Advertisement

Bacteriologically confirmed pulmonary tuberculosis patients: Loss to follow-up, death and delay before treatment initiation in Bulawayo, Zimbabwe from 2012–2016

Open AccessPublished:July 17, 2018DOI:https://doi.org/10.1016/j.ijid.2018.07.012

      Highlights

      • One in five people with drug-susceptible, bacteriologically confirmed tuberculosis was not initiated on treatment which constituted pre-treatment loss to follow-up (LTFU).
      • Pre-treatment deaths accounted for approximately half of all pre-treatment LTFU cases in a city with high HIV prevalence (18.7%).
      • There were long delays between sputum receipt to testing, and diagnosis to treatment initiation, though the delay was on a descending trajectory between 2012 and 2016.
      • Old age, male or missing gender, and being HIV-positive or having an unknown HIV status increased the risk of pre-treatment LTFU significantly.

      Abstract

      Objective

      To quantify and assess trends and risk factors for loss to follow-up (LTFU) and delays before treatment initiation among bacteriologically confirmed pulmonary tuberculosis (TB) patients (laboratory-diagnosed) in Bulawayo, 2012–16.

      Design

      Cohort study using secondary programme data. Presumptive TB patients’ sputum samples were sent to the laboratory from the 19 primary health care clinics. Laboratory-diagnosed patients (microscopy or Xpert MTB/RIF) were tracked for treatment registration at the clinics.

      Results

      Of 2443 laboratory-diagnosed patients, the mean (standard deviation, SD) delay from sputum receipt at the laboratory to testing was 2.7(1.6) days and from testing to result dispatch was 8.8(5.8) days. A total of 508(20.8%) were LTFU which included 252(10.3%) deaths. While the number of laboratory-diagnosed patients reduced over years, there was a significant increase in pre-treatment LTFU and death. Independent predictors of pre-treatment LTFU were age above 65 years, male gender and HIV positive/unknown. In addition, delay (≥3 days) between sputum receipt and testing was significantly associated with pre-treatment death. Among registered patients (n = 1935), the mean (SD) delay to initiate treatment was 29.1 (21.6) days which significantly declined over the years. Patients registered as new TB had significantly long treatment delay.

      Conclusions

      Interventions to mitigate the risk factors for high loss to follow-up, deaths and delays before TB treatment are urgently required.

      Keywords

      Introduction

      Tuberculosis (TB) is the single most important cause of death from an infectious disease. In 2016, approximately 1.7 million deaths were reported globally (
      • World Health Organization
      Global Tuberculosis Report 2017.
      ). An estimated 4.1 million people with TB were either not diagnosed or if diagnosed, not registered (
      • World Health Organization
      Global Tuberculosis Report 2017.
      ). Rapid identification and diagnosis of people with TB and their prompt initiation of appropriate treatment are essential to ending TB (
      • World Health Organization
      The End TB Strategy. Global strategy and targets for tuberculosis prevention, care and control after 2015.
      ).
      TB patients may be ‘lost’ after diagnosis and before treatment initiation. This is called initial or pre-treatment loss-to-follow-up (LTFU) (
      • Squire S.B.
      • Belaye A.K.
      • Kashoti A.
      • Salaniponi F.M.L.
      • Mundy C.J.F.
      • Theobald S.
      • et al.
      “Lost” smear-positive pulmonary tuberculosis cases: where are they and why did we lose them?.
      ,
      • Dowdy D.W.
      • Chaisson R.E.
      The persistence of tuberculosis in the age of DOTS: reassessing the effect of case detection.
      ,
      • Harries A.D.
      • Rusen I.D.
      • Chiang C.-Y.
      • Hinderaker S.G.
      • Enarson D.A.
      Registering initial defaulters and reporting on their treatment outcomes.
      ). Pre-treatment LTFU indicates that in spite of having attended health services and getting diagnosed, all patients do not necessarily transition to treatment, some ‘drop off’. In a systematic review and meta-analysis of 23 studies from low to middle-income countries, pre-treatment LTFU ranged from 4 to 38% and was more common in studies from Africa (
      • MacPherson P.
      • Houben R.M.
      • Glynn J.R.
      • Corbett E.L.
      • Kranzer K.
      Pre-treatment loss to follow-up in tuberculosis patients in low- and lower-middle-income countries and high-burden countries: a systematic review and meta-analysis.
      ). In this study, male sex, older age, urban residence and being diagnosed with smear-negative TB were identified as risk factors for pre-treatment LTFU.
      Zimbabwe in southern Africa is the epicentre of HIV-associated TB. It is one of the 30 high TB burden countries (
      • World Health Organization
      Global Tuberculosis Report 2017.
      ). Of the estimated 34 000 new TB patients in Zimbabwe in 2016, only 27 353 (80%) were notified. This resulted in a gap of approximately 6,647 (20%) people with TB who were not diagnosed and/or registered (
      • World Health Organization
      Global Tuberculosis Report 2017.
      ).
      A study from a rural district in the country in 2006 revealed a pre-treatment LTFU of 27% (
      • Chadambuka A.
      • Mabaera B.
      • Tshimanga M.
      • Shambira G.
      • Gombe N.T.
      • Chimusoro A.
      Low tuberculosis case detection in Gokwe North and South, Zimbabwe in 2006.
      ). There are no other assessments of pre-treatment LTFU in Zimbabwe, including urban areas. Understanding the extent and factors associated with LTFU and delays before treatment will help the programme to design interventions to ensure all TB patients are started on treatment. In spite of the fact that persons with pre-treatment LTFU experience high mortality (
      • Squire S.B.
      • Belaye A.K.
      • Kashoti A.
      • Salaniponi F.M.L.
      • Mundy C.J.F.
      • Theobald S.
      • et al.
      “Lost” smear-positive pulmonary tuberculosis cases: where are they and why did we lose them?.
      ), there is limited information in Zimbabwe on pre-treatment deaths.
      The objectives of this study from Bulawayo, Zimbabwe were to (i) quantify pre-treatment LFTU/deaths among patients with bacteriologically confirmed pulmonary TB, (ii) describe their yearly trends from 2012 to 2016, (iii) determine the factors associated with pre-treatment LTFU/death and (iv) among those initiated on treatment in the above cohort, determine the delay before the treatment initiation and the factors associated.

      Methods

      Study design

      This was a cohort study using secondary programme data.

