1 of 25

Risk-benefit analysis of TB infection testing for household contact management

Courtney M. Yuen, PhD

Brigham and Women’s Hospital

Harvard Medical School

2 of 25

Household contact management

  • Household contacts of TB patients are at high risk for developing TB themselves
    • High probability of recent infection
    • Best practice is to evaluate all for TB disease; preventive therapy can then prevent disease in those who are infected but not sick

​

Figure from Trauer JM et al. Chest 2016; 149(2):516-525.

0-4 years old

5-14 years old

≥15 years old

Days since infection

3 of 25

WHO LTBI guideline update 2018

  • Pre-2018 double standard of care
    • Low-incidence, high-resource settings: test contacts for TB infection and give preventive therapy to all who are infected
    • Other settings: give preventive therapy only to contacts who are <5 years old or who have HIV

​

  • 2018 consolidated guidelines
    • Algorithm for low-incidence, high-resource settings was recommended for all settings

4 of 25

Issues with TB infection testing for contact management in high-burden countries

  • TB infection testing may be less useful in high-burden settings
    • Negative predictive value may be poor if prevalence of infection is high
    • If the vast majority of contacts test positive, is it worth the trouble of testing?

​

  • Logistical barriers to TB infection testing
    • TST: tuberculin shortage, cold chain storage
    • IGRA: laboratory infrastructure

5 of 25

Open questions

  • In high burden settings, when is TB infection testing useful as part of the preventive therapy algorithm for household contacts?

​

  • How many incident TB cases and how many severe adverse events might one expect in a high-burden setting if one were to treat all contacts? If one were to treat only TST-positive contacts?

6 of 25

Methods

7 of 25

Objective

  • Compare two preventive therapy strategies:
    • Preventive therapy for all household contacts once TB disease ruled out
    • Preventive therapy for only TST-positive household contacts once TB disease ruled out

​

  • Compare risk of incident TB disease and risk of severe adverse event
    • Severe adverse event = grade 3-5
    • Separate analyses for different age groups and preventive therapy regimens, and assuming different prevalences of TST positivity

8 of 25

Model structure

TST-positive

TST-negative

Household contact of TB patient

Preventive therapy

No preventive therapy

Preventive therapy

No preventive therapy

Adverse event

No adverse event

TB disease

No TB disease

TB disease

No TB disease

TB disease

No TB disease

Adverse event

No adverse event

TB disease

No TB disease

TB disease

No TB disease

TB disease

No TB disease

9 of 25

No treatment

TST-positive

TST-negative

Household contact of TB patient

Preventive therapy

No preventive therapy

Preventive therapy

No preventive therapy

Adverse event

No adverse event

TB disease

No TB disease

TB disease

No TB disease

TB disease

No TB disease

Adverse event

No adverse event

TB disease

No TB disease

TB disease

No TB disease

TB disease

No TB disease

10 of 25

Treat all

TST-positive

TST-negative

Household contact of TB patient

Preventive therapy

No preventive therapy

Preventive therapy

No preventive therapy

Adverse event

No adverse event

TB disease

No TB disease

TB disease

No TB disease

TB disease

No TB disease

Adverse event

No adverse event

TB disease

No TB disease

TB disease

No TB disease

TB disease

No TB disease

11 of 25

Treat TST-positive

TST-positive

TST-negative

Household contact of TB patient

Preventive therapy

No preventive therapy

Preventive therapy

No preventive therapy

Adverse event

No adverse event

TB disease

No TB disease

TB disease

No TB disease

TB disease

No TB disease

Adverse event

No adverse event

TB disease

No TB disease

TB disease

No TB disease

TB disease

No TB disease

12 of 25

Data sources

  • Risk of TB disease
    • Natural progression once infected from Australian cohort
      • Different for children and adults, cumulative risk 3 months to 3 years post-infection
    • Risk ratio for progression in TST+ vs TST- from UK prospective cohort study
    • Protective effect of preventive therapy from network meta-analysis of efficacy trials

​

  • Age-stratified grade 3-4 adverse event risk attributable to treatment
    • Unpublished data from 4R vs 9H adult trial (Menzies et al) and from 3HP vs 9H adult trial (Sterling et al, TB Trials Consortium)
    • Published data from corresponding pediatric trials

13 of 25

Modeling

  • Different conditions modeled:
    • Age groups: 0-17, 18-34, 35-64, 65+
    • Regimens: no treatment, 3HP, 4R, 6H
    • Prevalence of TST-positivity: 10%, 25%, 50%, 75%

​

  • Modeled risks of incident TB disease and of severe adverse events
    • Modeled 10,000 sampled input parameter sets
    • Express uncertainty as interquartile range (IQR) around mean output

​

14 of 25

Results

15 of 25

3-year risks of TB and severe AE

Untreated TST-positive children have a ~15% risk of developing disease in 3 years

Treating children reduces disease risk to ~5% with minimal (<1%) risk of severe adverse events

Adverse event risk increases with age and differs by regimen

16 of 25

Risks of TB and severe AE for two approaches in children 0-17 years old with 25% TST+

Regimen

Treat all

Treat TST-positive only

TB cases per 1000 contacts (IQR)

Severe AE per 1000 contacts (IQR)

TB cases per 1000 contacts (IQR)

Severe AE per 1000 contacts (IQR)

No treatment

53.3 (37.3, 66.5)

0.0 (0.0, 0.0)

53.3 (37.3, 66.5)

0.0 (0.0, 0.0)

3HP

14.9 (8.8, 18.8)

5.6 (2.9, 7.4)

27.7 (18.2, 34.9)

