Skip to main content
Advertisement
  • Loading metrics

Clinic presentation delay and tuberculosis treatment outcomes in the Lake Victoria region of East Africa: A multi-site prospective cohort study

  • Grace E. Mulholland ,

    Roles Conceptualization, Data curation, Formal analysis, Methodology, Software, Visualization, Writing – original draft, Writing – review & editing

    gem@unc.edu

    Affiliation Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, United States of America

  • Michael E. Herce,

    Roles Conceptualization, Methodology, Writing – review & editing

    Affiliation Institute for Global Health and Infectious Diseases, School of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, United States of America

  • Brenda A. Okech,

    Roles Investigation, Project administration, Supervision, Writing – review & editing

    Affiliation UVRI-IAVI HIV Vaccine Program Limited, Entebbe, Uganda

  • Kidola Jeremiah,

    Roles Investigation, Project administration, Supervision, Writing – review & editing

    Affiliation Mwanza Research Centre, National Institute for Medical Research, Mwanza, Tanzania

  • Ubaldo M. Bahemuka,

    Roles Investigation, Project administration, Supervision, Writing – review & editing

    Affiliation Medical Research Council/Uganda Virus Research Institute & London School of Hygiene and Tropical Medicine, Uganda Research Unit, Entebbe, Uganda

  • Zachary A. Kwena,

    Roles Investigation, Project administration, Supervision, Writing – review & editing

    Affiliation Kenya Medical Research Institute, Nairobi, Kenya

  • Gertrude Nanyonjo,

    Roles Investigation, Project administration, Supervision, Writing – review & editing

    Affiliation UVRI-IAVI HIV Vaccine Program Limited, Entebbe, Uganda

  • Janet Seeley,

    Roles Writing – review & editing

    Affiliations Medical Research Council/Uganda Virus Research Institute & London School of Hygiene and Tropical Medicine, Uganda Research Unit, Entebbe, Uganda, Global Health and Development Department, London School of Hygiene and Tropical Medicine, London, United Kingdom

  • Audrey Pettifor,

    Roles Conceptualization, Writing – review & editing

    Affiliation Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, United States of America

  • Michael Emch,

    Roles Conceptualization, Writing – review & editing

    Affiliations Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, United States of America, Department of Geography, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, United States of America

  • Sharon S. Weir,

    Roles Conceptualization, Writing – review & editing

    Affiliation Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, United States of America

  • Jessie K. Edwards

    Roles Conceptualization, Funding acquisition, Methodology, Writing – review & editing

    Affiliation Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, United States of America

Abstract

In the Lake Victoria region of East Africa, little is known about delays between tuberculosis (TB) symptom onset and presentation at a clinic. Associations between clinic presentation delay and TB treatment outcomes are also poorly understood. In 2019, we abstracted data from routine TB treatment records for all adults (n = 776) initiating TB treatment in a 6-month period across 12 health facilities near Lake Victoria. We interviewed 301 cohort members and assessed whether they experienced a clinic presentation delay longer than 6 weeks. We investigated potential clinical and demographic correlates of clinic presentation delay and examined the association between clinic presentation delay and an unfavorable TB treatment outcome (death, loss to follow-up, or treatment failure). Clinic presentation delay was common, occurring among an estimated 54.7% (95% CI: 48.9%, 61.2%) of cohort members, though no specific correlates were identified. Clinic presentation delay was slightly associated with unfavorable TB treatment outcomes. The 180-day risk of an unfavorable outcome was 14.2% (95% CI: 8.0%, 20.4%) among those with clinic presentation delay, compared to 12.7% (95% CI: 5.1%, 20.3%) among those presenting earlier. Multi-level community-based interventions may be necessary to reduce clinic presentation delays in communities near Lake Victoria.

Introduction

Effective tuberculosis (TB) treatment rapidly reduces infectiousness [13], curtailing onward transmission of M. tuberculosis [4, 5] and substantially reducing the risk of mortality and other poor clinical outcomes [6]. TB remains, however, among the world’s most deadly infectious diseases and a leading cause of death in low- and lower-middle-income countries [7]. Many countries with high TB incidence rely heavily on passive case finding [8, 9]. This is a patient-initiated path to diagnosis which necessitates that a person experiences TB symptoms, determines a need for care, and seeks care at a health facility with TB diagnostic capabilities [10]. This path presents multiple opportunities for delays between the onset of symptoms and presentation at a health facility. Clinic presentation delay may be especially common in hard-to-reach populations with limited access to health care services, such as in fishing communities near the shores of Lake Victoria [11].