      Setting

      General setting

      In 2016, Zimbabwe had an estimated population of 16 million and ranked number 154 out of 189 countries on the human development index (
      • UNAIDS
      UNAIDS. Country: Zimbabwe.
      ,
      • UNDP and Government of Zimbabwe
      Zimbabwe Human Development Report 2017. Climate Change and Human Development: Towards Building a Climate Resilient Nation.
      ). In the same year, Bulawayo recorded the highest HIV prevalence in the country, at 18.7% compared with the national average of 14% (
      • Ministry of Health and Child Care (MoHCC), Zimbabwe, ZIMSTAT
      Zimbabwe Populations based HIV impact assessment, ZIMPHIA 2016.
      ).

      Zimbabwe National TB Programme

      In 2016, the national TB notification rate was reported to be 272 per 100 000 population. TB care and prevention activities are guided by the National TB Strategic Plan, 2017–2020 (
      • Zimbabwe Ministry of Health and Child Care
      Ministry of Health and Child Welfare, National Tuberculosis Program – Strategic Plan (2017-2020).
      ). TB services are implemented by the national unit, provincial and district coordinators and focal nurses. TB and HIV diagnosis, treatment and care, including ART initiation and follow-up services, are decentralised and integrated into general health services from primary to tertiary health services.
      Until 2016, sputum smear microscopy and chest radiography were the major investigations offered for diagnosis of pulmonary TB though expansion of rapid molecular techniques started in 2012. Sputum culture and line probe assay were available at two national reference laboratories.
      From 2015–16, rapid molecular test, Xpert MTB/RIF assay (Cepheid Inc, Sunnyvale, CA, USA) was provided to the high-risk populations, such as people living with HIV, people with risk factors for drug-resistant TB, health workers and miners. In 2017, Xpert MTB/RIF assay was adopted as the diagnostic standard for all presumptive TB (Zimbabwe Ministry of health/End TB Zimbabwe, 2018).
      All diagnosed patients are registered with a unique registration number, given standardised treatment and monitored for treatment outcomes according to the national recommendations (
      • World Health Organization
      Definitions and reporting framework for tuberculosis — December 2013 revision (updated December 2014).
      ,
      • Zimbabwe Ministry of Health and Child Care
      Zimbabwe National Tuberculosis and Leprosy Management Guidelines.
      ). TB investigations and all treatment are free of charge to patients in Zimbabwe.
      The national TB programme has a paper-based recording and reporting system that includes a presumptive TB, laboratory, directly observed treatment short-course (DOT) registers and district TB register. All presumptive TB patients are documented in facility-based presumptive TB registers with a record of the subsequent collection of sputum specimens and test results. This register is aimed at preventing attrition at all stages of the diagnostic and treatment cascade.

      Study site

      Bulawayo is Zimbabwe’s second largest city and had a population of 0.7 million in 2017 (
      • World Population Review
      Zimbabwe Population 2018.
      ). It has 19 primary health care clinics, one laboratory and one infectious disease hospital under the municipality.
      The national TB and HIV care/ART recording and reporting tools are used. Presumptive TB patients attend clinics closest to their residence and sputum specimens are collected and sent through a motorised specimen transport to the laboratory. At the laboratory, the following are documented: specimen receipt, test result and date of dispatch of the result back to respective clinics through the motorised system (once to twice weekly). Positive sputum examination results are communicated to respective clinics by phone, followed by printed copies of the results. At clinics, a TB focal nurse, in turn, telephones the patient inviting them to return for information and education about TB and start treatment.

      Patient population

      All patients in Bulawayo city clinics with bacteriologically confirmed (by microscopy or Xpert MTB/RIF assay or both) drug-susceptible pulmonary TB that were documented in the laboratory register from 1 January 2012 to 31 December 2016 were included in this study.

      Data variables, sources of data and data collection

      Data were collected between December 2017 and February 2018 using a structured data collection form. A line list of diagnosed patients was prepared from the laboratory register after removal of duplicate entries.
      The following variables were collected: laboratory serial number, name of clinic referring the sputum sample, name of patient, age, sex, HIV status, date of sputum receipt, date of testing, date of result dispatch, type of test (sputum microscopy/Xpert MTB/RIF assay), registered for TB treatment (yes/no) and date of treatment initiation (if registered). The sources of data included the laboratory register, master TB register of the health services department and clinic DOT registers. If the patient was not registered in district TB register, we confirmed this in the DOT register of the referring clinic. We collected information on death among those with pre-treatment LTFU from the presumptive TB register at clinics. Therefore, pre-treatment LTFU included deaths.
      Sputum grading was considered ‘severe’ if Xpert MTB/RIF assay grading was ‘high’ (microscopy grading of ‘3+’ if Xpert MTB/RIF was not used). Data for treatment registration were reviewed in the master TB register and clinic DOT TB register up to three months after the TB diagnosis. Tracking of the patients was done using laboratory number or name/age/sex if former was not recorded.

      Analysis and statistics

      Data were single-entered and analysed using EpiData (version 3.1 for entry and version 2.2.2.183 for analysis, EpiData Association, Odense, Denmark) for descriptive and unadjusted analysis (Supplementary Annex S1). The multivariable-adjusted analysis was done using STATA (version 12.1 STATA Corp., College Station, TX, USA).
      The trends of pre-treatment LTFU/death (among diagnosed) between 2012 and 2016 were summarised using bar/line diagram. Key analytic outputs were number (proportion) of pre-treatment LTFU (including deaths) and death on its own within three months of diagnosis. We described the mean (standard deviation-SD) treatment initiation delay (in days) between test-date and treatment initiation among those registered on treatment. The association of risk factors with pre-treatment LTFU/deaths were summarised using relative risks/adjusted relative risks (0.95 confidence interval (CI)). Adjusted relative risks were calculated using modified Poisson regression. The association of risk factors with delay in treatment initiation from diagnosis were summarised as Beta coefficients (0.95 CI) calculated using linear regression. The Beta coefficient indicated the adjusted mean difference of delay in treatment initiation from diagnosis (in days) between the sub-category of interest and the reference sub-category.
      Type of test for diagnosis was excluded from all the risk factor analyses because of a strong relationship with HIV status and diagnosis year. Delay from testing to result dispatch was excluded from the risk factor analysis of treatment initiation delay because the latter included the former.

