HOW WELL DO IGRAS PERFORM IN THE IDENTIFICATION OF PERSONS WHO ARE AT HIGH RISK OF PROGRESSING TO ACTIVE TB? Dr Pranab Haldar MD MRCP Senior Lecturer and Consultant Physician University Hospitals of Leicester UK THE ANSWER It is impossible for both The Ultimate Answer and The Ultimate Question to be known about in the same universe.. There is no single answer to this question The answer is dependent upon the context The answers are not known at this time
OVERVIEW Implications of the question to practice Review of the literature to date Two recent meta-analyses How should we interpret the outcomes? What may be learnt from the data? Contact t screening with single-step IGRA, the Leicester experience Looking ahead HOW WELL DO IGRAS PERFORM IN THE IDENTIFICATION OF PERSONS WHO ARE AT HIGH RISK OF PROGRESSING TO ACTIVE TB? Clinical need should drive the research agenda CLINICAL PRACTICE Inform the utility of IGRA based models for screening high risk populations Optimise the utilisation of IGRAs for implementation in clinical practice RESEARCH PRACTICE To understand the pathogenesis of disease progression To characterise different phenotypes of latent infection o Contribution i of IGRAs o Relate to primary outcome
CONTROLLING TB: GLOBAL PERSPECTIVE DOTS strategy: - Early identification - Early and effective treatment Incident TB Untreated Treated TIME Prevalent TB Death Clearance LTBI Population Exposure Eradication of TB can only be achieved through the control of LTBI Clinical infection Prevalent LTBI Sub-clinical infection PRINCIPLES OF SCREENING The disease has an established natural history that is predictable The disease has an early asymptomatic phase This stage may be identified reliably Safe interventions are available that can alter the natural history at this stage to prevent clinical disease Burden of infectious TB INFECTION CYCLE Infectious TB Exposure Contact screening Subclinical infection (LTBI)? Clinical infection (TB) Mortality Immigrant screening? Burden of Imported LTBI Noninfectious TB Adapted from Reider 1995
EFFICACY OF SCREENING NUMBER NEEDED TO TREAT (NNT): Number of people requiring intervention to prevent one adverse event ( 1/Δ absolute risk) NUMBER NEEDED TO SCREEN (NNS): Number of people that need to be screened to prevent one adverse event ( 1/risk prevalence) Efficiency of testing strategy Efficacy of treatment / intervention TARGETED SCREENING Subgroups at high risk of LTBI Population LTBI at high risk of progression LTBI TST No LTBI Progression TB IGRA No progression THEORETICAL BENEFITS OF IGRA OVER TST FOR SCREENING TST IGRA Immunological specificity Detection ti of cell-mediated d Detection ti of cell-mediated d immunoreactivity to antigens immunoreactivity to antigens Immunoreactivity to Mtb complex Immunoreactivity to Mtb complex Immunoreactivity to NTM and BCG Long term immunoreactivity Booster effect with repeated testing Requires two visits Requires expert (subjective) interpretation of induration Cheap Implementation Immunoreactivity to NTM and no BCG cross-reactivity? Shorter term immunoreactivity No booster effect Single visit Objective outcome on continuous scale effective quantitation Comparatively expensive
LTBI AND PROGRESSION RISK Progression risk is heterogeneous highest risk with recent infection TST +ve children Adapted from G Comstock et al. Am J Epidemiol. 1974 Feb;99(2):131-8 Recent contact P Hill et al. PLoS One. 2008 Jan 2;3(1):e1379 EFFECTS OF TREATMENT Prospective intervention studies in high risk cohorts Natural decline in risk with time Time of exposure not known TST + Fibrotic lesions on CXR Consistent with the idea that any cohort will have a mix of recent / remote infection
