consistent with findings time to diagnosis. The overarching theme of private from other low- and lower-middle-income countries health care seeking and/or self-medicating before TB [29–31], and high TB-burden settings. Despite the diagnosis derived from IDIs was a prominent barrier to similar risk profiles for TB between urban and rural early diagnosis. In Cambodia, free TB services are largely areas in Cambodia , delayed TB diagnosis in rural provided by the public healthcare sector, where persons Cambodia where large portions of the population are lo- diagnosed with TB are registered and receive treatment. cated raised concerns regarding the differences in The private health providers mostly do not provide direct structural and social determinants between the dwellings TB care and anti-TB medications, and the notification rate in obtaining timely diagnosis and treatment. IDI partici- remains poor [15, 40]. However, 67% of the Cambodian pants from rural areas highlighted physical access and population prefer to seek first treatment in the private logistical barriers in TB care-seeking. However, specific healthcare sector due to ease of access, drugs supply, and challenges faced by the population living in rural areas responsiveness [41, 42], with more than two-thirds of the should be further investigated. In Cambodia, different out-of-pocket payments for health care going to the pri- active case finding strategies targeting a specific setting vate sector. Hence, the private sector plays a signifi- and/or population have shown promising results in in- cant role in the cascade of care, and revitalizing and creasing case yields [4, 35], but the impact of these strat- contextualizing the public-private mix programs will be egies on early identification of TB should be thoroughly instrumental in finding the missing cases in Cambodia examined. Our analyses also showed that participants through referrals and public education [40, 44]. who had education above the primary level and those This study found that TB stigma contributed to de- with good TB knowledge (statistically not significant) layed TB diagnosis and IDI participants further elabo- had a shorter time to diagnosis, and it is in line with a rated on the shame, embarrassment, and discrimination Teo et al. Infectious Diseases of Poverty (2020) 9:49 Page 10 of 12 that they both perceived and experienced in the course survey. The distance from home to the nearest public of their TB care pathway. However, investigations on the health facility reported in this study is also comparable to relationship between stigma and diagnostic delay have other studies conducted in Cambodia [33, 54, 55]. Hence, produced mixed results. In Thailand, the impact of we believe the limitation on generalizability is minimal. In TB stigma on time to diagnosis was found to be non- the assessment of private health care engagement and significant. However, the effect of stigma on health- self-medication prior to TB diagnosis, we were not able to seeking behaviors was more pronounced in China , distinguish the frequency of these actions in this study. in accord with a 2014 systematic review. The dis- Therefore, we took the binary approach of either sought agreements could be contributed by the use of different private health care and/or had self-medicated before TB means of measurement. As stigmatization is a cul- diagnosis or not in the final analysis. To our knowledge, tural and social phenomenon , utilization of vali- this is the first study to assess the extent and determinants dated measurement tools and incorporation of of TB diagnostic and treatment delay in Cambodia. As qualitative assessment of TB stigma will provide a more there is no universal cut-off for diagnostic and treatment holistic insight into the role of stigma in affecting care- delay, we analyzed the outcome of interest in its continu- seeking behaviors. Nevertheless, the development of ous form to preserve as much information possible. contextualized interventions aimed at addressing stigma In the qualitative study, we undertook a long versus short will be pivotal in shortening the time to TB diagnosis. delay analysis approach—with equal representation by The median time from TB diagnosis to treatment initi- gender and urban/rural