Studying Rotavirus Vaccines and Intussusception in Mini-Sentinel W. - PowerPoint PPT Presentation

Studying Rotavirus Vaccines and Intussusception in Mini-Sentinel W. Katherine Yih, PhD, MPH Harvard Pilgrim Health Care Institute and Harvard Medical School January 31, 2013 info@mini-sentinel.org 1

Rotavirus vaccines  Live, attenuated, oral vaccines for infants, multiple doses  Rotashield licensed in August 1998  In 1999, Rotashield voluntarily withdrawn due to increased risk of intussusception • Excess risk: 1-2 cases/10,000 vaccine recipients • Risk highest 3-7 days after Dose 1  RotaTeq (2006) and Rotarix (2008) licensed after clinical trials with >60,000 infants info@mini-sentinel.org 2

Post-licensure studies, RotaTeq & Rotarix Dose 1, 1-7 d after vaccination 1 st author, date Site, system 1 st doses No. of RR (95% CI) cases RotaTeq Buttery 2011 Australia 115,657 3 5.3 (1.1, 15) Haber (abstract) 2011 U.S., VAERS n.a. 66 1.5 (0.96, 2.3) Shui 2012 U.S., VSD 309,844 1 1.2 (0.03, 6.8) Rotarix Buttery 2011 Australia 163,709 3 3.5 (0.7, 10) Patel 2011 Mexico n.a. 24 5.3 (3.0, 9.3) 5.8 (2.6, 13) Velázquez 2012 Mexico n.a. 56 6.5 (95.5% CI 4.2, 10) info@mini-sentinel.org 3

Post-licensure studies, RotaTeq & Rotarix Dose 1, 1-7 d after vaccination 1 st author, date Site, system 1 st doses No. of RR (95% CI) cases RotaTeq Buttery 2011 Australia 115,657 3 5.3 (1.1, 15) Haber (abstract) 2011 U.S., VAERS n.a. 66 1.5 (0.96, 2.3) Shui 2012 U.S., VSD 309,844 1 1.2 (0.03, 6.8) Rotarix Buttery 2011 Australia 163,709 3 3.5 (0.7, 10) Patel 2011 Mexico n.a. 24 5.3 (3.0, 9.3) 5.8 (2.6, 13) Velázquez 2012 Mexico n.a. 56 6.5 (95.5% CI 4.2, 10) info@mini-sentinel.org 4

Rotavirus vaccine doses in Mini-Sentinel study (for period for which charts reviewed, through 6/2011 maximum) 1st doses All doses RotaTeq 507,874 1,277,556 Rotarix 53,638 103,098 info@mini-sentinel.org 5

Intussusception case-finding algorithm First-ever of any of these in ED or inpatient setting: – ICD-9 560.0 (intussusception) – ICD-9 543.9 (unspec. diseases of appendix, including intussusception) – CPT 74283 (therapeutic enema, contrast or air, for reduction of intussusception or other intraluminal obstruction) info@mini-sentinel.org 6

Chart review  Purposes:  To confirm intussusception diagnoses  To confirm rotavirus vaccination (specific vaccine, dose, age) of intussusception cases  Standardized chart abstraction and adjudication forms  Pediatrician adjudicators reviewed chart material to determine if cases found by algorithm truly intussusception  Adjudicators blinded to vaccination status info@mini-sentinel.org 7

Chart review metrics Algorithm-identified potential cases = 343 Those for whom chart obtained = 267 (78%) Those confirmed as intussusception = 124 (46%) Cases are from whole infant population and include unexposed info@mini-sentinel.org 8