NON-UK-BORN TUBERCULOSIS CASE REPORTS BY TIME SINCE ENTRY TO THE UK TO TUBERCULOSIS DIAGNOSIS, UK, 2009 600 500 Number of cases 400 300 200 100 0 0 5 10 15 20 25 30 35 40 45 50+ Years since entry to diagnosis Sources: Enhanced Tuberculosis Surveillance (ETS), Enhanced Surveillance of Mycobacterial Infections (ESMI) THEORETICAL BENEFITS OF IGRA OVER TST FOR SCREENING TST IGRA Immunological specificity Detection ti of cell-mediated d Detection ti of cell-mediated d immunoreactivity to antigens immunoreactivity to antigens Immunoreactivity to Mtb complex Immunoreactivity to Mtb complex Immunoreactivity to NTM and BCG Long term immunoreactivity Booster effect with repeated testing Requires two visits Requires expert (subjective) interpretation of induration Cheap Implementation Immunoreactivity to NTM and no BCG cross-reactivity? Shorter term immunoreactivity No booster effect Single visit Objective outcome on continuous scale effective quantitation Comparatively expensive
IMMUNOPATHOGENESIS OF LTBI Adapted from Walzl et al 2011 PATHOGEN FACTORS -Virulence Active disease Maintained latency / Clearance HOST FACTORS -Immunocompetence - Susceptibility SPECTRUM OF MTB INFECTION GOALS: Characterise heterogeneity of the immune response to different phenotypes of LTBI Characterise the host-pathogen equilibrium in LTBI Potential phenotypes of LTBI Increasing disequilibrium Barry III et al. Nat Rev 2009
THE LITERATURE: TB PROGRESSION RISK AFTER IGRA TESTING Two published meta-analyses Diel et al. Eur Respir J 2011;37: 88-99 IFNγ release assays for the diagnosis of latent Mtb infection: a systematic review and meta-analysis Rangaka et al. Lancet ID 2012 Jan12(1):45-55 Predictive value of IFNγ release assays for incident active tuberculosis: a systematic review and meta-analysis OBJECTIVES NPV of IGRA for LTBI in low prevalence setting NPV and PPV of IGRA in active TB (gold standard of Mtb infection) NPV and PPV of IGRA for longitudinal TB risk in high risk populations Longitudinal TB risk with IGRAs in high risk populations Comparison of risk with TST Evaluate efficacy of chemoprevention Subgroup analyses for further IGRA based risk stratification SIGNIFICANCE Valuable synopsis and review of the existing literature Identify important limitations in study design that can help inform future practice Illustrate heterogeneity of the literature base THE LITERATURE: TB PROGRESSION RISK AFTER IGRA TESTING Diel et al. Eur Respir J 2011;37: 88-99 IFNγ release assays for the diagnosis of latent Mtb infection: a systematic review and meta-analysis Rangaka et al. Lancet ID 2012 Jan12(1):45-55 Predictive value of IFNγ release assays for incident active tuberculosis: a systematic review and meta-analysis When test performance was not restricted to prior positivity of another test, progression rates to tuberculosis among IGRA-positive individuals followed for 19 24 months varied 8 15%, exceeding those reported for the TST (2 3%). RESULTS Compared with test-negative results, IGRA-positive and TST-positive results were much the same with regard to the risk of tuberculosis pooled IRR 2 11 [95% CI 1 29 3 46] for IGRA vs 1 60 [0 94 2 72] for TST CONCLUSIONS The current evidence is, however, still limited in determining whether the IGRAs have a stronger predictive value for developing active disease later in life, as compared to the TST. Neither IGRAs nor the TST have high accuracy for the prediction of active tuberculosis, although use of IGRAs in some populations might reduce the number of people considered for preventive treatment.