      Results

      Baseline characteristics

      Of 2443 patients, females constituted 50.6% (n = 1 237) and 45.1% (n = 1,103) patients belonged to the age-group 25–44 years. TB diagnosis was done using sputum smear microscopy in 64.7% (n = 1,580), Xpert MTB/RIF assay in 22.4% (n = 548) and both in 12.9% (n = 315). Sputum grading was severe in 34.8% (n = 849) patients. HIV positivity was 74.2% (n = 1812) (Table 1). HIV positivity decreased from 2012 to 2016 (Supplementary Figure S1). Among HIV-positive patients, the diagnosis by Xpert MTB/RIF assay increased in 2015 and 2016 (Supplementary Figure S2).
      Table 1Clinical and demographic profile of the patients with bacteriologically confirmed pulmonary tuberculosis in Bulawayo, Zimbabwe, 2012–16. (N = 2443).
      VariableNumber(%)
      Column percentage.
      Total2443(100)
      Age in years
       ○ <1527(1.1)
       ○ 15–24330(13.5)
       ○ 25–34555(22.7)
       ○ 35–44548(22.4)
       ○ 45–54321(13.1)
       ○ 55–64247(10.1)
       ○  >/= 65414(16.9)
       ○ Unknown1(<0.1)
      Gender
       ○ Male1192(48.8)
       ○ Female1237(50.6)
       ○ Not documented14(0.6)
      HIV status
       ○ Positive1812(74.2)
       ○ Negative537(22.0)
       ○ Not documented94(3.8)
      Delay from sputum receipt to test
       ○ ≤2 days1408(57.6)
       ○ ≥3 days1035(42.4)
      Delay from testing to result dispatch
       ○ ≤8 days1112(45.5)
       ○ ≥9 days1331(54.5)
      Test type
       ○ Microscopy1580(64.7)
       ○ Xpert MTB/RIF
      Xpert MTB/RIF — genotypic diagnostic test for tuberculosis.
      548(22.4)
       ○ Both315(12.9)
      Severe sputum grade
      Xpert MTB/RIF grading as high; if Xpert MTB/RIF not used, sputum microscopy grade of 3+.
       ○ Yes849(34.8)
       ○ No1594(65.2)
      Year of diagnosis
       ○ 2012531(21.7)
       ○ 2013502(20.6)
       ○ 2014490(20.1)
       ○ 2015467(19.1)
       ○ 2016453(18.5)
      a Column percentage.
      b Xpert MTB/RIF grading as high; if Xpert MTB/RIF not used, sputum microscopy grade of 3+.
      c Xpert MTB/RIF — genotypic diagnostic test for tuberculosis.

      Laboratory delays

      The mean (SD) delay from sputum receipt in the laboratory to testing was 2.7(1.6) days and from testing to result dispatch was 8.8(5.8) days. There was a decreasing trend of laboratory delays over the years. In 2016, on an average sputum testing happened within a day of receipt and results were dispatched within three days (Figure 1).
      Figure 1
      Figure 1Trends of mean delay (in days) from sputum receipt to testing and from testing to result dispatch at laboratory among bacteriologically confirmed pulmonary tuberculosis patients in Bulawayo, Zimbabwe, 2012–16.

