Jason Andrews, MD, SM, DTM&HDivision of Infectious Diseases Massachusetts General Hospital
Harvard Medical School
International AIDS SocietyJuly 2, 2013
Intensified case finding in the era of new diagnostics: what is the impact?
Introduction: Active Case Finding (ACF)
• There are good data on yield of ACF for TB via community-based case finding or household contact investigations
• Meta-analysis of 27 studies: 4.5% HH contacts had active TB
• Shapiro et al. found at least one TB case in 19% of HHs where an index case had been found and 1% of random households visited in South Africa
• Fewer data on the individual level clinical benefits of early case detection.
• Even more limited evidence base (DETECTB, ZAMSTAR) on population level epidemiologic impact of ACF interventions
Morrison et al., Lancet Infect Dis 2008Shapiro et al., AJRCCM 2012
What can mathematical models tell us about the benefits of active case finding for TB?
Not much… in quantitative terms.
Limitations in understanding of:
1.Infectiousness over time
2.Duration of subclinical tuberculosis
3.Role of social contact structure
Overview
1. Explain critical assumptions and limitations of TB diagnostic models
2. How these assumptions impact projections for community-based ACF and household contact investigations
3. What we need to make better projections
Examining these assumptions:impact on ACF
Dowdy DW, Basu S, Andrews JR. Am J Resp Crit Care Med, 2012
Passive CFActive CF
Model and Results
• If individuals are asymptomatic (“subclinical”, and only 0.25x as infectious) for half of their duration of TB:
- 20% increase in passive diagnosis would reduce TB incidence by 11% over 10 years
- ACF targeted at 5% of population would reduce TB incidence by 16% over 10 years
• As duration/infectiousness subclinical period increases, efficacy of passive case detection decreases
• ACF needed if much of transmission occurs during a subclinical period
Dowdy DW, Basu S, Andrews JR. Am J Resp Crit Care Med, 2012
?
?
Examining these assumptions:household contact investigations
Household
InfectiousExposedSusceptible
Household
InfectiousExposedSusceptible
Household
InfectiousExposedSusceptible
Household
InfectiousExposedSusceptible
Household
InfectiousExposedSusceptible
Household Contacts
• Created a simple model, governed by ordinary differential equations, with two levels of transmission: within and between households
• Calibrated it to TB prevalence of 170/100,000• Varied proportion of TB transmitted within the
household versus within the community**• Projected the impact of:
1) HH contact investigations for active TB; 2) HH contact investigations + LTBI treatment
Assumed 100% coverage of intervention and “perfect” sensitivity of diagnostic testing
0%
10%
20%
30%
40%
50%Reduction in TB Prevalence over 10 years
Screening Contacts Screening + LTBI Treatment
Percentage of all TB transmitted within Household
Redu
ction
in T
B Pr
eval
ence
20% 40%0%
Cape Town
Conclusions
• We have poor data on role of subclinical TB in transmission, which is critical to projecting the impact of passive and active case finding approaches
• Epidemiologic impact of household contact investigations is somewhat attenuated due to mutual contacts between the index and secondary case
• Data on within-household versus between- household transmission could improve quantitative projections on contact investigation interventions
Acknowledgements
Robin Wood, FCP(SA), DSc – University of Cape Town
Rochelle Walensky, MD, MPH – MGH/Harvard
David Dowdy, MD, PhD – Johns Hopkins
Sanjay Basu, MD, PhD – Stanford
Megan Murray, MD, ScD - Harvard