Clinic presentation delay may contribute to or predict suboptimal treatment outcomes, yet few studies have explored this relationship. In this study, we describe clinic presentation delay among people who initiated TB treatment at 12 public sector health facilities in the Lake Victoria region of Kenya, Tanzania, and Uganda. We also estimate associations between clinical and demographic characteristics and clinic presentation delay, and we describe reported reasons for clinic presentation delay. Finally, we estimate the association between clinic presentation delay and unfavorable TB treatment outcomes.

Methods

Study design

We conducted a longitudinal cohort study with data collection proceeding in three main stages at each of the 12 health facilities included in the study. First, the study population (“cohort”) was enumerated from health facility records, and basic demographic and clinical data were abstracted for members of the cohort. Second, a subset of cohort members (a “subcohort”) was recruited for face-to-face interviews, and clinic presentation delay was ascertained among members of the subcohort. Third, at study closure, available TB treatment outcomes were abstracted for members of the cohort.

Setting and participants

The study was conducted at 12 public sector health facilities located near the shores of Lake Victoria. Facilities were selected according to the following criteria: high volume of people on TB treatment, availability of or linkage to Xpert MTB/RIF diagnostic testing, capacity to initiate anti-TB treatment for people with drug-susceptible and drug-resistant TB, maintenance of person-level TB treatment records, and willingness to facilitate records access. Facilities were in Kyotera, Kalangala, and Wakiso districts in Uganda; in Suba, Mbita, Karungu, and Macalder sub-counties in Kenya; and in Shirati, Tarime, and Bukoba districts in Tanzania (Fig 1). All people ages 18 and older who initiated TB treatment in the 6 months preceding the start of data collection at the selected health facilities were included in the cohort.

thumbnail
Fig 1. Health facilities where records were abstracted to enumerate the study cohort.

The study team abstracted basic clinical and demographic data for a census of people initiating TB treatment over a 6-month period at these 9 health centers and 3 hospitals. A subset of cohort members (a “subcohort”) was recruited from across the 12 facilities for additional data collection via face-to-face interviews. Data were collected in the East Africa TB/HIV and Mobility Study (2019). Map data are from OpenStreetMap (openstreetmap.org/copyright); map tiles are by Stamen Design (http://maps.stamen.com/#watercolor/8/-1.300/32.900).

https://doi.org/10.1371/journal.pgph.0002259.g001

Stage 1: Cohort enumeration and abstraction of demographic and clinical data for cohort members

The study team visited the 12 selected health facilities and consulted TB treatment registers to enumerate the cohort and abstract data on key variables. At each facility, the study team identified, as cohort members, all people who were 18 years of age or older and who initiated TB treatment within the 6-month period preceding the first date of data collection at the health facility. Because the timing of data collection varied somewhat, across the 12 facilities, TB treatment initiation dates among cohort members included in the study ranged from December 2018 to November 2019.

Stage 2: Recruitment of a subcohort and ascertainment of clinic presentation delay among members of the subcohort

The next stage of data collection involved interviewing a subset of cohort members (a “subcohort”) to assess clinic presentation delay and additional characteristics not captured in routine health records. We set health facility-specific recruitment targets to ensure that the distribution of subcohort members across the health facilities would reflect that of cohort members and that the subcohort would include an approximately equal number of people with and without TB-associated HIV. The study team coordinated with staff at each health facility to integrate the subcohort recruitment process into routine client intake procedures. When a person on TB treatment presented for care, the health facility staff identified whether the individual met the inclusion criteria for the study cohort. If so, they referred the client to the study team who confirmed cohort membership and administered consent and recruitment procedures. Subcohort recruitment continued consecutively as people presented for care until the facility-specific targets were reached.

In face-to-face interviews, subcohort members reported on characteristics not recorded in health facility records, including the exposure, clinic presentation delay. We defined clinic presentation delay categorically in the interview as waiting more than 6 weeks after TB symptom onset to seek care at a health facility. A six-week threshold has been used elsewhere to characterize a prolonged period between onset of TB symptoms and first consulting a healthcare provider [12]. Clinic presentation delay was assessed retrospectively; all cohort members had initiated TB treatment by the time of the interview. Subcohort members were asked if they experienced clinic presentation delay, and if so, why they did not present to a health facility sooner.