1.4 (0.7, 1.9)

4R

21.0 (12.9, 26.7)

3.0 (0.9, 4.2)

31.9 (21.3, 39.7)

0.7 (0.2, 1.0)

6H

23.3 (15.2, 29.2)

2.5 (0.7, 3.5)

33.4 (22.8, 41.6)

0.6 (0.2, 0.9)

17 of 25

Risks of TB and severe AE for two approaches in adults ≥65 years old with 25% TST+

Regimen

Treat all

Treat TST-positive only

TB cases per 1000 contacts (IQR)

Severe AE per 1000 contacts (IQR)

TB cases per 1000 contacts (IQR)

Severe AE per 1000 contacts (IQR)

No treatment

8.8 (6.5, 10.7)

0.0 (0.0, 0.0)

8.8 (6.5, 10.7)

0.0 (0.0, 0.0)

3HP

2.8 (1.9, 3.4)

84.0 (67.1, 98.7)

4.8 (3.4, 5.8)

21.0 (16.7, 24.7)

4R

3.3 (2.2, 4.1)

23.1 (12.0, 30.9)

5.1 (3.6, 6.3)

5.8 (3.0, 7.7)

6H

3.9 (2.7, 4.7)

55.1 (40.0, 67.6)

5.5 (4.0, 6.7)

13.8 (10.0, 16.9)

18 of 25

Treat-all versus treat-TST-only

Change in outcomes per 1000 contacts by not using TST

TB disease

Severe AE

3HP

4R

Age group

0-17

18-34

35-64

65+

0-17

18-34

35-64

65+

For a cohort of child contacts with 10% TST+, treating all with 3HP would lead to differences of:

-15 (IQR -22, -6) TB cases

+5 (IQR 3, 7) severe AEs

For a cohort of adults 65+ with 50% TST+, treating all with 3HP would lead to differences of:

-1 (IQR -2, -1) TB cases

+42 (IQR 34, 49) severe AEs

For a cohort of adults 65+ with 50% TST+, treating all with 4R would lead to differences of:

-1 (IQR -2, 0) TB cases

+12 (IQR 6, 15) severe AEs

19 of 25

Implications for high- and low-burden settings

Regimen

India hypothetical cohort

Netherlands hypothetical cohort

Difference in TB cases per 1000 contacts (IQR)

Difference in severe AE per 1000 contacts (IQR)

Difference in TB cases per 1000 contacts (IQR)

Difference in severe AE per 1000 contacts (IQR)

3HP

-4.2 (-5.7, -2.1)

9.1 (8.3, 9.8)

-3.1 (-3.9, -2.2)

37.9 (33.8, 41.4)

4R

-3.6 (-5.2, -1.5)

3.3 (2.7, 3.8)

-2.7 (-3.6, -1.8)

10.9 (8.2, 12.8)

6H

-3.3 (-5.0, -1.3)

6.2 (5.6, 6.8)

-2.5 (-3.4, -1.6)

25.7 (22.0, 28.8)

Difference between a treat-all policy and a treat-TST-only policy applied in hypothetical cohorts based on published cohorts from India and the Netherlands

20 of 25

Conclusions

21 of 25

Summary of findings

  • For children 0-17 years old, a treat-all approach would prevent more TB cases than treating only TST+
    • Reduction in incident TB cases was greater than increase in severe adverse events under all conditions modeled

​

  • For adults, a treat-all approach would prevent more TB cases, but also incur more adverse events
    • Least additional adverse events with 4R
    • Additional adverse events increases with age

22 of 25

Limitations

  • Choice of outcomes
    • TB disease and severe adverse events are not binary experiences
    • Different patients may value the two differently
    • Untreated disease impacts people other than the patient via transmission
    • Did not capture adverse events related to TB disease treatment

​

  • Limited data for model parameters
    • Limited age-stratified natural history data
    • Did not account for sex, HIV, other comorbidities
    • No data on 3HR or 1HP

​

23 of 25

Conclusions

  • For children <18 years old, there is no clear benefit to infection testing when prevalence of infection is high
    • Requiring a positive TST could lead to more cases due to limited sensitivity
    • Tests with better sensitivity than TST could reduce false negatives
    • Risk of serious adverse events is low with all regimens evaluated

​

  • For adults, minimizing adverse events can be achieved by:
    • Focusing treatment on those with a positive test for infection
    • Using a rifampin-only based regimen
    • Building robust systems for adverse event monitoring and management

​

​

24 of 25

Acknowledgements

Collaborators

  • Pete Dodd, University of Sheffield, UK
  • James Seddon, Imperial College London, UK, and Stellenbosch University, South Africa
  • Salmaan Keshavjee, Harvard Medical School, USA

​

​

Unpublished data

  • Jonathon Campbell and Dick Menzies, McGill University
  • Erin Sizemore, Nigel Scott, and the TB Trials Consortium, Centers for Disease Control and Prevention

Funding

  • Bill and Melinda Gates Foundation (TB Modelling and Analysis Consortium, OPP1084276)

25 of 25

References

  • References for model parameters
    • Trauer JM et al. Chest. 2016;149(2):516-25.
    • Zenner D et al. Ann Intern Med. 2017;167(4):248-55.
    • Abubakar I et al. Lancet Infect Dis. 2018;18(10):1077-87.
    • Villarino ME et al. JAMA Pediatr. 2015;169(3):247-55.
    • Sterling TR et al. N Engl J Med. 2011;365(23):2155-66.
    • Diallo T et al. N Engl J Med. 2018;379(5):454-63.
    • Menzies D et al. N Engl J Med. 2018;379(5):440-53.

​