A user-friendly, open-source tool to project impact and cost of diagnostic tests for tuberculosis

  1. David W Dowdy  Is a corresponding author
  2. Jason R Andrews
  3. Peter J Dodd
  4. Robert H Gilman
  1. Johns Hopkins Bloomberg School of Public Health, United States
  2. Johns Hopkins University, United States
  3. Massachusetts General Hospital, United States
  4. University of Sheffield, United Kingdom
4 figures, 2 tables and 2 additional files

Figures

Overview of user-friendly model.

Users are asked, via open-source computer script or Web interface, to select one of the nine diagnostic strategies and to provide unit costs and three basic epidemiological parameters (TB incidence, …

https://doi.org/10.7554/eLife.02565.003
Transmission model of TB diagnosis.

Boxes represent sub-populations in the model, and arrows represent rates of movement between those sub-populations. Parallel structures exist for: (a) HIV-infected vs HIV-uninfected; (b) …

https://doi.org/10.7554/eLife.02565.004
Incremental 5-year cost and impact of TB diagnostic strategies, high-incidence setting.

Shown are cumulative projected 5-year costs and impact (averted TB cases [panel A] or MDR-TB cases [panel B]) of each diagnostic strategy described in the Introduction, incremental to the baseline …

https://doi.org/10.7554/eLife.02565.006
Relative impact of diagnostic strategies in emblematic settings.

Shown are projected changes in TB incidence, MDR-TB incidence, TB mortality, and costs (in Year 1 and Year 5 after immediate implementation), relative to baseline (Strategy 1) after implementing …

https://doi.org/10.7554/eLife.02565.007

Tables

Table 1

Model input parameters*

https://doi.org/10.7554/eLife.02565.005
ParameterValueReference(s)/Rationale
TB and HIV Transmission
Transmission rate, per smear-positive/highly infectious person-yearCalibrated to user-defined TB incidence
Proportional reduction in per-case transmission rate, MDR-TBCalibrated to user-defined MDR-TB prevalence
Proportional reduction in fitness, isoniazid-monoresistant TB25% of MDR-TB reductionAssumption
HIV incidence rate, per yearCalibrated to user-defined HIV prevalence
Relative transmission rate from smear-negative/less infectious TB0.22(Behr et al., 1999)
Proportion of pulmonary TB that is smear-positive/highly infectious
 HIV-negative0.63(Steingart et al., 2006a; Steingart et al., 2006b)
 HIV-infected0.50(Getahun et al., 2007)
TB Progression
Endogenous reactivation rate
 HIV-negative0.0005/year(Horsburgh et al., 2010)
 HIV-infected0.05/year(Antonucci et al., 1995)
Proportion of recent infections resulting in rapid progression
 HIV-negative0.14(Vynnycky and Fine, 1997; Dye et al., 1998)
 HIV-infected0.470.75 without ART, (Daley et al., 1992)
75% reduction if on ART, (Williams et al., 2010) 50% ART coverage
Reduction in TB rapid progression probability due to latent TB infection (HIV-negative only)0.79(Andrews et al., 2012)
TB Mortality and Resolution
Life expectancy at age 1555 years(World Bank, 2012)
Annual mortality from HIV0.05/year(UNAIDS, 2012)
Annual mortality from TB
 HIV-negative, smear-positive/highly infectious0.23/year(Tiemersma et al., 2011)
 HIV-negative, smear-negative/less infectious0.07/year(Tiemersma et al., 2011)
 HIV-infected1.0/year(Corbett et al., 2003; Corbett et al., 2007; Wood et al., 2007)
Rate of spontaneous TB resolution (HIV-negative only)
 Smear-positive/highly infectious0.1/year(Tiemersma et al., 2011)
 Smear-negative/less infectious0.27/year(Tiemersma et al., 2011)
TB Treatment Outcomes and Emergence of Drug Resistance
Probability of failure or relapse (within 1 year)
 Drug-susceptible0.04(World Health Organization, 2012)
 INH-monoresistant, first-line therapy0.21(Menzies et al., 2009b)
 INH-monoresistant, retreatment or 2nd-line0.16(Menzies et al., 2009b)
 MDR-TB, first-line or retreatment0.50(Espinal et al., 2000)
 MDR-TB, second-line therapy0.30(World Health Organization, 2010)
Proportion of one-year recurrence due to failure
 Drug-susceptible0.14(Lew et al., 2008)
 INH-monoresistant0.33
 MDR-TB0.56
Probability of acquired drug resistance (per treatment course)
 Susceptible becoming INH-monoresistant0.001(Menzies et al., 2009a; Menzies et al., 2009b)
 Susceptible becoming MDR-TB0.002
 INH-monoresistant becoming MDR-TB0.045
 If treated with 2 effective drugs for >6 mos0.017
Behavioral Parameters
Infectious months before starting to seek care
 HIV-negative9 months(Dowdy et al., 2013)
 HIV-infected1 month(Corbett et al., 2004)
Diagnostic frequency while seeking care5.0/year(Storla et al., 2008; Sreeramareddy et al., 2009)
Probability of treatment in a TB patient whose microbiological test is negative0.25(Wilkinson et al., 2000; Dowdy et al., 2008)
Loss to follow-up between diagnostic presentation and treatment initiation
 Sputum smear or GXP (not same-day)0.15(MacPherson et al., 2014)
 Culture (microcolony or commercial liquid)0.25(Dowdy et al., 2008)
 Same-day diagnosis0Assumption
Diagnostic Accuracy
Sensitivity for smear-negative/less-infectious TB
 Sputum smear microscopy0
 Xpert MTB/RIF0.72(Brownell et al., 2012)
 Culture (microcolony or commercial liquid)0.85(Cruciani et al., 2004; Leung et al., 2012)
Specificity for TB(Steingart et al., 2006; Boehme et al., 2011; Leung et al., 2012)
 Sputum smear microscopy0.98
 Xpert MTB/RIF0.98
 Microcolony culture0.98
Sensitivity for drug resistance (if TB detected)
 Microcolony culture (rifampin and isoniazid)0.98(Minion et al., 2010)
 Xpert MTB/RIF (rifampin only)0.94(Boehme et al., 2011)
Specificity for drug resistance (if TB detected)
 Microcolony culture (isoniazid)0.96(Minion et al., 2010)
 Microcolony culture (rifampin)0.99(Minion et al., 2010)
 Xpert MTB/RIF (rifampin)0.98(Boehme et al., 2011)
Diagnostic Delay and non-TB Care-Seeking
Days from presentation to treatment initiation
 Sputum smear or Xpert MTB/RIF7 daysAssume 1 week
 Microcolony or commercial liquid culture30 days(Boehme et al., 2011)
Months of therapy before a failing regimen will be changed, or before default and recurrence6 monthsAssumption
Annual rate of diagnostic evaluation for TB, among people who do not have active TB0.01/year10% of suspects have TB, high-incidence setting
Cost Parameters (user-defined; values below for comparison purposes only)
Per-patient cost of TB therapy
 First-lineUS$500User-defined
 RetreatmentUS$1000(Vassall et al., 2011)
 Second-line/MDRUS$5000(Vassall et al., 2011)
 Outpatient visit (diagnosis or follow-up)US$10(Vassall et al., 2011)
Per-test cost:
 Sputum smearUS$2(Vassall et al., 2011)
 Same-day sputum smearUS$10Assumption
 Xpert MTB/RIFUS$15(Vassall et al., 2011)
 Same-day Xpert MTB/RIFUS$30Assumption
 Microcolony culture (with DST)US$5(Solari et al., 2011)
 Commercial liquid-media cultureUS$20(Vassall et al., 2011)
 Commercial liquid-media culture + DSTUS$40(Vassall et al., 2011)
  1. *