Stage 3: Abstraction of TB treatment outcomes for cohort members

At study closure, which occurred between 3 and 6 months after enumeration of the cohort, the study team returned to the health facilities to abstract cohort members’ TB treatment outcomes. A phone-based tracing study was conducted to confirm treatment outcomes among subcohort members who either initiated treatment at least 6 months before study closure but had no outcome recorded, or had recorded outcomes of “lost to follow-up” or “not evaluated.” Outcomes were updated to the tracing-confirmed outcome for each of the 75 subcohort members successfully traced.

Ethical considerations

The protocol was approved by the Institutional Review Board of the University of North Carolina at Chapel Hill; the Uganda Virus Research Institute (UVRI) Research Ethics Committee; the Uganda National Council for Science and Technology and the London School of Hygiene and Tropical Medicine Research Ethics Committee; the Kenya Medical Research Institute (KEMRI) Scientific and Ethics Review Unit; and the Medical Research Coordinating Committee of the National Institute for Medical Research (NIMR), Tanzania. The cohort was enumerated from TB treatment registers under a waiver of informed consent; therefore, the health facility staff redacted sensitive information such as name, patient record number, and contact information from the materials and the study team created the cohort from the redacted records. The study team obtained written informed consent from the cohort members who were recruited for the subcohort survey. Members of the subcohort were compensated for transportation and their time.

Inclusivity in global research

Additional information regarding the ethical, cultural, and scientific considerations specific to inclusivity in global research is included in the S1 Checklist.

Statistical analysis

We conducted our analyses using SAS 9.4 [13] and R 4.0.3 [14]. One cohort member with missing HIV status was excluded from all analyses. We used map data from OpenStreetMap, licensed under the Open Data Commons Open Database License (ODbL) by the OpenStreetMap Foundation (OSMF) [15]. Map tiles are by Stamen Design, licensed under CC BY 3.0 [16].

Imputation of missing clinic presentation delay and covariate data.

Clinic presentation delay and detailed covariate data were missing by design for cohort members who did not participate in the subcohort survey, and a small number of subcohort members (n = 9; 3.0%) had an undefined time to clinic presentation. Missing values were imputed using multiple imputation by fully conditional specification [17]. We included expected predictors of the value or missingness of clinic presentation delay and all variables from the analysis models. The full set of variables included in the imputation process, coding details, and the variable-specific imputation methods used are provided in the S1 Table. Models for each incomplete variable included all other variables in the table. We generated 200 complete data sets, and for all analyses using the multiply imputed data, we used Rubin’s rules [18] to summarize point estimates and standard errors.

Predictors of clinic presentation delay.

We estimated the prevalence of clinic presentation delay in the cohort in each imputation by fitting an intercept-only log-binomial model. We also fit separate log-binomial models within strata of clinical and demographic characteristics, and we computed prevalence ratios (PRs) to assess the association between clinic presentation delay and these characteristics. Finally, we reported the proportion of subcohort members with clinic presentation delay who experienced delay for various reasons.

Estimating risk of an unfavorable TB treatment outcome by clinic presentation delay.

We used the Nelson-Aalen estimator [19, 20] to compute the risk of an unfavorable TB treatment outcome within 180 days of treatment initiation. We followed cohort members from their date of TB treatment initiation until the date of an unfavorable treatment outcome, transfer out, outcome recorded as “unknown” or “not evaluated,” study closure (i.e., the date of outcome data collection), or 180 days, whichever occurred first. Unfavorable TB treatment outcome was a composite outcome that included death, loss to follow-up, and treatment failure. We estimated risks up to 180 days to align with the typical 6-month drug-susceptible TB treatment duration. Missing death dates (n = 6) were imputed as the midpoint between the last recorded contact with the health facility and study closure. Missing dates for loss to follow-up (n = 14) were imputed as 60 days beyond the last recorded contact with the health facility, to approximate a treatment interruption of two months [21].

We estimated the 180-day risk of an unfavorable TB treatment outcome overall and within strata defined by HIV status and sex. Then, we calculated risk differences, comparing risks among those with and without clinic presentation delay. We used the delta method [22] to estimate standard errors of the risk differences. We also conducted two sensitivity analyses. In one, we repeated the main analysis among only the subset of cohort members with measured (i.e., non-imputed) values for clinic presentation delay. In a second sensitivity analysis, we estimated the association between clinic presentation delay and the most common unfavorable TB treatment outcome, death.