COMPARING THE STUDIES Diel Rangaka Comments IGRAs Commercial only Commercial and noncommercial Manufacturer Prolonged incubation instructions only allowed Study design TB diagnosis 5 of 8 studies included by Rangaka used in-house assays 2 studies included by Rangaka used in house assays with prolonged incubation instructions only allowed used in-house assays with Retrospective and prospective Stratification based on diagnostic certainty Retrospective and prospective Stratification based on culture confirmation Additional exclusion criteria Incorporation and work-up bias Outcomes assessed TB risk presented as NPV Longitudinal TB risk Vs Difference between studies in and PPV TST, presented as IRR the way TB risk and therefore IGRA performance is presented Studies included 4 8 No overlap between papers Follow up period 19 to 24 months Median 4 years Conclusion IGRA better than TST with a PPV of 8-15% (4 to 5 fold higher) IGRA and TST equivalent IRR 2.11 vs 1.6 (1.3 fold difference) Only IGRA as a test shows a significant difference in IRR for positive and negative results COMPARING THE STUDIES Diel Rangaka Comments IGRAs Commercial only Commercial and noncommercial Manufacturer Prolonged incubation instructions only allowed Study design TB diagnosis 5 of 8 studies included by Rangaka used in-house assays 2 studies included by Rangaka used in house assays with prolonged incubation instructions only allowed used in-house assays with Retrospective and prospective Stratification based on diagnostic certainty Retrospective and prospective Stratification based on culture confirmation Additional exclusion criteria Incorporation and work-up bias Outcomes assessed TB risk presented as NPV Longitudinal TB risk Vs Difference between studies in and PPV TST, presented as IRR the way TB risk and therefore IGRA performance is presented Studies included 4 8 No overlap between papers Follow up period 19 to 24 months Median 4 years Conclusion IGRA better than TST with a PPV of 8-15% (4 to 5 fold higher) IGRA and TST equivalent IRR 2.11 vs 1.6 (1.3 fold difference) Only IGRA as a test shows a significant difference in IRR for positive and negative results
Studies included by Rangaka First author Country Setting Income WHO Prevalence Population High risk group High risk progression Doherty Ethiopia Low 359 Contacts (Sm+) Yes Hill Gambia Low 269 Contacts (Sm+) Yes Bakir Turkey Middle 29 Contacts (Sm+) Yes Del-Corral Colombia Middle 35 Contacts Yes Lienhardt Senegal Low 282 Contacts (Sm+/C+) Yes Jonnalagadda Kenya Low 305 HIV Yes Joshi India Middle 168 HCW Uncertain Costa Portugal High 30 HCW Uncertain Aichelburg Austria High 11 HIV Yes Studies included Kik Netherlands High 8 Contacts (Sm+) Yes by Diel Diel Germany High 5 Contacts t (Sm+) Yes Clark UK High 14 HIV Yes Yoshiyama Japan High 21 Contacts (95% Sm+) Yes Studies excluded Leung China Middle 96 Silicosis Yes Harstad Norway High 6 Immigrants Uncertain Mahomed S. Africa Middle 971 High prevalence area Uncertain Diel N=4 Rangaka N=8 Not included N=4 Country Income High 4 1 2 Middle 0 3 2 Low 0 4 0 TB Prevalence (/ 100 000) 7.8 155 590 Study Group High risk Recent contacts 2 5 1 group HIV 2 1 0 HCW 0 2 0 Residence in high prevalence area 0 0 2 Other 0 0 1 Progression Very high NS (Low) 3.10% NS risk High 4 6 2 Intermediate / unknown 0 2 2 Age Mean 33.7 23.7 19 Children only 0 1 1 Adult only 3 5 3 All (mixed) 1 2 Total population 2882 12540 10619 % Population followed up 77 95 83 Duration of follow-up 19-24 months 4 years (2-6)
THE LITERATURE: TB PROGRESSION RISK AFTER IGRA TESTING Two published meta-analyses Diel et al. Eur Respir J 2011;37: 88-99 Rangaka et al. Lancet ID 2012 Jan12(1):45-55 IFNγ release assays for the diagnosis of latent Mtb infection: a systematic review and meta-analysis Predictive value of IFNγ release assays for incident active tuberculosis: a systematic review and meta-analysis Low prevalence setting High income countries High prevalence setting Low / middle income countries COMPARING THE STUDIES Diel Rangaka Comments IGRAs Commercial only Commercial and noncommercial Manufacturer Prolonged incubation instructions only allowed Study design TB diagnosis 5 of 8 studies included by Rangaka used in-house assays 2 studies included by Rangaka