Designs and analysis approaches  Vaccinated infants only (self-controlled risk interval)  Uses just vaccinated cases with intussusception in either pre-specified risk interval or comparison interval  Analysis by logistic regression  All infants (cohort)  Uses exposed and unexposed person-time of whole infant population  Analysis by Poisson regression info@mini-sentinel.org 9

info@mini-sentinel.org Intussusception incidence by age Rate per 100,000 infants 10 20 30 40 50 60 70 0 from Tate et al. Pediatrics 2008;121:e1125-e1132 0-6 DAYS 14-20 DAYS 28-34 DAYS 42-48 DAYS 56-62 DAYS 70-76 DAYS 84-90 DAYS 98-104 DAYS 112-118 DAYS 126-132 DAYS 140-146 DAYS 154-160 DAYS 168-174 DAYS 182-188 DAYS 196-202 DAYS 210-216 DAYS 224-230 DAYS 238-244 DAYS 252-258 DAYS 266-272 DAYS 280-286 DAYS 294-300 DAYS 308-314 DAYS 322-328 DAYS 336-342 DAYS 350-356 DAYS 10

info@mini-sentinel.org Intussusception incidence by age Rate per 100,000 infants 10 20 30 40 50 60 70 0 from Tate et al. Pediatrics 2008;121:e1125-e1132 0-6 DAYS 14-20 DAYS 28-34 DAYS 42-48 DAYS Dose 1 56-62 DAYS 70-76 DAYS 84-90 DAYS 98-104 DAYS Dose 2 112-118 DAYS 126-132 DAYS Dose 3 (for RotaTeq) 140-146 DAYS 154-160 DAYS 168-174 DAYS 182-188 DAYS 196-202 DAYS 210-216 DAYS 224-230 DAYS 238-244 DAYS 252-258 DAYS 266-272 DAYS 280-286 DAYS 294-300 DAYS 308-314 DAYS 322-328 DAYS 336-342 DAYS 350-356 DAYS 11

Self-controlled risk interval design  Each subject serves as own control; adjusts for individuals’ characteristics that don’t change vaccinated ____ ↓Ι ___ Ι _______ Ι __________ Ι _______ 0 1 7 22 42 days after vaccination primary risk window comparison window  Adjust for age-specific risk of intussusception using logistic regression with offset term info@mini-sentinel.org 12

Cohort design  Uses exposed and unexposed person-time in 1 st year of life from whole population Pt 1 __ ↓ Pt 2 ______________ Pt 3 _____RV ___ __________ ↓ Pt 4 __RV ___ ____________________ Pt 5 __________________RV _ ↓  Adjust for age-specific risk of intussusception using Poisson regression with polynomial risk function info@mini-sentinel.org 13

Complementarity of designs Design Pros Cons Self-controlled Controls well for Requires accurate fixed risk factors, age-specific e.g. race/ethnicity incidence for age adjustment Cohort Higher statistical Could be affected power; extrinsic by residual background rates confounding not needed info@mini-sentinel.org 14

www.mini-sentinel.org/work_products/Statistical_Methods/Mini-Sentinel_Methods_Framework- for-Assessment-of-Signal-Refinement-Positive-Results.pdf info@mini-sentinel.org 15

Concern To address concern 1. Data validity Examine descriptive statistics in detail 2. Systematic bias a. Misclassification i. Of exposure Review charts to confirm RV exposure (type, dose number) Use 2 risk windows, 1-7 d and 1-21 d ii. Of outcome Review charts to confirm intussusception b. Selection bias Use exposed and unexposed person-time from same people (with self-controls and with the cohort) c. Confounding Use SCRI analysis to adjust for fixed risk factors Use multivariate adjustment in regression modeling Age: Adjust for age in all analyses, using either age- specific incidence from literature or in M-S data List of concerns adapted from Mini-Sentinel Framework for Assessment of Positive Results (1st of 2 slides) info@mini-sentinel.org 16

Concern To address concern 3. Magnitude of Quantitative bias analysis influence of systematic error on risk estimates Examples: Re-do analyses including possible cases (neither confirmed nor ruled out) Re-do analyses taking into consideration cases whose charts were not obtained Additional secondary analysis: examine pattern in timing of onset after vaccination, using age-adjusted temporal scan statistics Adapted from Mini-Sentinel Framework for Assessment of Positive Results (2nd of 2 slides) info@mini-sentinel.org 17

 Final results available by fall 2013 info@mini-sentinel.org 18