Results

Cohort and subcohort characteristics are presented in Table 1. Most (86.1%) of the 775 cohort members had pulmonary TB, over 90% were new TB cases, and half (49.7%) had TB-associated HIV. About 40% of cohort members were female, and the median age was 39 (IQR: 29–50) years. TB was bacteriologically confirmed among 47.4% of the cohort members, and drug-resistant TB was recorded for less than 3%. The subcohort (n = 301, 38.8% of the cohort) was generally similar to the full cohort with respect to the distribution of abstracted variables.

thumbnail
Table 1. Characteristics of subcohort and cohort members.

Basic demographic and clinical characteristics are presented for all people who initiated TB treatment over a 6-month period (the “cohort”). Additional characteristics are presented for cohort members who participated in face-to-face interviews (the “subcohort”).

https://doi.org/10.1371/journal.pgph.0002259.t001

Clinic presentation delay was reported by 54.1% (n = 158) of subcohort members with a defined time to clinic presentation. We estimated that 54.7% (95% CI: 48.9%, 61.2%) of cohort members experienced clinic presentation delay. Table 2 presents prevalence ratios estimating associations between clinic presentation delay and other characteristics. The prevalence ratios furthest from the null compared prevalence by educational attainment, with, for example, those who completed Form 6 1.19 times (95% CI: 0.88, 1.62) as likely to have experienced this delay as those who had not completed primary school. Clinic presentation delay among people who worked in the fishing industry in the preceding 12 months was 0.87 (95% CI: 0.62, 1.22) times the prevalence among those who did not recently work in this industry, and people with TB-associated HIV were 0.87 (95% CI: 0.70, 1.09) times as likely to have experienced clinic presentation delay as those without HIV. Other associations were closer to the null (1), and all 95% confidence intervals included the null.

thumbnail
Table 2. Prevalence of clinic presentation delay by characteristics, and associations between characteristics and delay.

https://doi.org/10.1371/journal.pgph.0002259.t002

Among people who experienced clinic presentation delay, 64 (40.5%) said they did not present to a health facility sooner because they first tried other treatments. An additional 33 people (20.9%) indicated that they did not think their symptoms were caused by TB, due to limited knowledge of TB symptoms or their perception of their symptoms as non-severe. Thirty-five people (22.2%) said they did not know where to go for care. All other reasons were reported by less than 5% of respondents.

Among cohort members with recorded treatment outcomes (n = 596, 76.9%), 103 had an unfavorable TB outcome: 64 died, 35 were lost to follow-up, and 4 experienced treatment failure. The full distribution of outcomes is presented in the S2 Table. The mean duration of follow-up in the risk period was 144 (standard deviation: 48.6) days. The 180-day risk of an unfavorable TB treatment outcome in the cohort overall was 13.6% (95% CI: 11.0%, 16.1%) (Table 3). The 180-day risk was 12.7% (95% CI: 5.1%, 20.3%) among those with clinic presentation delay, and 14.2% (95% CI: 8.0%, 20.4%) among those who presented to a clinic sooner (Fig 2), for a risk difference of 1.5 (95% CI: -8.3, 11.3) percentage points. The association between clinic presentation delay and 180-day risk of an unfavorable outcome was slightly stronger among those without (vs. with) TB-associated HIV, and among female (vs. male) cohort members; however, estimates were imprecise (Table 3). The HIV- and sex-stratified risks of an unfavorable TB treatment outcome over the 180-day period since TB treatment initiation among those who did and did not experience clinic presentation delay are presented in the S1 File.

thumbnail
Fig 2. Risk of an unfavorable TB treatment outcome by clinic presentation delay.

Risks were estimated from the time of TB treatment initiation up to 180 days, among people who did and did not experience clinic presentation delay. An unfavorable TB treatment outcome was defined as any outcome of: treatment failed, died, or lost to follow-up. Clinic presentation delay was defined as presenting to a health facility more than 6 weeks after the onset of TB symptoms. Data are from the East Africa TB/HIV and Mobility Study (2019).

https://doi.org/10.1371/journal.pgph.0002259.g002

thumbnail
Table 3. Risk of an unfavorable TB treatment outcome by clinic presentation delay.