used in house assays with prolonged incubation instructions only allowed used in-house assays with Retrospective and prospective Stratification based on diagnostic certainty Retrospective and prospective Stratification based on culture confirmation Additional exclusion criteria Incorporation and work-up bias Outcomes assessed TB risk presented as NPV Longitudinal TB risk Vs Difference between studies in and PPV TST, presented as IRR the way TB risk and therefore IGRA performance is presented Studies included 4 8 No overlap between papers Follow up period 19 to 24 months Median 4 years Conclusion IGRA better than TST with a PPV of 8-15% (4 to 5 fold higher) IGRA and TST equivalent IRR 2.11 vs 1.6 (1.3 fold difference) Only IGRA as a test shows a significant difference in IRR for positive and negative results
PRESENTING OUTCOMES Measure Sensitivity Specificity PPV NPV Incidence rate ratio (IRR) Likelihood ratio (LR) Diagnostic odds ratio (DOR) Definition Efficacy of test to identify progressors Efficacy of test to identify non-progressors Proportion of positive tests associated with progression Proportion of negative tests not associated with progression Progression rate if IGRA positive/progression rate if IGRA negative Change in odds of progression given positive (LR+) / negative result (LR-) LR+/LR- Population 'independent' Yes Yes No No No Yes Yes Comments Important variables that inform IGRA performance. Clinically meaningful after deriving likelihood ratios (LR) Measure of TB risk given positive result. Measure of TB risk given negative result. Easy to interpret for use in clinical practice but limited by context dependent applicability of research findings. Derived from PPV and NPV {PPV/(1-NPV)}. A useful summary statistic that describes discriminatory performance of test. Derived from sensitivity and specificity and therefore largely population independent. A useful statistic for clinical interpretation of test result Similar to IRR without dependence on population context. PRESENTING OUTCOMES Measure Sensitivity Specificity PPV NPV Incidence rate ratio (IRR) Likelihood ratio (LR) Diagnostic odds ratio (DOR) Definition Efficacy of test to identify progressors Efficacy of test to identify non-progressors Proportion of positive tests associated with progression Proportion of negative tests not associated with progression Progression rate if IGRA positive/progression rate if IGRA negative Change in odds of progression given positive (LR+) / negative result (LR-) LR+/LR- Population 'independent' Yes Yes No No No Yes Yes Comments Important variables that inform IGRA performance. Clinically meaningful after deriving likelihood ratios (LR) Measure of TB risk given positive result. Measure of TB risk given negative result. Easy to interpret for use in clinical practice but limited by context dependent applicability of research findings. Derived from PPV and NPV {PPV/(1-NPV)}. A useful summary statistic that describes discriminatory performance of test. Derived from sensitivity and specificity and therefore largely population independent. A useful statistic for clinical interpretation of test result Similar to IRR without dependence on population context.
CUMULATIVE INCIDENCE RISK RATIO FOR IGRAS Rangaka et al: Difference in IRR PROSPECTIVE TB INCIDENCE BY IGRA Rangaka et al: No significant difference between assays for IGRA result to discriminate TB risk
INCIDENCE RATE RATIO TST VS IGRA Rangaka et al: No significant difference between IGRA and TST (95% CIs overlap) But only IGRA IRR is significant PROSPECTIVE TB INCIDENCE IN IGRA +VE Rangaka et al: * * * * * * Recent Contacts High risk comorbidities High risk LTBI Results restricted to papers for which incidence / 1000 PYs could be evaluated Heterogeneity in TB progression rates across high risk groups - Importance of context highlights limitations of IGRAs - Are meta-analyses across risk groups meaningful?
IMPACT OF PREVALENCE ON LONGITUDINAL OUTCOME: RECENT VS REMOTE INFECTION Average annual risk of infection = 1 - (1- Prev t ) 1/t tion Probability of infec High prevalence Low prevalence Ratio P (Recent infection) : P (Remote infection) falls with: Age Prevalence A 1 A 2 Age/ Time (t) If test of LTBI discriminates poorly between recent and remote infection then risk of progression falls in a high prevalence setting ( PPV) FACTORS ASSOCIATED WITH A POSITIVE IGRA RESULT * * Diel et al. AJRCCM 2008