      Pre-treatment loss to follow up and deaths from 2012–16

      A steady decrease (14.7%) in TB diagnoses was observed: it was 531 in 2012 and 453 in 2016 (Figure 2).
      Figure 2
      Figure 2Number of bacteriologically confirmed pulmonary tuberculosis diagnosed, pre-treatment loss to follow up and pre-treatment deaths in Bulawayo, Zimbabwe, 2012–16*.
      *Pre-treatment loss to follow up includes pre-treatment death.
      Pre-treatment LTFU was 20.8% (508/2443) which included 252 deaths, that is, 10.3% (252/2443) (Table 2, Table 3). The yearly proportion of pre-treatment LTFU and death from 2012–16 is depicted in Figure 3. Pre-treatment LTFU ranged from 12.1% in 2012 to 24.9% in 2016 with an overall increasing trend. Pre-treatment death increased from 6.4% in 2012 to 15.5% in 2016. Pre-treatment deaths constituted 49.6% (252/508) of all the pre-treatment LTFU: it was 56.1% (189/337) and 43.5% (37/85) among HIV-positive and HIV-negative TB patients, respectively.
      Table 2Factors associated with pre-treatment loss to follow up among bacteriologically confirmed pulmonary tuberculosis in Bulawayo, Zimbabwe, 2012–16. (N = 2442).
      One patient was excluded from the model due to missing data; type of test for diagnosis was excluded from the model because of a strong relationship with HIV status and diagnosis year.
      VariableTotalPre-treatment loss to follow up
      Includes pre-treatment deaths.
      RR (95 CI)aRR (95 CI)
      Modified Poisson regression for aRR.
      Nn(%)
      Row percentages.
      Total2442508(20.8)
      Age in years
       ○ <15278(29.6)2.32 (1.25,4.32)1.51 (0.79, 2.88)
       ○ 15–2433066(20.0)1.57 (1.15,2.13)1.56 (1.17,2.08)
      p<0.05.
       ○ 25–34555104(18.7)1.47 (1.11,1.94)1.51 (1.16,1.97)
      p<0.05.
       ○ 35–4454870(12.8)RefRef
       ○ 45–5432152(16.2)1.27 (0.91,1.77)1.38 (1.00,1.97)
      p<0.05.
       ○ 55–6424747(19.0)1.49 (1.06,2.09)1.46 (1.06, 2.01)
      p<0.05.
       ○ >/= 65414160(38.6)3.02 (2.36,3.89)2.71 (2.12,3.46)
      p<0.05.
      Gender
       ○ Male1191285(23.9)1.36 (1.16,1.60)1.21 (1.04,1.41)
      p<0.05.
       ○ Female1237217(17.5)RefRef
       ○ Not documented146(42.9)2.44 (1.32,4.53)1.78 (1.03, 3.09)
      p<0.05.
      HIV status
       ○ Positive1813338(18.7)1.18 (0.95,1.46)1.27 (1.02,1.56)
      p<0.05.
       ○ Negative53585(15.8)RefRef
       ○ Unknown9485(90.4)5.71 (4.65,7.02)4.78 (3.81,6.01)
      p<0.05.
      Delay from sputum receipt to test
       ○ ≤2 days1407314(22.3)RefRef
       ○ ≥3 days1035194(18.7)0.85 (0.72,0,99)1.10 (0.93,1.31)
      Delay from testing to result dispatch
       ○ ≤8 days1111291(26.2)RefRef
       ○ ≥9 days1331217(16.3)0.62 (0.53, 0.73)0.94 (0.74, 1.20)
      Severe sputum grade
      Xpert MTB/RIF grading as high; if Xpert MTB/RIF not used, sputum microscopy grade of 3+.
       ○ Yes848169(19.9)0.94 (0.79, 1.10)0.86 (0.74, 1.01)
       ○ No1594339(21.3)RefRef
      Year of diagnosis
       ○ 201253164(12.1)RefRef
       ○ 2013502102(20.3)1.69 (1.26,2.25)1.75 (1.33, 2.31)
      p<0.05.
       ○ 201449080(16.3)1.35 (0.99,1.84)1.19 (0.88, 1.61)
       ○ 2015466149(31.9)2.65 (2.03,3.45)2.03 (1.45, 2.85)
      p<0.05.
       ○ 2016453113(24.9)2.07 (1.56,2.74)2.16 (1.53, 3.04)
      p<0.05.
      RR — relative risk; aRR — adjusted relative risk; CI — confidence interval.
      a One patient was excluded from the model due to missing data; type of test for diagnosis was excluded from the model because of a strong relationship with HIV status and diagnosis year.
      b Includes pre-treatment deaths.
      c Row percentages.
      d Modified Poisson regression for aRR.
      e Xpert MTB/RIF grading as high; if Xpert MTB/RIF not used, sputum microscopy grade of 3+.
      ^ p < 0.05.
      Table 3Factors associated with pre-treatment death among patients with bacteriologically confirmed pulmonary tuberculosis in Bulawayo, Zimbabwe, 2012–16. (N = 2442).
      One patient was excluded from the model due to missing data; type of test for diagnosis was excluded from the model because of a strong relationship with HIV status and diagnosis year.
      VariableTotalPre-treatment deathRR (95 CI)aRR (95 CI)
      Modified Poisson regression for aRR.
      Nn(%)
      Row percentages.
      Total2442252(10.3)
      Age in years
       ○ <15273(11.1)3.20 (1.01, 10.17)2.01 (0.62, 6.50)
       ○ 15–2433026(7.9)2.27 (1.28, 4.04)2.30 (1.30, 4.06)
      p<0.05.
       ○ 25–3455533(5.9)1.71 (0.99, 2.98)1.83 (1.06, 3.17)
      p<0.05.
       ○ 35–4454819(3.5)RefRef
       ○ 45–5432126(8.1)2.34 (1.31, 4.15)2.49 (1.40, 4.42)
      p<0.05.
       ○ 55–6424722(8.9)2.57 (1.42, 4.66)2.60 (1.43, 4.71)
      p<0.05.
       ○ >/= 65414122(29.5)8.50 (5.33, 13.55)8.66 (5.37, 13.98)
      p<0.05.
      Gender
       ○ Male1191144(12.1)1.45 (1.14,1.84)1.29 (1.02,1.62)
      p<0.05.
       ○ Female1237103(8.3)RefRef
       ○ Not documented145(35.7)4.29 (2.07, 8.87)3.29 (1.85, 5.86)
      p<0.05.
      HIV status
       ○ Positive1813189(10.4)1.51 (1.08, 2.12)1.66 (1.20, 2.30)
      p<0.05.
       ○ Negative53537(6.9)RefRef
       ○ Unknown9426(27.7)4.01 (2.56, 6.30)3.03 (1.97, 4.64)
      p<0.05.
      Delay from sputum receipt to test
       ○ ≤2 days1407153(10.9)RefRef
       ○ ≥3 days103599(9.6)0.88 (0.69,1,12)1.44 (1.09,1.90)
      p<0.05.
      Delay from testing to result dispatch
       ○ ≤8 days1111153(13.8)RefRef
       ○ ≥9 days133199(7.4)0.54 (0.43, 0.69)0.94 (0.65, 1.35)
      Severe sputum grade
      Xpert MTB/RIF grading as high; if Xpert MTB/RIF not used, sputum microscopy grade of 3+.
       ○ Yes84880(9.4)0.87 (0.68, 1.12)0.80 (0.63, 1.02)
       ○ No1594172(10.8)RefRef
      Year of diagnosis
       ○ 201253134(6.4)RefRef
       ○ 201350235(7.0)1.09 (0.69, 1.72)1.29 (0.85, 1.96)
       ○ 201449041(8.4)1.31 (0.84,2.02)1.24 (0.78, 1.95)
       ○ 201546672(15.4)2.41 (1.63, 3.55)2.03 (1.23, 3.35)
      p<0.05.
       ○ 201645370(15.5)2.41 (1.63, 3.56)3.25 (1.99, 5.32)
      p<0.05.
      RR — relative risk; aRR — adjusted relative risk; CI — confidence interval.
      a One patient was excluded from the model due to missing data; type of test for diagnosis was excluded from the model because of a strong relationship with HIV status and diagnosis year.
      b Row percentages.
      c Modified Poisson regression for aRR.
      d Xpert MTB/RIF grading as high; if Xpert MTB/RIF not used, sputum microscopy grade of 3+.
      ^ p < 0.05.
      Figure 3
      Figure 3Trends of pre-treatment loss to follow up and pre-treatment deaths among patients with bacteriologically confirmed drug-sensitive pulmonary tuberculosis in Bulawayo, Zimbabwe, 2012–16*.
      *Pre-treatment loss to follow up includes pre-treatment death.

      Pre-treatment loss to follow up and death: risk factors

      The risk factors for pre-treatment LTFU and death are summarised in Table 2, Table 3. The risk factors for both pre-treatment LTFU and death were: age groups 15-34 or ≥45 years; male or missing gender; HIV-positive or unknown; and diagnosis after 2014. In addition, diagnosis in 2013 was also a risk factor for pre-treatment LTFU when compared with 2012, whereas delay in testing in the laboratory by ≥3 days was a risk factor for pre-treatment death.