The 180-day risk of an unfavorable TB treatment outcome was estimated for the cohort overall and according to experience of clinic presentation delay. Results are also shown for strata defined by HIV status and sex. Risk differences present the associations between clinic presentation delay and an unfavorable TB treatment outcome at 180 days.

https://doi.org/10.1371/journal.pgph.0002259.t003

When restricting to cohort members with measured values for clinic presentation delay, we found that throughout the 180-day follow-up period, the risk of an unfavorable TB treatment outcome was higher among those who experienced clinic presentation (S2 File). When examining mortality as the outcome of interest, we estimated that the 180-day risk of death was 5.8 (95% CI: -2.1, 13.6) percentage points higher among cohort members who experienced clinic presentation delay as compared to those who presented to a health facility within 6 weeks of the onset of TB symptoms (S3 File).

Discussion

In this study, we reported prevalence of and reasons for clinic presentation delay, explored factors associated with this delay, and estimated the association between clinic presentation delay and an unfavorable TB treatment outcome. We found that the risk of an unfavorable outcome was slightly higher among those who waited more than 6 weeks to seek care at a health facility, though the risk difference was not statistically significant. None of the assessed clinical or sociodemographic characteristics was strongly associated with clinic presentation delay.

We estimated that more than half (54.7%; 95% CI: 48.9%, 61.2%) of people initiating TB treatment at the selected health facilities experienced clinic presentation delay. This type of delay (elsewhere, often termed “patient delay”) has been described previously in some areas of East Africa, including Pwani region [23], Dar es Salaam [24], and Mwanza region [25], Tanzania. In this multi-county study, we estimated that the prevalence of clinic presentation delay was similar across cohort members in Kenya, Tanzania, and Uganda. This indicates the pervasive nature of clinic presentation delay throughout the region. Studies elsewhere in East Africa have found clinic presentation delay to be associated with sex [24, 26, 27], older age [25], lower educational attainment [25], marital status [28], unemployment [27], and health beliefs [23]; however, we did not identify any strong correlates of clinic presentation delay in this cohort.

In this study, we found a small association between clinic presentation delay and unfavorable TB treatment outcomes. This is compatible with a range of hypotheses, including the null. Other studies assessing the association between clinic presentation delay or total delay (i.e., time from TB symptoms to treatment initiation) and sub-optimal TB treatment outcomes have reported the same direction of effect as what we report here [2936].

Assessment of clinic presentation delay was subject to interviewer and recall biases, which may have caused us to underestimate the prevalence of clinic presentation delay. We also assessed clinic presentation delay as a categorical rather than continuous variable in the subcohort survey. We are therefore unable to describe time to clinic presentation in further detail or consider alternative thresholds that could be relevant to TB treatment outcomes. In addition, because clinic presentation delay was assessed only in the subcohort, inferences in the full cohort depend on the validity of the process we used to impute missing clinic presentation delay data. Our results could be biased if we omitted important variables associated with both clinic presentation delay and an unfavorable outcome. We found, however, that the direction of the association between clinic presentation delay and an unfavorable TB treatment outcome was consistent whether including or excluding cohort members with an imputed value for clinic presentation delay. Some TB treatment outcomes may have been misclassified due to incomplete records, incomplete death ascertainment, or silent transfers. Furthermore, we used a composite outcome, which may obscure associations between clinic presentation delay and individual treatment outcomes. The composite outcome is operationally useful, however, given the desirability of reducing all component outcomes (death, loss to follow-up, and treatment failure), and given that similar interventions would likely be used to prevent each outcome. The direction of association between clinic presentation delay and mortality was consistent with the direction of association between clinic presentation delay and the composite outcome. The point estimate of the 180-day risk difference was larger when considering mortality as compared to the composite outcome; however, the risk estimates for mortality may be sensitive to incomplete death ascertainment, particularly among people lost to follow-up. Finally, our findings may not generalize to people in care at health facilities dissimilar to those selected for this study, such as at facilities that are not linked to Xpert MTB/RIF testing, or where the profile of clinic presentation delay is dissimilar.