FACTORS ASSOCIATED WITH PROGRESSION RISK Diel et al 2010 FACTORS ASSOCIATED WITH A POSITIVE IGRA RESULT Kik et al. Int J Tuberc Lung Dis 2009
IGRA RESULTS IN RECENT CONTACTS (LEICESTER) Non-pulmonary index Smear +ve index 100 90 80 70 60 50 40 30 20 10 0 Indeterminate Negative Positive 27.2 7.6 2.2 0.0 <16 16-35 <16 16-35 80 70 60 50 40 30 20 10 0 Negative Positive 62.5 36.8 40.0 28.6 <16 16-35 <16 16-35 UK born Foreign born UK born Foreign born Prevalence of LTBI rises significantly with age in FB but not UK born Recent infection is detectable as a significant increase in LTBI prevalence for UK born contacts of both age groups Recent infection is not readily identified for older foreign born contacts FACTORS AFFECTING IGRA RESULT (LEICESTER) Variable Adjusted O.R. (95% C.I.) Wald Statistic Sig. Contact age Index disease Non-pulmonary (vs Smear positive) Index disease Smear neg pulmonary (vs Smear positive) UK Born (vs Foreign born) 1.036.(1.015 1.059) 0.180 (0.110 0.293) 0.232 (0.143 0.379) 0.517 (0.339 0.790) 11.2.001 47.5 <0.0001 34.3 9.3.002 Factors not significant: Ethnicity BCG vaccination status Household contact Partner
TB RISK AND DURATION OF FOLLOW UP Ratio P (Recent infection) : P (Remote infection) falls with duration of follow up del Corral et al 2009 High prevalence increases the probability of acquiring new infection during longitudinal follow up ( NPV) Correlation between IFNγ response and TB risk may be most predictive for early secondary disease Risk is not linear. Correcting data for mixed periods of follow up may be misleading EFFECT OF TB PREVALENCE ON INCIDENCE RISK RATIO (IRR) High prevalence setting PPV (1- NPV) IRR Likel to be an important factor in the IRR differences obser ed Likely to be an important factor in the IRR differences observed between included and excluded studies of the Rangaka metaanalysis
IN WHICH POPULATION IS IGRA LIKELY TO BE MOST INFORMATIVE OF PROGRESSION RISK? Recent contacts Young cohort In a low prevalence setting * * COMPARING IGRA RELATED PROGRESSION RISK IN A LOW PREVALENCE SETTING Study Objective High risk group Follow up period (months) Diel et al. AJRCCM 2010 (Germany) PPV of QFT for contact screening Recent contact of smear pos ptb Kik et al. Eur Resir J 2010 (Netherlands) Progression risk with mixed recent and remote infection Recent contact of smear pos ptb 42.9 22 PPV (%) 12.9 28 2.8 IGRA neg contact cases 0 3 NPV (%) 100 98 Sensitivity 100 63 Specificity 85 46
Diel (Germany) Kik (Netherlands) Demographics Mean age 29.0 37.2 Age groups <16 11.1 0 >35 34.7 60.8 % Immigrant proportion 40.6 100 Screening details Contact:case ratio 1417/101 (14.1) 3086/380 (8.1) % household contacts 35.8 33.9 Protocol QFT in all QFT / T.SPOT if TST + Timing of IGRA >8 weeks after last exposure Variable (Median 37 days) IGRA pos% 22.22 54 Sample representative % total screened 954/1417 (67.3%) 327/3086 (10.6%) % meeting inclusion criteria 954/1335 (71.5%) 327/570 (57.4%) Contact TB cases identified 19 23 Contact TB cases excluded 0 14 + 1 IMPACT OF DUAL STEP TESTING When test performance was not restricted to prior positivity of another test, progression rates to tuberculosis among IGRA-positive individuals followed for 19 24 months varied 8 15%, exceeding those reported for the TST (2 3%).? LTBI for screening LOW RISK POPULATION - TST + IGRA - IGRA + IGRA - IGRA + Rationale for dual testing: To filter out a proportion of subgroup without LTBI and very low risk of TB IGRA positive proportion in tested population IRR for TST-/TST+ is < 1 IRR for IGRA +ve (TST-/TST+) < 1 Increased PPV IRR for IGRA ve (TST-/TST+) <1 Reduced NPV