      Delay in treatment initiation from diagnosis

      Among patients registered for TB treatment (n = 1 935), the mean (SD) delay was 29.1 (21.6) days. Treatment was initiated within seven days in 2.1% (41/1,935) and fourteen days in 13% (252/1935) of the patients (Table 4). The delay stratified across sub-groups is depicted in Table 5. The risk factors for the delay have been summarised in Table 6. When compared to 2012, there was a significant reduction in delay (in days) in 2013 (nine days), 2014 (10 days), 2015 (14 days) and 2016 (19 days). Compared with previously treated patients, new patients had a significant higher delay by 14 days.
      Table 4Distribution of delay from diagnosis to treatment initiation among patients with bacteriologically confirmed pulmonary tuberculosis registered for treatment in Bulawayo, Zimbabwe, 2012–16. (N = 1935).
      Mean (SD) delay was 29.1 (21.6) days.
      Delay in daysn(%)Cumulative %
      <741(2.1)2.1
      7–13211(10.9)13.0
      14–20394(20.4)33.4
      21–27409(21.1)54.5
      ≥28878(45.4)99.9
      Missing2(0.1)100.0
      a Mean (SD) delay was 29.1 (21.6) days.
      Table 5Delay in treatment initiation from diagnosis (in days), stratified by subgroups, among patients with bacteriologically confirmed pulmonary tuberculosis registered for treatment in Bulawayo, Zimbabwe, 2012–16. (N = 1935).
      Delay from testing to result dispatch was not assessed as a baseline factor because the delay in treatment initiation included the above delay.
      VariableN(%)
      Column percentage.
      Mean delay(SD)
      Total1935(100)29.1(21.6)
      Age in years
       ○ <1519(1.0)20.6(11.7)
       ○ 15–24264(13.6)27.5(13.8)
       ○ 25–34451(23.3)32.0(32.7)
       ○ 35–44478(24.7)28.9(21.7)
       ○ 45–54269(13.9)29.1(14.5)
       ○ 55–64200(10.3)28.3(13.2)
       ○ >/= 65254(13.1)27.0(13.7)
      Gender
       ○ Male1020(52.6)30.0(21.8)
       ○ Female907(46.8)28.1(21.5)
       ○ Not documented8(0.4)24.5(10.5)
      HIV status
       ○ Positive1474(76.2)32.7(32.4)
       ○ Negative452(23.3)28.0(16.9)
       ○ Unknown9(0.5)22.6(5.8)
      Delay from sputum receipt to test
       ○ ≤2 days1094(56.5)27.1(17.4)
       ○ ≥3 days841(43.5)31.6(25.9)
      Test type
       ○ Microscopy1319(68.2)32.3(22.2)
       ○ Xpert MTB/RIF
      Delay from testing to result dispatch was not assessed as a baseline factor because the delay in treatment initiation included the above delay.
      383(19.8)22.0(11.7)
       ○ Both233(12.0)22.2(26.0)
      Severe sputum grade
      Xpert MTB/RIF grading as high; if Xpert MTB/RIF not used, sputum microscopy grade of 3+.
       ○ Yes680(35.1)28.3(23.9)
       ○ No1255(64.9)29.5(20.3)
      TB type
       ○ New1248(64.4)34.3(24.7)
       ○ Previously treated679(35.1)19.57(8.1)
       ○ Missing8(<1)20.75(11.7)
      Year of diagnosis
       ○ 2012467(24.1)38.93(22.8)
       ○ 2013400(20.7)31.42(12.8)
       ○ 2014410(21.2)28.08(20.9)
       ○ 2015318(16.4)22.44(30.2)
       ○ 2016340(17.6)20.17(10.3)
      SD — standard deviation.
      a Delay from testing to result dispatch was not assessed as a baseline factor because the delay in treatment initiation included the above delay.
      b Column percentage.
      c Xpert MTB/RIF grading as high; if Xpert MTB/RIF not used, sputum microscopy grade of 3+.
      Table 6Factors associated with delay in treatment initiation from diagnosis (in days) among patients with bacteriologically confirmed pulmonary tuberculosis registered for treatment in Bulawayo, Zimbabwe, 2012–16. (N = 1933).
      Two patients with missing data were excluded from the model; type of test for diagnosis was excluded from the model because of a strong relationship with HIV status and diagnosis year; delay from testing to result dispatch was excluded because delay in treatment initiation included the above delay.
      VariablesBeta coefficient
      Linear regression, the Beta coefficient indicated the adjusted mean difference of delay in treatment initiation from diagnosis between the sub-category of interest and the reference sub-category.
      (95% CI)p-Value
      Age in years
       ○ <15−2.8(−11.8, 6.3)0.551
       ○ 15–24−2.9(−5.9, 0.0)0.051
       ○ 25–340.6(−1.9, 3.1)0.644
       ○ 35–44RefRef
       ○ 45–54−0.8(−3.7, 2.1)0.596
       ○ 55–64−1.1(−4.4, 2.1)0.488
       ○ >/= 65−2.2(−5.2, 0.8)0.146
      Gender
       ○ Male0.2(−1.6, 1.9)0.845
       ○ FemaleRefRef
       ○ Not documented2.7(−10.9, 16.4)0.695
      HIV status
       ○ Positive−1.9(−4.0, 0.3)0.092
       ○ NegativeRefRef
       ○ Unknown0.7(−12.3, 13.7)0.915
      Delay from sputum receipt to test
       ○ ≤2 days−0.1(−2.1,1.8)0.903
       ○ ≥3 daysRefRef
      Severe sputum grade
      Xpert MTB/RIF grading as high; if Xpert MTB/RIF not used, sputum microscopy grade of 3+.
       ○ Yes0.7(−1.2, 2.6)0.477
       ○ NoRefRef
      Year of diagnosis
       ○ 2012RefRef
       ○ 2013−8.8(−11.5, −6.1)<0.001
      p<0.05.
       ○ 2014−9.6(−12.3, −6.9)<0.001
      p<0.05.
       ○ 2015−14.2(−17.3, −11.2)<0.001
      p<0.05.
       ○ 2016−18.6(−21.7, −15.6)<0.001
      p<0.05.
      TB type
       ○ New13.5(11.5,15.4)<0.001
      p<0.05.
       ○ Previously treatedRefRef
       ○ Not documented5.8(−7.9,19.5)0.408
      Constant31.8(27.9, 35.7)<0.001
      p<0.05.
      CI — confidence interval.
      a Two patients with missing data were excluded from the model; type of test for diagnosis was excluded from the model because of a strong relationship with HIV status and diagnosis year; delay from testing to result dispatch was excluded because delay in treatment initiation included the above delay.
      b Linear regression, the Beta coefficient indicated the adjusted mean difference of delay in treatment initiation from diagnosis between the sub-category of interest and the reference sub-category.
      c Xpert MTB/RIF grading as high; if Xpert MTB/RIF not used, sputum microscopy grade of 3+.
      ^ p < 0.05.

      Discussion

      This study revealed high loss to follow-up, deaths and delay before TB treatment initiation. The proportion of diagnosed patients that was either LTFU or died before treatment increased significantly between 2012 and 2016. Risk factors for pre-treatment LTFU/death and delay were identified.

      Strengths

      Availability of presumptive TB registers with complete data that included information on patients’ deaths facilitated capturing of pre-treatment deaths. The five year study duration and the large sample size enabled us to study the trend and factors that influenced pre-treatment LTFU, death and delays. The study was done using routinely collected programmatic data which was representative of the ground reality.

