To Pool or Not to Pool: POC Testing for Hepatitis B and C Sushmita Shivkumar MSc candidate, McGill University Department of Epidemiology and Biostatistics Sushmita.shivkumar@mail.mcgill.ca 1
We have no conflicts of interest with industry. 2
Background: Hepatitis B 350 million worldwide with chronic Hepatitis B (WHO). 4 million acute cases every yy year (WHO). http://www.cdc.gov/ncphi/disss/nndss/casedef/h epatitiscacutecurrent.htm 3
Systematic Review Objective: To synthesize evidence on the diagnostic accuracy (sensitivity and specificity) of all POC tests used globally to diagnose Hepatitis B. To critique the quality of studies that have been conducted to date using the QUADAS and STARD checklists for studies of diagnostic accuracy. 4
Time period: 1985-2010 Search strategy Databases searched: Pubmed/Medline, Embase, Biosis, Web of Science. Search string : ("Hepatitis B"[Mesh] OR "Hepatitis B Antibodies"[Mesh] OR "Hepatitis B Surface Antigens"[Mesh] OR "Hepatitis B Core Antigens"[Mesh] OR "Hepatitis B Antigens"[Mesh] OR "Hepatitis B e Antigens"[Mesh]) AND ("Point-of-Care Systems"[Mesh] OR "rapid test*" OR "diagnostic") AND ("Sensitivity and Specificity"[Mesh] OR "diagnostic accuracy" OR "validity") 5
Systematic review-search tion Screening Identificat Records identified through database searching (n = 183) Records after duplicates removed (n = 145) EXCLUDED: 90 Not looking at Hepatitis B = 46 Other outcomes reported = 36 Not relevant = 8 Eligibility Full text articles assessed for eligibility (n = 55) Full text articles excluded (n =38) Not POC tests = 23 Reviews = 4 Not diagnostic accuracy studies = 2 Prevalence studies = 9 Included Studies included in qualitative synthesis (n = 17) 6
Study ID Author Year Country Sample size Index Test 1. Clement 2010 Bli Belgium 942 Amrad 2. Lau 2003 USA 2627 Amrad 3. Lien 2000 Vietnam 328 Determine, Dainascreen, Serodia 4. Nakata 1990 NR 300 Hybritech 5. Ansari 2007 Iran 240 Acon, Atlas, Intec, Blue Cross, DIMA, Cortez 6. Lin 2008 UK 1250 DRW, Determine 7. Randrianirina 2008 Madagascar 200 Determine, Virucheck, Hexagon, Cypress 8. Kaur 2000 India 2754 Hepacard 9. Akanmu 2006 Nigeria 238 Amrad 10. Oh 1999 Korea 250 Genedia (HBsAg + anti HBsAg) 11. Whang 2005 Korea 400 Genedia (HBsAg + anti HBsAg), Daewoong (HBsAg + anti HBsAg), 12. Cha 2006 Korea 80 SD (HBsAg + anti HBsAg), Genedia (HBsAg + anti HBsAg), Asan (HBsAg + anti HBsAg), 13. Palmer 1999 Honduras, DR, 298 Determine Trinidad, Jamaica 14. Davies 2010 Malawi 75 Determine, Onecheck 15. Khan 2010 Pkit Pakistan 57 Onecheck, Accurate 16. Torane 2008 India 60 Hepacard 17. Raj 2001 India 1000 Hepacard 7
Quality assessment of studies Quality QUADAS scores range from 3-10 out of 14. STARD scores range from 7-14 out of 25. Poor to moderate quality Conflict of interest Only 2 studies explicitly reported (no conflict) 8
Diagnosing Hepatitis B Double stranded DNA virus HBsAg surface of virus HBcAg encloses the virus HBeAg surrounds the core and is a marker of active replication. Antibodies to HBsAg present with onset of symptoms. http://www.hivandhepatitis.com/2008ic r/aids2008/docs/082208_e.html Long: Principles and Practice of Pediatric Infectious Diseases, 3rd ed.