dwellings for multifarious per- ation was two days, and this finding showed that people spectives. We adopted the explanatory sequential design with TB in Cambodia were timely treated in accordance in this study for a comprehensive understanding of the with the national guideline (not exceeding three days findings, and the cogency of findings is further strength- from the day of diagnosis). In this study, 75% of the ened through the corroboration of both quantitative and participants-initiated TB treatment within 3 days from qualitative data. the day of diagnosis. We did not find significant differ- ences between urban and rural dwellers regarding treat- Conclusions ment delay. This reflects the success of a decentralized In conclusion, we found that people with TB experi- DOTS program to all health centers in Cambodia, enced substantial delays before being diagnosed with TB. resulting in greater access and shorter delay to TB treat- Seeking private healthcare and self-medication before ment [15, 48]. We found participants diagnosed with TB TB diagnosis; absence of TB symptoms such as cough, by smear microscopy were 50% more likely to experi- hemoptysis, and night sweats; stigma; rural residence; and ence a longer time to treatment initiation. This is con- a lower education level were responsible for longer delays. gruent with a longer reporting time required by smear Delays in TB care-seeking for more than 2 months have microscopy compared with GeneXpert MTB/RIF. shown to elevate the risk of household infection, and in- The utilization of point-of-care diagnostic tools such as fectivity declined rapidly after the initiation of treatment GeneXpert MTB/RIF should be further optimized to im-. Longer treatment delay was noted among TB diagno- prove case detection, timely diagnosis, and treatment of ses made using sputum-smear microscopy. Increasing TB, and cost-effectiveness [50–52]. public awareness about TB and consciousness regarding This study has several limitations. The onset of symp- stigma, engaging the private healthcare providers, and tai- toms was self-reported by study participants and subjected loring approaches to target the rural areas could further to recall bias. To reduce the recall bias, we trained data improve early detection of TB and narrowing the gap of collectors to interview participants within 1 month of missing cases in Cambodia. diagnosis. Significant cultural and public holidays in Cambodia were used as prompts to aid recall and facilitate Supplementary information the estimation of the date of the onset of symptoms. Sec- Supplementary information accompanies this paper at https://doi.org/10. ond, our sampling frame did not include all the ODs in 1186/s40249-020-00665-8. Cambodia, hence limiting the generalizability of the find- Additional file 1. ings to all people with TB in Cambodia. To assess the rep- resentativeness of our sample, we compared our findings with other nationally representative studies. Despite an Abbreviations ACF: Active case finding; aHR: Adjusted hazard ratios; CENAT: National Center overrepresentation of participants living in urban settings for Tuberculosis and Leprosy Control; CI: Confidence interval; DOTS: Directly (36.9% vs 21.4% in the national census ), the ratio of observed treatment, short-course; GHQ: General health questionnaire; men to women (1.3), and the proportion of bacteriologic- HIV: Human immunodeficiency virus; IDI: In-depth interview; IQR: Interquartile range; MTB/RIF: Mycobacterium tuberculosis/rifampicin resistance; ally confirmed TB by age groups (Supplementary Table 2) NECHR: National Ethic Committee for Health Research; NUS IRB: National were comparable to the previous national TB prevalence University of Singapore Institutional Review Board; OD: Operational district; Teo et al. Infectious Diseases of Poverty (2020) 9:49 Page 11 of 12 PCF: Passive case finding; SD: Standard deviation; TB: Tuberculosis; 6. Jassal MS, Bishai WR. The epidemiology and challenges to the elimination USD: United States dollar; WHO: World Health Organization of global tuberculosis. Clin Infect Dis. 2010;50:S156–64. 7. The Global Fund to Fight AIDS, Tuberculosis and Malaria. Make a global Acknowledgments priority of finding missing cases of tuberculosis. 