      Limitations

      We do not know what happened to patients who underwent pre-treatment LTFU and were not recorded as dead. Variables with a potential influence on pre-treatment LTFU, such as ART status and duration, are not documented in the presumptive TB register and were not collected. History of residing in neighbouring South Africa, identified as a factor for late presentation in other studies, and socio-economic status of the patients was not available for analysis (
      • Chadambuka A.
      • Mabaera B.
      • Tshimanga M.
      • Shambira G.
      • Gombe N.T.
      • Chimusoro A.
      Low tuberculosis case detection in Gokwe North and South, Zimbabwe in 2006.
      ,
      • Takarinda K.C.
      • Harries A.D.
      • Nyathi B.
      • Ngwenya M.
      • Mutasa-Apollo T.
      • Sandy C.
      Tuberculosis treatment delays and associated factors within the Zimbabwe national tuberculosis programme.
      ). Further, no analysis of the relationship between pre-treatment LTFU and staffing at clinics and the laboratory, availability of laboratory consumables and the distance between clinics and patients’ residences was done.

      Interpretation of key findings

      This study provided important insights into the performance of the TB programme in Bulawayo.
      First, one in five people with bacteriologically confirmed TB was LTFU before treatment initiation. The extent of pre-treatment LTFU is of serious concern, especially because the proportion increased over the study years. It was lower than reported (27%) previously in a rural district in Zimbabwe (2006) (
      • Chadambuka A.
      • Mabaera B.
      • Tshimanga M.
      • Shambira G.
      • Gombe N.T.
      • Chimusoro A.
      Low tuberculosis case detection in Gokwe North and South, Zimbabwe in 2006.
      ). High pre-treatment LTFU has also been reported globally: 22–25% in South Africa (2010–12), 22% in India (2010–12 and 2015), 17% in Cameroon, 38% in Ghana (2009) (
      • Afutu F.K.
      • Zachariah R.
      • Hinderaker S.G.
      • Ntoah-Boadi H.
      • Obeng E.A.
      • Bonsu F.A.
      • et al.
      High initial default in patients with smear-positive pulmonary tuberculosis at a regional hospital in Accra, Ghana.
      ,
      • Claassens M.M.
      • du Toit E.
      • Dunbar R.
      • Lombard C.
      • Enarson D.A.
      • Beyers N.
      • et al.
      Tuberculosis patients in primary care do not start treatment. What role do health system delays play?.
      ,
      • Claassens M.M.
      • Dunbar R.
      • Yang B.
      • Lombard C.J.
      Scanty smears associated with initial loss to follow-up in South African tuberculosis patients.
      ,
      • Mehra D.
      • Kaushik R.M.
      • Kaushik R.
      • Rawat J.
      • Kakkar R.
      Initial default among sputum-positive pulmonary TB patients at a referral hospital in Uttarakhand, India.
      ,
      • Mwansa-Kambafwile J.
      • Maitshotlo B.
      • Black A.
      Microbiologically confirmed tuberculosis: factors associated with pre-treatment loss to follow-up, and time to treatment initiation.
      ,
      • Onyoh E.F.
      • Kuaban C.
      • Lin H.-H.
      Pre-treatment loss to follow-up of pulmonary tuberculosis patients in two regions of Cameroon.
      ,
      • Thomas B.E.
      • Subbaraman R.
      • Sellappan S.
      • Suresh C.
      • Lavanya J.
      • Lincy S.
      • et al.
      Pretreatment loss to follow-up of tuberculosis patients in Chennai, India: a cohort study with implications for health systems strengthening.
      ). The high pre-treatment LTFU is an indicator of insufficient tracking of presumptive patients to diagnosis and treatment start if found to have TB, coupled with the lack of same-day sputum testing and delays in results dispatch. Long distance, long travel time and urban location of treatment units have been associated with high pre-treatment LTFU in Cameroon (
      • Onyoh E.F.
      • Kuaban C.
      • Lin H.-H.
      Pre-treatment loss to follow-up of pulmonary tuberculosis patients in two regions of Cameroon.
      ).
      Second, pre-treatment deaths accounted for approximately half of all pre-treatment LTFU cases. There is limited documented evidence on pre-treatment deaths among persons with TB. This high proportion can be explained by the high HIV prevalence although we did not have information on ART uptake.
      Third, in addition to delay during testing, there was a long delay from diagnosis to treatment initiation. Considering the high HIV prevalence among TB patients in Bulawayo and the fact that in many the HIV status would have been known at the time of TB diagnosis, there should be urgency in ensuring that persons living with both HIV and TB are started on TB treatment promptly after laboratory diagnosis. We do not know what happened to persons who did not die but were not started on treatment. They may have experienced even longer delays in accessing treatment elsewhere, possibly contributing to additional adverse pre-treatment outcomes. Long treatment initiation delay is contrary to what has been reported globally (median 2.5 days) (
      • Sreeramareddy C.T.
      • Qin Z.Z.
      • Satyanarayana S.
      • Subbaraman R.
      • Pai M.
      Delays in diagnosis and treatment of pulmonary tuberculosis in India: a systematic review.
      ).
      Fourth, the risk factors for pre-treatment LTFU in this study were old age, male or missing information on gender, and being HIV-positive or having an unknown HIV status. These findings were consistent with previous studies (
      • Storla D.G.
      • Yimer S.
      • Bjune G.A.
      A systematic review of delay in the diagnosis and treatment of tuberculosis.
      ,
      • Claassens M.M.
      • du Toit E.
      • Dunbar R.
      • Lombard C.
      • Enarson D.A.
      • Beyers N.
      • et al.
      Tuberculosis patients in primary care do not start treatment. What role do health system delays play?.
      ,
      • Claassens M.M.
      • Dunbar R.
      • Yang B.
      • Lombard C.J.
      Scanty smears associated with initial loss to follow-up in South African tuberculosis patients.
      ,
      • Mehra D.
      • Kaushik R.M.
      • Kaushik R.
      • Rawat J.
      • Kakkar R.
      Initial default among sputum-positive pulmonary TB patients at a referral hospital in Uttarakhand, India.
      ,
      • Onyoh E.F.
      • Kuaban C.
      • Lin H.-H.
      Pre-treatment loss to follow-up of pulmonary tuberculosis patients in two regions of Cameroon.
      ,
      • Thomas B.E.
      • Subbaraman R.
      • Sellappan S.
      • Suresh C.
      • Lavanya J.
      • Lincy S.
      • et al.
      Pretreatment loss to follow-up of tuberculosis patients in Chennai, India: a cohort study with implications for health systems strengthening.
      ). In addition to the above, delay in testing in the laboratory was also a risk factor for pre-treatment deaths. When compared to previously treated patients, new patients had a higher risk for the delay in treatment initiation, possibly because of the latter’s first experience with TB services.
      Finally, the increasing proportion of persons who were LTFU or died before treatment initiation in 2012–16 in a setting with decreasing trend of TB diagnosis is worrisome (Figure 2, Figure 3). This was despite the decreasing laboratory delays (Figure 1), decrease in HIV prevalence among TB patients (Supplementary Figure S1), substantial increase in use of Xpert MTB/RIF assay among HIV-positive TB patients during the same period (Supplementary Figure S1) and substantial increase in the ART coverage over the past years (
      • UNAIDS
      UNAIDS. Country: Zimbabwe.
      ,
      • Zimbabwe Ministry of Health and Child Care
      Ministry of Health and Child Welfare, National Tuberculosis Program – Strategic Plan (2017-2020).
      ).

