Vaccine Safety: the Importance of Going Beyond Passive Surveillance

Optimal Vaccine Pharmacovigilance: Why it is important and how to do it Steven Black, MD Professor of Pediatrics Division of Infectious Diseases Cincinnati Children’s Hospital Cincinnati, Ohio USA Presentation Overview • Passive Reporting Systems: VAERS as an example • Broad post licensure evaluations of safety: Phase Four studies, VSD Datalink, and rapid cycle studies • Background rates of events versus vaccine related events • Passive versus Active Surveillance • Global collaborative efforts Passive Reporting: VAERS as an example • VAERS (Vaccine Adverse Event Reporting System) relies on passive reporting by physicians or patients. – No real denominator data. – Only cases after vaccine are reported • Statistical techniques have been developed for signal detection using “data mining”). – Relies on comparing expected versus observed numbers of events following vaccine using Bayesian statistics. – In a retrospective analysis evaluating this technique, intussusception following Rotashield vaccine identified early. • EBGM ( Empirical Bayesian Geometric Mean) score elevated for – Intussusception – Diarrhea, GI hemorrhage Passive Reporting: Intussusception Cases Reported to VAERS after Rotashield 45 40 MMWR 35 30 25 20 Cases Cases 15 10 5 0 gen- feb- mar- apr- mag- giu- lug- ago- set- ott- nov- 99 99 99 99 99 99 99 99 99 99 99 4 Empirical Bayesian Geometric Mean Score following Rotashield 20 18 16 14 12 10 EBGM for Intussusception 8 6 4 2 0 Jan Feb Mar April May June July 1 5 Passive Reporting Summary: Intussusception and VAERS • VAERS was excellent at signal detection for intussusception. • Additional Statistical techniques can lead to earlier detection of vaccine safety signals. • Passive reporting systems, however: – Are prone to bias in that events closest to vaccine are reported preferentially – Underreporting is a major issue. – Misclassification of events can occur. – Information on cases may be limited. – Recognized adverse events are more likely to be reported. – Do not allow assessment of relative or attributable risk. Active Post Marketing Surveillance Phase Four studies by manufacturers Datalink Studies Rapid cycle analyses Global Collaborations 7 Vaccine Data Link studies vs Passive Reporting Systems. • Unlike passive reporting systems, data link studies have the potential to – Identify cases of events in an unbiased way. – Allows analysis of unexposed cases – Allow access to medical records to better characterize and understand cases. – Calculate rates of events, relative risk and attributable risk. – Evaluate vaccine impact and changes in disease epidemiology What Data do you Need? Computerized Vaccine Data on an individual level Exposure Need a unique identifier to link Outcome or Possible “Adverse Event” Computerized Hospital and/or Clinic Diagnoses NOTE: Can do case series with On an individual level Outcome Data alone to assess risk and hence causality Demographic Data on a Population Adjust for confounders Evaluation of Uncommon Events Resulting in Medical Utilization: Case Series • Allows calculation of relative risk using only cases of the outcome. • Probability of being vaccinated within a specified time window prior to vaccination is compared to probability of being vaccinated at other times. If there is no association, then these two probabilities should be equal yielding an odds ratio of one. • For self control comparison, exposed and unexposed follow up time are defined for the same individual. • Resolves issue of differences between groups • Issue of “well child effect” bias for pre-exposure time interval • Does not work if event prevents vaccination vax vax Pre Exposed Post 1 Exposed 2 Post /Pre 2 Developed by Farrington Rapid Cycle Techniques • Requires rapid access to automated data sources to establish rates of events in baseline and following a new vaccine. Currently in use in VSD with weekly data pulls. • Uses sequential probability testing to test the likelihood of an observed number of cases versus expected. • A “stopping value” is established to either rule in or rule out an association based upon the expected number of cases and risk level. • For intussusception, a risk established after 2589 doses of vaccine given… a similar time frame as VAERS data mining. • Also able to detect a decrease in risk of seizures, fever and other abnormal neurologic events within 12 weeks of introduction of DTaP as compared to DTwP. Rapid Cycle Techniques Sequential Probability Testing: Two Patterns Intussusception and Rotashield Facial Paralysis and Men ACWY 12 What do these analyses look like? An example Prelicensure Study of MMR-V vs Post Licensure Evaluation in VSD A Case Study: MMR-V Pre-licensure vs Post Licensure Overview • Pre-licensure studies of safety are usually small – Focus on common local and systemic events – Analyses done within predefined windows • Post licensure studies are usually much