2017. https://www. We would like to thank the study participants for their contribution to this theglobalfund.org/en/blog/2017-10-10-make-a-global-priority-of-finding- project. We would also like to acknowledge the contribution of the National missing-cases-of-tuberculosis/. Accessed 11 July 2018. Center for Tuberculosis and Leprosy Control Cambodia, Dr. Chansophal Ly, 8. Getnet F, Demissie M, Assefa N, Mengistie B, Worku A. Delay in diagnosis of Mr. Seyha Ong, and the field staff in coordinating the training and data pulmonary tuberculosis in low-and middle-income settings: systematic collection processes. We would like to thank Dr. Kiesha Prem for her advice review and meta-analysis. BMC Pulm Med. 2017;17:202. on the study design and statistical methods used in this study. 9. de Vries SG, Cremers AL, Heuvelings CC, Greve PF, Visser BJ, Bélard S, et al. Barriers and facilitators to the uptake of tuberculosis diagnostic and Authors’ contributions treatment services by hard-to-reach populations in countries of low and AKJT, LYH, and SY conceptualized the study, designed the study and medium tuberculosis incidence: a systematic review of qualitative literature. developed the study methods. AKJT, LYH, and SY wrote the grant Lancet Infect Dis. 2017;17:e128–43. application to obtain funding. AKJT, NS, LYH, and SY obtained ethical 10. Teo AKJ, Singh SR, Prem K, Hsu LY, Yi S. Determinants of delayed diagnosis and approvals for the project. AKJT, OC, SE, NS, and ST oversaw project treatment of tuberculosis in high-burden countries: a mixed-methods implementation, data collection, curation, and processing. AKJT and SY systematic review and meta-analysis. 2020. https://www.researchsquare.com/ analyzed the data and interpreted the results. AKJT, CO, SE, and ST article/6dbe6955-78e6-49ec-9249-a0ae9baffe8c/v1. Accessed 14 Mar 2020. contributed to the drafting of the manuscript. All authors reviewed and 11. Sundaram N, James R, Sreynimol U, Linda P, Yoong J, Saly S, et al. A strong approved the final manuscript. TB programme embedded in a developing primary healthcare system is a lose-lose situation: insights from patient and community perspectives in Cambodia. Health Policy Plan. 2017;32:ii32–42. Funding 12. Craig GM, Daftary A, Engel N, O’Driscoll S, Ioannaki A. Tuberculosis stigma as This project is supported by the Saw Swee Hock School of Public Health a social determinant of health: a systematic mapping review of research in Infectious Diseases Program Research Grant and the National University of low incidence countries. Int J Infect Dis. 2017;56:90–100. Singapore President’s Graduate Fellowship. The funders had no role in the 13. Kipp A, Pungrassami P, Stewart P, Chongsuvivatwong V, Strauss R, Van Rie design of the study and collection, analysis, and interpretation of data and in A. A prospective study of TB and AIDS stigma as barriers to TB treatment writing the manuscript. adherence using validated stigma scales. Int J Tuberc Lung Dis. 2011;15: 1540–i. Availability of data and materials 14. Courtwright A, Turner AN. Tuberculosis and stigmatization: pathways and The datasets used and/or analyzed during the current study are available interventions. Public Health Rep. 2010;125:34–42. from the corresponding author on reasonable request. 15. National Center for Tuberculosis and Leprosy Control (CENAT). Technical guidelines on tuberculosis control. 2nd ed. Phnom Penh: Ministry of Health; 2016. Ethics approval and consent to participate 16. United Nations Population Fund. Report on urbanization and its linkage to This study was approved by the National Ethics Committee for Health socio-economic and environmental issues. Phnom Penh: Cambodia; 2014. Research Cambodia (NECHR reference: 024/NECHR) and the National 17. Asian Development Bank. Cambodia: urban sector assessment, strategy and University of Singapore Institutional Review Board (REF No. N-19-015). In- road map. Manila: Asian Development Bank; 2012. formed consent was obtained from all respondents before study enrolment. 18. World Health Organization. Advocacy, communication and social mobilization for TB control: a guide to developing knowledge, attitude, and Consent for publication practice surveys. Geneva: World Health Organization; 2008. Not applicable. 