    In the actual model program (Supplementary file 1), users can change any parameter based on local values.

  2. For reference, the transmission rate (in infections per person-year during diagnosis-seeking active TB) is 36.9 in the reference scenario, 14.0 in the low-incidence scenario, 25.4 in the high MDR scenario, and 12.9 in the high HIV scenario. Corresponding proportional reductions in MDR-TB transmission rate are 0.23, 0.23, 0.21, and 0.19; and HIV incidence estimates (per 1000 adult person-years) are 0.7, 0.6, 0.6, and 18.9.

Table 2

Model parameters and symbolic representations

https://doi.org/10.7554/eLife.02565.008
ParameterRepresentationBaseline value (see Table 1)
Transmission rate (transmission events per highly infectious person-year)βCalibrated to TB incidence
Proportional reduction in per-case transmission rate
 Drug-susceptible TBφ01.0
 Isoniazid-monoresistant TBφ125%* of φ2
 MDR-TBφ2Calibrated
HIV incidence rate, per yearθCalibrated to HIV prevalence
Relative transmission rate from smear-negative/less infectious TBζ0.22
Proportion of pulmonary TB that is smear-positive/highly infectious
 HIV-negativeψ00.63
 HIV-infectedψ10.50
Endogenous reactivation rate, per year
 HIV-negativeε00.005
 HIV-infectedε10.05
Proportion of recent infections resulting in rapid progression
 HIV-negativeπ00.14
 HIV-infectedπ10.47
Reduction in TB rapid progression probability due to latent TB infection
 HIV-negativeι0.79
 HIV-infectedNot included0
Baseline mortality rate, per yearμbl1/55 = 0.018
Additional HIV-related mortality rate, per yearμh0.05
Additional untreated TB-related mortality rate, per year
 HIV-negative, smear-positive/highly infectiousμt10.23
 HIV-negative, smear-negative/less infectiousμt00.07
 HIV-infectedμth1.0
Rate of spontaneous TB resolution, per year
 Smear-positive/highly infectiousν10.1
 Smear-negative/less infectiousν00.27
 HIV-infectedNot included0
Rate of starting diagnosis-seeking in active TB, per year
 HIV-negativeδe01.33 (9 months)
 HIV-infectedδe112 (1/month)
Rate of progression: ineffective therapy to repeat therapy (failure) or active TB (relapse), per yearδf6/12 = 0.5
Rate of diagnostic evaluation for TB, per year
 Late active TBInput into decision tree5.0
 No active TBτ00.01
Decision tree outputs (in addition to unit costs):Vary by intervention
 Successful diagnosis rate of late active TB, per yearσhdip
 Rate of movement from successful diagnosis to treatment (1/diagnostic delay), per yearρhdip
 Ineffective diagnosis rate of late active TB, per yearκhdip
 Rate of diagnosis and treatment leading to new resistance, per year
 susceptible to INH-monoresistantαsihip
 susceptible to MDRαsmhip
 INH-monoresistant to MDRαimhip
  1. *

    Calculated such that (1−φ1) = 0.25*(1−φ2).

Additional files

Download links