larger and have ability to look at events more flexibly MMR-V Pre-licensure Safety Black et al. PIDJ 24:8-12, 2005 MMRV MMR & V P-value N=323 N=157 Fever 0-42 39.6% 34.8% ns days s/p vax Fever 5-12 27.7% 18.7% 0.034 days s/p vax Seizures 1 1 on day 1 ns on day 9 15 VSD Rapid Cycle Evaluation of MMRV Outpatient Visits for Fever by Day after Vaccine at Northern California Kaiser Permanente: 1995-2008 Age 12-23 months 6241 total fever visits after 302,670 MMR+V, 147,762 MMR, 46,390 MMRV, 38,251 VZV 350 MMR 300 MMR+V MMRV 250 V 200 150 100 Events Events / Doses 100,000 50 0 0 5 10 15 20 25 30 35 40 Days after Immunization VSD Rapid Cycle Evaluation of MMRV Temporal distribution of seizures after MMRV vaccination vs after simultaneous MMR and varicella vaccination • 25 60 50 20 40 15 30 10 20 5 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 Days Post-MMRV Vaccine Days Post-MMR + Varicella Vaccine (2/06-9/07) (1/00-8/07) Nicola Klein as presented at PAS/SPR in Honolulu May 2008 VSD Rapid Cycle Evaluation of MMRV Evaluating for the risk of seizure 7-10 days post- vaccination using chart verified febrile seizures 95% Confidence P-value Odds ratio Interval MMRV versus MMR + 2.3 1.6, 3.2 <0.0001 Varicella* Attributable 95% Confidence Risk Interval MMRV versus MMR + 5.2 / 10,000 2.2, 8.1 Varicella* N= 42 for MMR-V out of 43,353 MMR-V and 124 for 314,599 doses of MMR + V for chart review confirmed seizures. Adjusted for age and influenza season. Increased risk with MMRV cannot be explained by concomitant vaccines, temporal trends in seizure, VSD site, age or influenza season. 18 A Comparison of Active versus Passive Surveillance in Burkino Faso • Men A vaccine was introduced into the Sub- Saharan African meningitis belt in December 2010 and almost 12 million people vaccinated in one campaign in Burkino Faso • Two safety surveillance systems established – Nation wide passive surveillance – Active Surveillance in one district VACCINE in press MenA Introduction in Burkino Faso: Passive Surveillance Data Per 100,000 MenA Introduction in Burkino Faso: Active Surveillance Data Per 100,000 Per 100,000 21 MenA Introduction in Burkino Faso: Active Surveillance Data Summary: Active vs Passive Surveillance • Passive surveillance can be useful for signal detection • Under-reporting can severely limit the usefullness of passive reporting • Active surveillance in a defined sub-population can be more useful than passive surveillance in a larger population. • Both types of surveillance are feasible in developed and developing countries. 23 Summary of Advantages of Population Based Database Studies • Allow calculation of incidence on AEFI and background rates of disease without vaccination • Allow calculation of relative risk • Allow calculation of attributable risk. • Allow adjustment for confounders • Allow assessment of trends including vaccine impact on disease for risk-benefit analyses What to do if you observe a possible signal? • Evaluation of a possible consistent time association of the event with vaccination • Evaluation in a different analytic framework: self- control analysis or other reference group • Possible associations can serve as a source of hypothesis generation for further studies – ie case- control study conducted for intussusception. The first step in looking at a possible signal: What happens without vaccines? The importance of knowing background epidemiology 26 What happens without vaccines? “AEFI” without Vaccines: Autoimmune < 30days Outpatient Events in Teens # events Rate/100,000 Rate Outpatient care Adolescents py Teens Adults Thyroid (…) disorders 859 396 1412.05 Ulcerative 556.x colitis 76 35.4 117.52 Regional 555.x enteritis 68 31.6 97.18 Systemic lupus 7100 erythematosus 63 52.9 120.23 Rheumatoid 7140 arthritis 29 13.5 119.33 37730 Optic neuritis 10 4.7 13.56 Multiple 340 sclerosis 9 4.2 64.18 71659 Polyarthritis 7 3.3 30.74 Claire Anne Seigrist PIDJ Pandemic Flu Safety The importance of background rates of disease in assessment of vaccine safety during mass immunization with H1N1 influenza vaccines Where to from here? • Increasing the sample size and geographic reach – Regional and global consortiums • Maximizes statistical power • Increases local expertise – EXAMPLES: • VAESCO • SOMNIA • WHO GVSI Pilot SOMNIA STUDY: EVALUATING THE RISK OF NARCOLEPSY FOLLOWING ADJUVANTED 2009 H1N1 PANDEMIC VACCINES BACKGROUND AND RATIONALE 30 Narcolepsy • CNS disorder characterized by excessive daytime sleepiness, abnormal manifestations of REM sleep o Sleep attacks, disrupted nocturnal sleep, sleep paralysis, hypnagogic hallucinations, cataplexy o Chronic disease, treated with medication and behavior modification, no cure • Two diagnostic entities o Narcolepsy with cataplexy, narcolepsy without cataplexy • Brighton Collaboration case definition exists* o Primarily based on presence of symptoms and an abnormal multiple sleep latency test (MSLT) characteristic of narcolepsy • Poli et al.

Load more