      Implications for policy and practice

      Recording and reporting processes in the TB programme in Bulawayo are performing well and documentation was found to be good especially in the presumptive TB registers which enabled us to find out what happened to patients who were not initiated on treatment. However, high pre-treatment LTFU and delay before treatment initiation may contribute to continued transmission of TB bacilli in the community.
      There is need to migrate to the provision of point of care diagnostics for TB, including Xpert MTB/RIF assay, in line with the country’s strategic plan (2017–2020) (
      • Lessells R.J.
      • Cooke G.S.
      • McGrath N.
      • Nicol M.P.
      • Newell M.-L.
      • Godfrey-Faussett P.
      Impact of point-of-care Xpert MTB/RIF on tuberculosis treatment initiation. A cluster-randomized trial.
      ,
      • Zimbabwe Ministry of Health and Child Care
      Ministry of Health and Child Welfare, National Tuberculosis Program – Strategic Plan (2017-2020).
      ). One laboratory for 0.7 million population served by 19 clinics appears insufficient. Until this happens the programme may devise mechanisms to convey results to the clinics in a timely manner. The use of short messaging servce and e-mails to communicated results may be explored. The programme may consider additional laboratories in some clinics which could at least perform the follow-up sputum examinations of drug-susceptible TB patients on treatment to reduce workload from diagnostic tests in the main laboratory and prevent testing delay.
      The slightly increased trend of pre-treatment LTFU/deaths despite a reduction in laboratory delays, improvement in HIV/ART indicators and expansion of Xpert MTB/RIF use may be due to delayed case finding. In addition to improving access to point of care diagnostics, this calls for community-based active finding strategies, especially among marginalised/vulnerable population. This also includes strengthening of intensified TB finding among people living with HIV (
      • World Health Organization
      Systematic Screening for Active Tuberculosis: An Operational Guide. Geneva, Switzerland, 2015.
      ).
      The programme in Bulawayo should consider cohort-wide reporting from bacteriological-confirmation instead of TB registration. Reporting of pre-treatment LTFU, deaths and delays in the quarterly reports is the need of the hour and should be closely monitored.

      Future research

      First, a qualitative study to explore the drivers of high pre-treatment delay among bacteriologically confirmed TB in a high HIV prevalence setting is required. Second, once the measures for improved access to diagnosis are implemented, operational research may be considered for tracking the care cascade in the pre-diagnosis phase-in programme setting (presumptive TB patients from sputum referral to testing and result receipt). Considering Bulawayo is one of the best performing areas of Zimbabwe, there is a need for similar studies over a long period of time (enabling assessment of trends) in other parts of Zimbabwe.

      Conclusion

      In Bulawayo, Zimbabwe (2012–16), we identified high pre-treatment deaths that have increased in the past five years. Delay in treatment initiation was significantly longer than in other studies. The TB programme in Bulawayo needs to address these urgently if we are to move towards ending TB by 2035 (
      • World Health Organization
      The End TB Strategy. Global strategy and targets for tuberculosis prevention, care and control after 2015.
      ).

      Ethics approval and consent to participate

      Approval to conduct this study was obtained from the Director of the Health Services Department and the Institutional Ethics Review Board for Bulawayo City, the Medical Research Council of Zimbabwe and the Ethics Advisory Group of the International Union against Tuberculosis and Lung Disease, Paris, France. As this study involved secondary data, waiver of informed consent was sought and obtained from the ethics committees.

      Consent for publication

      Not applicable.

      Availability of data and material

      The dataset used in this study has been provided in Supplementary Annex S1.

      Competing interests

      The authors declare that they have no competing interests.

      Funding

      The training course under which this study was conducted was funded by the United Kingdom’s Department for International Development (DFID) ; La Fondation Veuve Emile Metz-Tesch (Luxembourg) ; the United States Agency for International Development (USAID) through Challenge TB; The Global Fund to Fight AIDS, Tuberculosis and Malaria (GFATM) and the National AIDS Council Zimbabwe . The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

      Authors’ contributions

      Conception and design: HM, HDS, RAD; development of data capture tools: HM, HDS; data collection: HM, ES; SH; data entry: HM; data analysis and interpretation: all authors; preparing first draft of manuscript: HM, HDS, RAD; critical review and approval of final draft: all authors.

      Acknowledgements

      We would like to acknowledge the Structured Operational Research and Training Initiative (SORT IT) through which this research was conducted. It is a global partnership led by the Special Programme for Research and Training in Tropical Diseases at the World Health Organization (WHO/TDR). The training model is based on a course developed jointly by the International Union Against Tuberculosis and Lung Disease (The Union) and Medécins sans Frontières (MSF). The specific SORT IT program which resulted in this publication was implemented by the Centre for Operational Research, The Union, Paris, France. Mentorship and the coordination/facilitation of this particular SORT IT course was provided through the Centre for Operational Research, The Union, Paris, France; The Union South-East Asia Office, New Delhi, India; the Department of Tuberculosis and HIV, The Union, Paris, France; the University of Washington, School of Public Health, Department of Global Health, Seattle, Washington, USA; and AMPATH, Eldoret, Kenya.
      We would like to thank Joyline Nkala and Emely Mthethwa for their support in extracting records and following up clients for treatment initiation. We also acknowledge the National AIDS and TB Programme, Ministry of Health and Child Care and Bulawayo Health Services Department for the support while implementing the study.

      Disclaimer

      The contents of this paper do not necessarily reflect the views of Ministry of Health and Child Care, Zimbabwe; Bulawayo Health Services Department or The Union.

      Appendix A. Supplementary data

      The following are Supplementary data to this article:
      • Figure S2

        The proportion of patients diagnosed using Xpert MTB/RIF* assay among HIV-positive bacteriologically confirmed drug-sensitive pulmonary tuberculosis in Bulawayo, Zimbabwe, 2012–16. *Genotypic diagnostic test for tuberculosis – cartridge-based nucleic acid amplification test.