Subgroups Index tests detecting: True Positives False positives False negatives HBsAg: 31 data points Determine (Abbott Laboratories): 6 data points True negatives HBsAg + HBeAg (AMRAD, Binax Inc.): 7 data points Antibody to HBsAg: 6 data points 10
Diagnosing Hepatitis B Acute Hepatitis B diagnosis Chronic Hepatitis B diagnosis HBsAg + Anti HBsAg HBcAg+ IgM Anti HBcAg + IgM anti HBcAg + AND HBsAg + OR HBeAg + OR HBV DNA + OR HBsAg + OR HBV DNA + OR HBeAg + At least 6 months apart http://www.who.int/csr/disease/hepatitis/hepatitisb_w hocdscsrlyo2002_2.pdf 11
Forest plots - Sensitivity 12
Forest plot - Specificity 13
Bayesian meta-regression Program in R: Hierarchical model Subgroup Assuming perfect reference Assuming imperfect Standard (100%) reference standard (<90 100%) Sensitivity (95% CI) HBsAg 94.76% (90.08 98.23%) Determine HBsAg 98.24% (94.74 99.98%) Specificity (95% CI) 99.54% (99.03 99.953%) 99.96% (99.31 100%) Sensitivity (95% CI) 96.77% (92.92 99.26%) 98.76% (94.67 99.99%) Specificity (95% CI) 99.89% (99.55 100%) 99.94% (99.49 100%) Amrad HBs+eAg 95.47% 99.81% 98.04% 99.95% 95% (88.88 (99.33 (93.39 (99.71 99.44%) 100%) 99.83%) 100%) Anti HBs 93.15% 93.08% 99.77% 96.08% (85.04 (81.9 (94.18 (86.38 98.5%) 99.99%) 100%) 100%) 14
Conclusion Studies use different and imperfect reference standards, different samples (oral and blood) tested. Variability in study design and reporting quality. Incomplete reporting of conflict with industry. What are the antibody tests detecting? Lack of research on secondary testing outcomes Most studies conducted in developing countries, role in developed countries was unclear 15
POC tests for Hepatitis C 16
Hepatitis C - Background WHO estimates that 170 million people worldwide are infected with the Hepatitis C virus (HCV). 70-90% of those infected go on to become chronic carriers. HCV is a Singlestranded RNA virus http://www.hepatitis.org/hepcslides/powerpoi.ppt/ 01PVANDA/sld005.htm 17
Diagnosis CDC algorithm IgM Anti Hepatitis A and Hepatitis B Anti HCV + (EIA) HCV Recombinant Immunoblot Assay + Nucleic acid testing for HCV RNA + http://www.cdc.gov/ncphi/disss/nndss/casedef/hepatit iscacutecurrent.htm 18
Objective To synthesize the available evidence on diagnostic accuracy of rapid tests for Hepatitis C To rate the quality of studies using the QUADAS and STARD checklists for diagnostic accuracy. 19
Search Identification Records identified through database searching (n =215) Screening Records after duplicates removed (n = 168) EXCLUDED based on title and abstract: 118 Not Hepatitis C = 51 Other outcomes= 50 Prevalence = 17 Eligibility Full text articles assessed for eligibility (n = 50) Full text articles excluded (n =43) Not rapid tests = 32 Reviews = 5 Not relevant = 5 Inc cluded Studies included in qualitative synthesis (n = 7) 20
Results Study ID Author Year Country Sample Size Index Test 1. Lee 2010 USA 571 Oraquick 2. Daniel 2005 India 5290 HCV Tri Dot 3. Kaur 2000 India 2754 HCV Bidot 4. Montebugnoli 1999 Italy 100 Therma Ricerca 5. Yuen 2001 China 195 SM HCV 6. Khan 2010 Pakistan 136 Onecheck, Accurate 7. Torane 2008 India 60 Goldspot 21
Quality QUADAS score ranged from 7-10 out of 14. STARD scores ranged from 8-12 out of 25. Only l one study (Khan et al. 2010) reported whether h test readers were blinded. Conflict of Interest: Lee et al. s study on Oraquick financial relationship with OraSure Technologies, Inc. 22
Results 23
Results Results Including Oraquick study (Lee et al. 2010) Assuming perfect reference standard Assuming imperfect reference standard Sensitivity (95%CI) Specificity (95% CI) Sensitivity (95%CI) Specificity (95%CI) 91.5% (89.7 93.1%) 99.6% (99.5 99.8%) 92.72% (72.11 11 99.88% (99.56 56 99.93%) 100%) Assuming perfect reference standard Results excluding Oraquick study Assuming imperfect reference standard Sensitivity (95%CI) Specificity (95% CI) Sensitivity (95%CI) Specificity (95%CI) 72.38% (36.66 98.2%) 99.88% (98.62 100%) 77.11% (45.49 99.61%) 99.99% (99.82 100%) 24
Hepatitis C - Conclusion Populations and countries No studies in co-infected populations Majority of evidence from developing countries. Accuracy: Limited evidence. Oraquick appears to be best but conflict of interest with industry. Lack of differentiation between acute and chronic cases. No research on secondary outcomes In comparison to HIV, Syphilis, Hepatitis B, HCV tests have a long way to go. 25
ACKNOWLEDGENTS This work was supported by a Knowledge Syntheses grant from the Canadian Institute for Health Research. CIHR KRS 102067 Thank you! Dr. Rosanna Peeling Dr. Jorge M Cajas Dr. Christiane Claessens Dr. Marina B Klein Dr. Madhukar Pai Dr. Gilles Lambert Dr. Lawrence Joseph ( THANK YOU) Dr Tom Wong Dr. Nitika Pant Pai 26