The high prevalence and generalized nature of clinic presentation delay in communities near Lake Victoria suggests a need for multi-level community-based interventions that address known barriers to care. Such interventions may involve, for example, providing patient-centered services close to where people with TB live and work. Given that members of the subcohort commonly cited non-specificity and mildness of symptoms as reasons for clinic presentation delay, time to diagnosis and treatment may be improved by increasing public awareness of the range and potential mild presentation of early TB symptoms. About one in five subcohort members with clinic presentation delay reported that they were unsure where to go for care. Informational campaigns about where to seek care for TB symptoms could help to mitigate delays among people experiencing this barrier. Many (40.5%) of those with clinic presentation delay reported seeking traditional medicine or medications that were not specifically anti-TB drugs before they visited a healthcare facility. This has been previously described in the region [9, 26]. Increasing or introducing TB screening into these non-clinical settings may be effective in reducing clinic presentation delays. Still, the often low positive predictive value of symptom-based TB screening [3740] warrants consideration. Especially in highly resource-limited settings, the costs and potential benefits of expanded TB screening should be weighed against those of other efforts to improve individual and public health [41].

This study demonstrates an efficient method to study yet-unmeasured factors in a defined population by collecting new data among a subset of a population while leveraging extant data to produce estimates generalizable to the broader population. This approach can be especially useful in healthcare settings, where routine medical records exist and results can inform interventions among an accessible population (i.e., people in care). Substantively, our results provide important insights into clinic presentation delays occurring in communities near Lake Victoria and into the relationship between clinic presentation delay and unfavorable TB treatment outcomes.

Supporting information

S1 Checklist. Inclusivity in global research checklist.

https://doi.org/10.1371/journal.pgph.0002259.s001

(DOCX)

S1 Table. Variables used in multiple imputation.

https://doi.org/10.1371/journal.pgph.0002259.s002

(DOCX)

S2 Table. TB treatment outcomes in the full study cohort.

https://doi.org/10.1371/journal.pgph.0002259.s003

(DOCX)

S1 File. Risk of an unfavorable TB treatment outcome by clinic presentation delay, stratified by HIV status and sex.

https://doi.org/10.1371/journal.pgph.0002259.s004

(DOCX)

S2 File. Risk of an unfavorable TB treatment outcome by clinic presentation delay, restricting to cohort members with measured values for clinic presentation delay.

https://doi.org/10.1371/journal.pgph.0002259.s005

(DOCX)

S3 File. Risk of death by clinic presentation delay.

https://doi.org/10.1371/journal.pgph.0002259.s006

(DOCX)

Acknowledgments

We thank members of the Lake Victoria Consortium for Health Research: UVRI-IAVI HIV Vaccine Program, Kenya Medical Research Institute, Medical Research Council/Uganda Virus Research Institute and London School of Hygiene & Tropical Medicine Uganda Research Unit, and the Mwanza Intervention Trials Unit; and the data collection teams in each country. We thank the leadership and staff of the health facilities where the data were extracted for their guidance and assistance. Our sincere thanks to all our participants in Kenya, Tanzania, and Uganda who gave their time and information in the interviews.