      References

        • Afutu F.K.
        • Zachariah R.
        • Hinderaker S.G.
        • Ntoah-Boadi H.
        • Obeng E.A.
        • Bonsu F.A.
        • et al.
        High initial default in patients with smear-positive pulmonary tuberculosis at a regional hospital in Accra, Ghana.
        Trans R Soc Trop Med Hyg. 2012; 106: 511-513
        • Chadambuka A.
        • Mabaera B.
        • Tshimanga M.
        • Shambira G.
        • Gombe N.T.
        • Chimusoro A.
        Low tuberculosis case detection in Gokwe North and South, Zimbabwe in 2006.
        Afr Health Sci. 2011; 11: 190-196
        • Claassens M.M.
        • Dunbar R.
        • Yang B.
        • Lombard C.J.
        Scanty smears associated with initial loss to follow-up in South African tuberculosis patients.
        Int J Tuberc Lung Dis. 2017; 21: 196-201
        • Claassens M.M.
        • du Toit E.
        • Dunbar R.
        • Lombard C.
        • Enarson D.A.
        • Beyers N.
        • et al.
        Tuberculosis patients in primary care do not start treatment. What role do health system delays play?.
        Int J Tuberc Lung Dis. 2013; 17: 603-607
        • Dowdy D.W.
        • Chaisson R.E.
        The persistence of tuberculosis in the age of DOTS: reassessing the effect of case detection.
        Bull World Health Organ. 2009; 87: 296-304
        • Harries A.D.
        • Rusen I.D.
        • Chiang C.-Y.
        • Hinderaker S.G.
        • Enarson D.A.
        Registering initial defaulters and reporting on their treatment outcomes.
        Int J Tuberc Lung Dis. 2009; 13: 801-803
        • Lessells R.J.
        • Cooke G.S.
        • McGrath N.
        • Nicol M.P.
        • Newell M.-L.
        • Godfrey-Faussett P.
        Impact of point-of-care Xpert MTB/RIF on tuberculosis treatment initiation. A cluster-randomized trial.
        Am J Respir Crit Care Med. 2017; 196: 901-910
        • MacPherson P.
        • Houben R.M.
        • Glynn J.R.
        • Corbett E.L.
        • Kranzer K.
        Pre-treatment loss to follow-up in tuberculosis patients in low- and lower-middle-income countries and high-burden countries: a systematic review and meta-analysis.
        Bull World Health Organ. 2014; 92: 126-138
        • Mehra D.
        • Kaushik R.M.
        • Kaushik R.
        • Rawat J.
        • Kakkar R.
        Initial default among sputum-positive pulmonary TB patients at a referral hospital in Uttarakhand, India.
        Trans R Soc Trop Med Hyg. 2013; 107: 558-565
        • Ministry of Health and Child Care (MoHCC), Zimbabwe, ZIMSTAT
        Zimbabwe Populations based HIV impact assessment, ZIMPHIA 2016.
        2017
        • Mwansa-Kambafwile J.
        • Maitshotlo B.
        • Black A.
        Microbiologically confirmed tuberculosis: factors associated with pre-treatment loss to follow-up, and time to treatment initiation.
        PLoS One. 2017; 12e0168659
        • Onyoh E.F.
        • Kuaban C.
        • Lin H.-H.
        Pre-treatment loss to follow-up of pulmonary tuberculosis patients in two regions of Cameroon.
        Int J Tuberc Lung Dis. 2018; 22: 378-384
        • Squire S.B.
        • Belaye A.K.
        • Kashoti A.
        • Salaniponi F.M.L.
        • Mundy C.J.F.
        • Theobald S.
        • et al.
        “Lost” smear-positive pulmonary tuberculosis cases: where are they and why did we lose them?.
        Int J Tuberc Lung Dis. 2005; 9: 25-31
        • Sreeramareddy C.T.
        • Qin Z.Z.
        • Satyanarayana S.
        • Subbaraman R.
        • Pai M.
        Delays in diagnosis and treatment of pulmonary tuberculosis in India: a systematic review.
        Int J Tuberc Lung Dis. 2014; 18: 255-266
        • Storla D.G.
        • Yimer S.
        • Bjune G.A.
        A systematic review of delay in the diagnosis and treatment of tuberculosis.
        BMC Public Health. 2008; 8: 15
        • Takarinda K.C.
        • Harries A.D.
        • Nyathi B.
        • Ngwenya M.
        • Mutasa-Apollo T.
        • Sandy C.
        Tuberculosis treatment delays and associated factors within the Zimbabwe national tuberculosis programme.
        BMC Public Health. 2015; 15: 29
        • Thomas B.E.
        • Subbaraman R.
        • Sellappan S.
        • Suresh C.
        • Lavanya J.
        • Lincy S.
        • et al.
        Pretreatment loss to follow-up of tuberculosis patients in Chennai, India: a cohort study with implications for health systems strengthening.
        BMC Infect Dis. 2018; 18: 142
        • UNAIDS
        UNAIDS. Country: Zimbabwe.
        2016 (Available from: http://www.unaids.org/en/regionscountries/countries/zimbabwe. [cited 2018 April 17].)
        • UNDP and Government of Zimbabwe
        Zimbabwe Human Development Report 2017. Climate Change and Human Development: Towards Building a Climate Resilient Nation.
        2017 (Harare, Zimbabwe)
        • World Health Organization
        Definitions and reporting framework for tuberculosis — December 2013 revision (updated December 2014).
        2014 (Geneva, Switzerland)
        • World Health Organization
        The End TB Strategy. Global strategy and targets for tuberculosis prevention, care and control after 2015.
        2014 (Geneva, Switzerland)
        • World Health Organization
        Systematic Screening for Active Tuberculosis: An Operational Guide. Geneva, Switzerland, 2015.
        2015 (Geneva, Switzerland)
        • World Health Organization
        Global Tuberculosis Report 2017.
        2017 (Geneva, Switzerland)
        • World Population Review
        Zimbabwe Population 2018.
        2018 (Available from: http://worldpopulationreview.com/countries/zimbabwe-population/. [cited 2018 April 17])
        • Zimbabwe Ministry of Health and Child Care
        Ministry of Health and Child Welfare, National Tuberculosis Program – Strategic Plan (2017-2020).
        2018 (Harare, Zimbabwe)
        • Zimbabwe Ministry of Health and Child Care
        Zimbabwe National Tuberculosis and Leprosy Management Guidelines.
        2018 (Harare, Zimbabwe)