19. Goldberg DP, Gater R, Sartorius N, Ustun TB, Piccinelli M, Gureje O, et al. The validity of two versions of the GHQ in the WHO study of mental illness in Competing interests general health care. Psychol Med. 1997;27:191–7. The authors declare that they have no competing interests. 20. Goldberg D, Williams P. A user’s guide to the general health questionnaire. Windsor: NFER-Nelson; 1988. Author details 21. Desale A, Ali I, Esmael A, Yaregal Z, Desta K, Agonafir M. Assessment of 1 Saw Swee Hock School of Public Health, National University of Singapore, patients’ knowledge, attitude, and practice regarding pulmonary National University Health System, Singapore, Singapore. 2KHANA Center for tuberculosis in eastern Amhara regional state, Ethiopia: cross-sectional Population Health Research, Phnom Penh, Cambodia. 3Yong Loo Lin School study. Am J Trop Med Hyg. 2013;88:785–8. of Medicine, National University of Singapore and National University Health 22. Van Rie A, Sengupta S, Pungrassami P, Balthip Q, Choonuan S, Kasetjaroen System, Singapore, Singapore. 4Center for Global Health Research, Touro Y, et al. Measuring stigma associated with tuberculosis and HIV/AIDS in University California, Vallejo, USA. 5School of Public Health, National Institute southern Thailand: exploratory and confirmatory factor analyses of two new of Public Health, Phnom Penh, Cambodia. scales. Tropical Med Int Health. 2008;13:21–30. 23. Meershoek A, Zwerling A, Daftary A, Citro B, Smyth C, Lewis D, et al. TB stigma Received: 4 March 2020 Accepted: 23 April 2020 measurement guidance. Den Haag: KNCV Tuberculosis Foundation; 2018. 24. KoBoToolbox. KoBoToolbox: data collection tools for challenging environments. 2019. https://kobotoolbox.org/. Accessed 4 Oct 2019. References 25. Akaike H. A new look at the statistical model identification. IEEE Trans 1. World Health Organization. Global tuberculosis report 2019. Geneva: World Autom Control. 1974;19:716–23. Health Organization; 2019. 26. Schoenfeld D. Chi-squared goodness-of-fit tests for the proportional hazards 2. Khieu K, Ito K, Hamajima N. Experience in tuberculosis treatment through regression model. Biometrika. 1980;67:145–53. directly observed therapy short course in health centres and communities 27. Martinez L, Xu L, Chen C, Sekandi JN, Zhu Y, Zhang C, et al. Delays and in Cambodia. Public Health. 2007;121:696–9. pathways to final tuberculosis diagnosis in patients from a referral Hospital 3. World Health Organization. Systematic screening for active tuberculosis: in Urban China. Am J Trop Med Hyg. 2017;16:0358. principles and recommendations. Geneva: World Health Organization; 2013. 28. Ahmad RA, Mahendradhata Y, Utarini A, de Vlas SJ. Diagnostic delay 4. Morishita F, Eang MT, Nishikiori N, Yadav R-P. Increased case notification amongst tuberculosis patients in Jogjakarta Province, Indonesia is related to through active case finding of tuberculosis among household and the quality of services in DOTS facilities: diagnostic delay amongst TB Neighbourhood contacts in Cambodia. PLoS One. 2016;11:e0150405. patients in Jogjakarta Province. Tropical Med Int Health. 2011;16:412–23. 5. Golub JE, Bur S, Cronin WA, Gange S, Baruch N, Comstock GW, et al. 29. Kahsay A, Gedefaw M, Asres M, Weldu Y. Patients’ delay in seeking health Delayed tuberculosis diagnosis and tuberculosis transmission. Int J Tuberc Care for Tuberculosis Diagnosis in East Gojjam zone, Northwest Ethiopia. Lung Dis. 2006;10:24–30. Am J Trop Med Hyg. 2017;96:1071–5. Teo et al. Infectious Diseases of Poverty (2020) 9:49 Page 12 of 12 30. Satyanarayana S, Nair SA, Chadha SS, Sharma G, Yadav S, Mohanty S, 53. National Center for Tuberculosis and Leprosy Control (CENAT). Second et al. Health-care seeking among people with cough of 2 weeks or national prevalence survey, 2011. Phnom Penh: Ministry of Health; 2012. more in India. Is passive TB case finding sufficient? Public Health 54. Jacobs B, Bajracharya A, Saha J, Chhea C, Bellows B, Flessa S, et al. Making Action. 2012;2:157–61. free public healthcare attractive: optimizing health equity funds in 31. Chanda-Kapata P, Kapata N, Masiye F, Maboshe M, Klinkenberg E, Cambodia. Int J Equity Health. 2018;17:88. Cobelens F, et al. Health Seeking Behaviour among Individuals with 55. Hwang WJ, Park YM. Factors influencing the accessibility of maternal health Presumptive Tuberculosis in Zambia. PLoS One. 2016;11:e0163975 van Service in Cambodia. Int J Environ Res Public Health. 