References

  1. 1. Escombe AR, Moore DAJ, Gilman RH, Pan W, Navincopa M, Ticona E, et al. The infectiousness of tuberculosis patients coinfected with HIV. PLoS Med. 2008;5: 1387–1396. pmid:18798687
  2. 2. Riley RL, Mills CC, Nyka W, Weinstock N, Storey PB, Sultan LU, et al. Aerial dissemination of pulmonary tuberculosis: A two-year study of contagion in a tuberculosis ward. Am J Epidemiol. 1995;142: 3–14. pmid:7785671
  3. 3. Churchyard G, Kim P, Shah NS, Rustomjee R, Gandhi N, Mathema B, et al. What We Know About Tuberculosis Transmission: An Overview. J Infect Dis. 2017;216: S629–S635. pmid:29112747
  4. 4. Dye C, Glaziou P, Floyd K, Raviglione M. Prospects for Tuberculosis Elimination. Annu Rev Public Health. 2013;34: 271–286. pmid:23244049
  5. 5. World Health Organization. Global Tuberculosis Report 2019. Geneva; 2019. Available: https://apps.who.int/iris/bitstream/handle/10665/329368/9789241565714-eng.pdf?ua=1
  6. 6. Tiemersma EW, van der Werf MJ, Borgdorff MW, Williams BG, Nagelkerke NJD. Natural history of tuberculosis: duration and fatality of untreated pulmonary tuberculosis in HIV negative patients: a systematic review. PLoS One. 2011;6: e17601. pmid:21483732
  7. 7. World Health Organization. The Top 10 Causes of Death. 2020. Available: https://www.who.int/news-room/fact-sheets/detail/the-top-10-causes-of-death
  8. 8. Harries AD, Lin Y, Kumar AMV, Satyanarayana S, Takarinda KC, Dlodlo RA, et al. What can National TB Control Programmes in low- and middle-income countries do to end tuberculosis by 2030? [version 1; referees: 2 approved]. F1000Research. Faculty of 1000 Ltd; 2018. pmid:30026917
  9. 9. Mbuthia GW, Olungah CO, Ondicho TG. Health-seeking pathway and factors leading to delays in tuberculosis diagnosis in West Pokot County, Kenya: A grounded theory study. Fair E, editor. PLoS One. 2018;13: e0207995. pmid:30485379
  10. 10. World Health Organization. Systematic screening for active tuberculosis: principles and recommendations. Geneva; 2013. Available: https://apps.who.int/iris/bitstream/handle/10665/84971/9789241548601_eng.pdf?sequence=1
  11. 11. Nash S, Tittle V, Abaasa A, Sanya RE, Asiki G, Hansen CH, et al. The validity of an area-based method to estimate the size of hard-to-reach populations using satellite images: The example of fishing populations of Lake Victoria. Emerg Themes Epidemiol. 2018;15. pmid:30123310
  12. 12. Huong NT, Vree M, Duong BD, Khanh VT, Loan VT, Co N V., et al. Delays in the diagnosis and treatment of tuberculosis patients in Vietnam: A cross-sectional study. BMC Public Health. 2007;7. pmid:17567521
  13. 13. SAS Institute Inc. SAS Software, Version 9.4. Cary, NC; 2016.
  14. 14. R Foundation for Statistical Computing. R: A Language and Environment for Statistical Computing. Vienna, Austria; 2020. Available: https://www.r-project.org/
  15. 15. OpenStreetMap Copyright and License. Available: openstreetmap.org/copyright
  16. 16. Map tiles: Watercolor. Stamen Design; Available: http://maps.stamen.com/#watercolor/8/-1.300/32.900
  17. 17. Liu Y, De A. Multiple Imputation by Fully Conditional Specification for Dealing with Missing Data in a Large Epidemiologic Study. Int J Stat Med Res. 2015;4: 287–295. pmid:27429686
  18. 18. Richards LE, Little RJA, Rubin DB. Statistical Analysis with Missing Data. 3rd ed. John Wiley & Sons; 2019.
  19. 19. Nelson W. Theory and Applications of Hazard Plotting for Censored Failure Data. Technometrics. 1972;14: 945–966.
  20. 20. Odd A. Nonparametric Inference for a Family of Counting Processes. The Annals of Statistics. 1978;6: 701–726.
  21. 21. World Health Organization. Definitions and reporting framework for tuberculosis– 2013 revision (updated December 2014 and January 2020). Geneva; 2014. Available: http://apps.who.int/iris/bitstream/10665/79199/1/9789241505345_eng.pdf
  22. 22. Agresti A. Analysis of Ordinal Categorical Data: Second Edition. 2nd ed. Wiley , editor. Analysis of Ordinal Categorical Data: Second Edition. Hoboken; 2012.
  23. 23. Ngadaya ES, Mfinanga GS, Wandwalo ER, Morkve O. Delay in tuberculosis case detection in Pwani region, Tanzania. A cross sectional study. BMC Health Serv Res. 2009;9: 196. pmid:19863823