IMPACT OF DUAL STEP TESTING Diel et al 2010 17 2 Single-step IGRA: PPV = 19/198 = 9.6% Dual-step (TST 5mm): PPV = 17/191 = 8.9% Kik et al 2010: No cases recorded in TST ve contacts Dual step testing would have increased the PPV in this study LTBI PHENOTYPES AND PROGRESSION RISK BASED ON DUAL TESTING Assay Type Phenotype N (% of total) % IGRA+ in TST subgroups Incidence/1000 persons years The Gambia In-House IGRA-/TST- 183 (17.98) 4 0 (0 8 7 2) n=1018 IGRA+/TST- 177 (17.39) 29.2626 12 4 (0 3 24 5) IGRA+/TST+ 428 (42.04) 70.74 8 86 (2 4 15 4) IGRA-/TST+ 230 (22.59) 9 6 (0 2 19) Turkey In-House IGRA-/TST- 314 (34.58) 5 1 (0 6 18 4) n=908 IGRA+/TST- 44 (4.85) 11.55 11 7 (0 3 65 1) IGRA+/TST+ 337 (37.11) 88.45 22 2 (10 6 40 8) IGRA-/TST+ 213 (23.46) 7 4 (0 9 26 6) Senegal In-House IGRA-/TST- 187 (20.94) 9 85 (3 79 26 25) n=893 IGRA+/TST- 77 (8.62) 15.01 5 9 (0 83 41 87) IGRA+/TST+ 436 (48.82) 84.99 14 74 (8 73 24 89) IGRA-/TST+ 193 (21.61) 4 93 (1 23 19 7) India Commercial IGRA-/TST- 336 (49.85) 3.5 (1.7-7.4) n=674 IGRA+/TST- 58 (8.61) 21.09 2 89 (0 4 20 5) IGRA+/TST+ 217 (32.20) 78.91 3 9 (1 6 9 4) IGRA-/TST+ 63 (9.35) 2 67 (0 4 19 0) Adapted from Rangaka et al IGRA+ IRR (TST-/TST+) IGRA- IRR (TST-/TST+) 1.4 0.42 0.53 0.69 1.2 2.0 0.74 1.3
IMPORTANT DIFFERENCES BETWEEN STUDIES Diel (Germany) Kik (Netherlands) Demographics Mean age 29.0 37.2 Age groups <16 (%) 11.1 0 >35 (%) 34.7 60.8 % Immigrant proportion 40.6 100 Screening details Contact:case ratio 1417/101 (14.1) 3086/380 (8.1) % household contacts 35.8 33.9 Protocol QFT in all QFT / T.SPOT if TST + Timing of IGRA >8 weeks after last exposure Variable (Median 37 days) IGRA pos% 22.2 54 Sample representative % total screened 954/1417 (67.3%) 327/3086 (10.6%) % meeting inclusion criteria 954/1335 (71.5%) 327/570 (57.4%) Contact TB cases identified 19 23 Contact TB cases excluded 0 14 + 1 HETEROGENEITY IN LOW PREVALENCE SETTINGS Screening of high risk populations as a TB prevention strategy is most realistic in high income countries with low prevalence The utility of IGRA as a significant predictor of TB progression is likely l to be greatest t in a low prevalence setting BUT A paucity of available data exists that is discordant. In the UK ~75% of TB cases occur in immigrants, most often from high prevalence countries There is considerable variation in TB prevalence at a regional and sub-regional level There is little data in the literature on predictive capacity of IGRAs in non-smear positive pulmonary TB.
HETEROGENEITY OF TB PREVALENCE: LEICESTERSHIRE AND LEICESTER CITY Catchment population ~ 1 million Overall incidence for catchment population = 28 / 100,000 COUNTY 7 / 100,000 CITY 72 / 100,000 LEICESTER CITY ~300,000 Population 30% Foreign Born (UK average 11%) 25% Indian Origin First UK plural city 2019
HETEROGENEITY WITHIN LEICESTER CITY Top 10 GP practices (7% of practices) account for 45% of TB cases Estimated incidence 150 / 100,000 Progression risk in IGRA studies of LTBI in low prevalence settings are susceptible to the influence of heterogeneity of TB prevalence PROFILE OF CONTACTS FOR SCREENING VARIES WITH ORIGIN OF INDEX CASE P<0.001 76.4% 39.7%
UNDERSTANDING LOCAL EPIDEMIOLOGY 100 80 Contact FB Contact UK Born 60 40 71% 20 0 36% Index UK Index FB Index UK Index FB Index UK Born Born Born 8% Index FB Contact age <16 yrs Contact age 16-35 yrs Contact age >35 yrs IGRA based contact screening is likely to have greater predictive value in younger people because: Age association with probability of recent infection Contact more likely to be UK born IMPROVING RISK CHARACTERISATION WITH IGRAS Recent contact represents the purest model for study - Biomarkers developed in the contact model may be validated in more heterogeneous populations Serial testing - To identify recent infection - Better characterisation of the dynamics of serial testing needed More quantitation thresholds of risk - Threshold for identification of LTBI < threshold for significant progression risk - Trade off between PPV and NPV Progression risk with IGRAs not considered in isolation: - Bayesian approach to IGRA use - Inclusion of a more complete panel of variables for modelling progression risk
CONCLUSIONS There is no single answer to the question of predictive value with IGRA testing Heterogeneity precludes accurate interpretation of outcomes in meta-analyses - Heterogeneity of TB prevalence - Heterogeneity of at-risk groups - Heterogeneity of study design (assays, follow up) - Heterogeneity of data presentation There is a need for consensus in study designs to address the question of progression rates with IGRAs IGRA testing is likely to be most predictive in low prevalence settings due to poor discrimination of LTBI heterogeneity