2019;16:2909. Zyl-Smit R, editor. 56. Altman DG, Royston P. The cost of dichotomising continuous variables. BMJ. 32. Li Y, Ehiri J, Tang S, Li D, Bian Y, Lin H, et al. Factors associated with patient, 2006;332:1080. and diagnostic delays in Chinese TB patients: a systematic review and meta- 57. Riley RL, Moodie AS. Infectivity of patients with pulmonary tuberculosis in analysis. BMC Med. 2013;11:156. inner city homes. Am Rev Respir Dis. 1974;110:810–2. 33. Prem K, Pheng SH, Teo AKJ, Evdokimov K, Nang EEK, Hsu LY, et al. Spatial and temporal projections of the prevalence of active tuberculosis in Cambodia. BMJ Glob Health. 2019;4:e001083. 34. National Institute of Statistics. Cambodia inter-censal population survey 2013. Phnom Penh: Ministry of Planning; 2014. 35. Eang MT, Peou S, Yadav RP, Morishita F, Nishikiori N, van Maaren P, et al. Early detection of tuberculosis through community-based active case finding in Cambodia. BMC Public Health. 2012;12:1–9. 36. Yimer S, Holm-Hansen C, Yimaldu T, Bjune G. Health care seeking among pulmonary tuberculosis suspects and patients in rural Ethiopia: a community-based study. BMC Public Health. 2009;9:454. 37. Machado AC, Steffen RE, Oxlade O, Menzies D, Kritski A, Trajman A. Fatores associados ao atraso no diagnóstico da tuberculose pulmonar no estado do Rio de Janeiro. J Bras Pneumol. 2011;37:512–20. 38. Xia D, Zhang Z, Li X, Jiang C, Ma J, Ding S, et al. Factors associated with patient delay among new tuberculosis patients in Anqing, China. Biomed Res. 2016;27:8. 39. Storla DG, Yimer S, Bjune GA. A systematic review of delay in the diagnosis and treatment of tuberculosis. BMC Public Health. 2008;8:15. 40. Bell CA, Eang MT, Dareth M, Rothmony E, Duncan GJ, Saini B. Provider perceptions of pharmacy-initiated tuberculosis referral services in Cambodia, 2005–2010. Int J Tuberc Lung Dis. 2012;16:1086–91. 41. National Institute of Statistics (NIS). Demographic health survey 2014. Phnom Penh: Ministry of Planning; 2015. 42. Berendes S, Heywood P, Oliver S, Garner P. Quality of private and public ambulatory health care in low and middle income countries: systematic review of comparative studies. PLoS Med. 2011;8:e1000433 Jenkins R, editor. 43. Ministry of Health Cambodia, World Health Organization Regional Office for the Western Pacific. Health service delivery profile. Cambodia; 2012. http:// www.wpro.who.int/health_services/service_delivery_profile_cambodia.pdf. Accessed 14 Oct 2019. 44. Lei X, Liu Q, Escobar E, Philogene J, Zhu H, Wang Y, et al. Public–private mix for tuberculosis care and control: a systematic review. Int J Infect Dis. 2015; 34:20–32. 45. Pungrassami P, Kipp A, Stewart P, Chongsuvivatwong V, Strauss R, Van Rie A. Tuberculosis and AIDS stigma among patients who delay seeking care for tuberculosis symptoms. Int J Tuberc Lung Dis. 2010;14:181–7. 46. Duan Q, Wang P, Lv J, Zhong R, Wang W. Health care seeking among pulmonary tuberculosis suspects in Wuhan: A community-based study. Int J Med Med Sci. 2013;5:273–8. 47. Chang S-H, Cataldo JK. A systematic review of global cultural variations in knowledge, attitudes and health responses to tuberculosis stigma. Int J Tuberc Lung Dis. 2014;18(6):168–73. 48. Saly S, Onozaki I, Ishikawa N. Decentralized DOTS shortens delay to TB treatment significantly in Cambodia. Kekkaku. 2006;81:467–74. 49. Lee H-S, Kee S-J, Shin J-H, Kwon Y-S, Chun S, Lee JH, et al. Xpert MTB/RIF assay as a substitute for smear microscopy in an intermediate-burden setting. Am J Respir Crit Care Med. 2019;199:784–94. 50. Creswell J, Codlin AJ, Andre E, Micek MA, Bedru A, Carter EJ, et al. Results from early programmatic implementation of Xpert MTB/RIF testing in nine countries. BMC Infect Dis. 2014;14:2. 51. Laushkina Z, Cherednichenko A. Estimation of the role of the test GeneXpert MTB-RIF in the diagnosis of pulmonary tuberculosis. Eur Respir J. 2014;44:P2677. 52. Pallas SW, Courey M, Hy C, Killam WMP, Warren D, Moore B. Cost analysis of tuberculosis diagnosis in Cambodia with and without Xpert® MTB/RIF for people living with HIV/AIDS and people with presumptive multidrug-resistant tuberculosis. Appl Health Econ Health Policy. 2018; 16:537–48.