  24. 24. Mfinanga SG, Mutayoba BK, Kahwa A, Kimaro G, Mtandu R, Ngadaya E, et al. The magnitude and factors associated with delays in management of smear positive tuberculosis in Dar es Salaam, Tanzania. BMC Health Serv Res. 2008;8: 158. pmid:18655730
  25. 25. Wandwalo ER, Mørkve O. Delay in tuberculosis case-finding and treatment in Mwanza, Tanzania. International Journal of Tuberculosis and Lung Disease. 2000;4: 133–138. pmid:10694091
  26. 26. Mhalu G, Weiss MG, Hella J, Mhimbira F, Mahongo E, Schindler C, et al. Explaining patient delay in healthcare seeking and loss to diagnostic follow-up among patients with presumptive tuberculosis in Tanzania: a mixed-methods study. BMC Health Serv Res. 2019;19: 217. pmid:30953502
  27. 27. Muttamba W, Kyobe S, Komuhangi A, Lakony J, Buregyeya E, Mabumba E, et al. Delays in diagnosis and treatment of pulmonary tuberculosis in patients seeking care at a regional referral hospital, Uganda: a cross sectional study. BMC Res Notes. 2019;12: 589. pmid:31533804
  28. 28. Ayuo P, Diero L, Owino-Ong’or W, Mwangi A. Causes Of Delay In Diagnosis Of Pulmonary Tuberculosis In Patients Attending A Referral Hospital In Western Kenya. East Afr Med J. 2008;85: 263–268. pmid:18817022
  29. 29. Woya AA, Tekile AK, Basha GW. Spatial Frailty Survival Model for Multidrug-Resistant Tuberculosis Mortality in Amhara Region, Ethiopia. Tuberc Res Treat. 2019;2019: 1–7. pmid:30693105
  30. 30. Xie Y, Han J, Yu W, Wu J, Li X, Chen H. Survival Analysis of Risk Factors for Mortality in a Cohort of Patients with Tuberculosis. Can Respir J. 2020;2020. pmid:32963642
  31. 31. Asres A, Jerene D, Deressa W. Delays to treatment initiation is associated with tuberculosis treatment outcomes among patients on directly observed treatment short course in Southwest Ethiopia: A follow-up study. BMC Pulm Med. 2018;18: 64. pmid:29716569
  32. 32. Virenfeldt J, Rudolf F, Camara C, Furtado A, Gomes V, Aaby P, et al. Treatment delay affects clinical severity of tuberculosis: a longitudinal cohort study. BMJ Open. 2014;4: e004818. pmid:24916087
  33. 33. Kassim SA, Cote A, Kassim SM, Abbas M, Baig MMFA, Ahmed AM, et al. Factors influencing treatment outcomes of tuberculosis patients attending health facilities in Galkayo Puntland, Somalia. J Public Health (Bangkok). 2020. pmid:32880632
  34. 34. Lee CH, Wang JY, Lin HC, Lin PY, Chang JH, Suk CW, et al. Treatment delay and fatal outcomes of pulmonary tuberculosis in advanced age: A retrospective nationwide cohort study. BMC Infect Dis. 2017;17. pmid:28646854
  35. 35. Gebreegziabher SB, Bjune GA, Yimer SA. Total delay is associated with unfavorable treatment outcome among pulmonary tuberculosis patients in West Gojjam Zone, Northwest Ethiopia: A prospective cohort study. PLoS One. 2016;11. pmid:27442529
  36. 36. Lisboa M, Fronteira I, Colove E, Nhamonga M, Martins M do RO. Time delay and associated mortality from negative smear to positive Xpert MTB/RIF test among TB/HIV patients: a retrospective study. BMC Infect Dis. 2019;19. pmid:30616533
  37. 37. Menberu MA. Performance of the WHO 2011 TB symptom screening algorithm for pulmonary TB diagnosis among HIV-infected patients in Gondar University Referral Hospital, Ethiopia. Int J Microbiol. 2016;2016. pmid:28058048
  38. 38. Vonasek B, Kay A, Devezin T, Bacha JM, Kazembe P, Dhillon D, et al. Tuberculosis symptom screening for children and adolescents living with HIV in six high HIV/TB burden countries in Africa. AIDS. 2021;35: 73–79. pmid:33048868
  39. 39. Brennan A, Maskew M, Larson BA, Tsikhutsu I, Bii M, Vezi L, et al. Prevalence of TB symptoms, diagnosis and treatment among people living with HIV (PLHIV) not on ART presenting at outpatient clinics in South Africa and Kenya: Baseline results from a clinical trial. BMJ Open. 2020;10. pmid:32895266
  40. 40. Kagujje M, Mwanza W, Somwe P, Chilukutu L, Creswell J, Muyoyeta M. Sensitivity and specificity of CRP and symptom screening as tuberculosis screening tools among HIV-positive and negative outpatients at a primary healthcare facility in Lusaka, Zambia: a prospective cross-sectional study. BMJ Open. 2023;13: e061907. pmid:37072353
  41. 41. World Health Organization. Systematic screening for active tuberculosis: an operational guide. Geneva; 2015. Available: https://www.who.int/publications/i/item/9789241549172