Vaccine Safety Research Methods

1
(No Transcript)
2
Vaccine Safety Research Methods

• Hypothesis generating
• Vs.
• Hypothesis testing

3
Vaccine Safety Research Methods

• Hypothesis generating
• Passive surveillance systems
• Vaccine Adverse Event Reporting System (VAERS)
• Active surveillance systems
• Rapid Cycle Analysis (RCA)
• Post-marketing screening studies
• Safety of the Pediatric Trivalent Inactivated
  Influenza Vaccination (TIV) (I II)

4
Safety of Pediatric TIV (I)

• Objective
• To screen a large cohort of children ages 6
  months to 18 years who received the TIV for
  evidence of medically attended events (MAE)
  following vaccination

5
Safety of Pediatric TIV (I)

• Vaccine Safety Datalink (VSD)
• 5 Managed Care Organizations (MCOs) Kaiser
  Permanente (KP) Northern California, KP Southern
  California, Group Health Cooperative, KP
  Northwest, and KP Colorado
• gt3,500,000 children under 18 years of age

6
Safety of Pediatric TIV (I)

• Data ranging 1/1/93 12/31/99
• Outpatient/ED visits, 1/1/95 12/31/99
• Inpatient visits, 1/1/93 12/31/99
• Children who received at least one TIV shot were
  included in the cohort
• 251,600 children who received 438,167 TIV
  vaccinations

7
Safety of Pediatric TIV (I)

• 1,165 diagnosis codes occurred during the 2 weeks
  following flu vaccination
• Only analyzed those who experienced an event
  during the follow-up period

8
Risk and Control Periods
Risk Period Days 1-14
Flu Vaccination Day 0
Control Group 2 Days 15 - 28
Control Group 1 Days -28 -15
9
Inpatient
438,167 Shots
Sample 2 219,083 Shots
Sample 1 219,084 Shots
Remove codes not biologically plausible
P lt .05 compared to EITHER control group
Confirm against Sample 2
10
Final Steps

• ORs gt 1 and p lt .05 in both samples
• Medical chart review on the cases to confirm case
  status
• Reanalyze data with confirmed incident cases

11
Conclusions

• No signals of any serious events noted
• Confirms that a signal has not been missed by
  VAERS

12
Safety of Pediatric TIV (II)

• Same objective as the first study, but the
  analysis focused on children 6 to 23 months
• 8 MCOs
• Data ranging from 1/1/1991 5/31/2003
• Used similar self-control methodology as the
  first study

13
Safety of Pediatric TIV (II)

• Did not randomly split the cohort
• Required a significant association (P lt 0.05)
  against both control periods before and after
  vaccination
• Conducted an additional self-controlled case
  series (SCCS) analysis on biologically plausible
  significant associations, controlling for age and
  season

14
Safety of Pediatric TIV (II)

• Screened more than 45,356 children, representing
  69,359 TIV vaccinations
• Analyzed more than 1200 ICD-9 codes and 20,000
  MAEs
• Examined MAEs in post-vaccination risk windows of
  3, 14, and 42 days

15
Conclusions

• Similar conclusions to the first safety of TIV
  study
• Possible association with fever and pain crisis
  in 14 days following vaccination in children with
  sickle cell anemia
• Case-control study in VSD underway

16
Strengths of Screening Studies

• Relatively fast turnaround
• Analyze multiple outcomes
• Denominator data in contrast to VAERS
• Control for potential confounders
• Can follow-up positive signals with chart review
• Bridges the gap between passive surveillance and
  formal hypothesis testing studies

17
Limitations of Screening Studies

• Multiple comparisons type 1 error
• Many outcomes not biologically plausible
• Difficult to control for age and seasonal
  effects
• Biases inherent to administrative data

18
Summary

• Good method for screening a large cohort in a
  relatively short amount of time
• Requires chart review for positive signals
• May have a role in post-licensure safety
  screening of new vaccines

19
Acknowledgments

• Vaccine Safety Datalink
• Co-authors on TIV studies
• Simon J. Hambidge, David McClure, Eric K France,
  Stanley Xu, Kristi Yamasaki, Lisa Jackson, John
  P. Mullooly, Kenneth M. Zangwill, S. Michael
  Marcy, Steven B. Black, Edwin M. Lewis, MPH,
  Henry R. Shinefield, Edward Belongia, James
  Nordin, Robert T. Chen, David K. Shay, Robert L.
  Davis, and Frank DeStefano.

20
References

• France EK, Glanz JM, Xu S, et al. Safety of the
  trivalent inactivated vaccine among children A
  population-based study. Arch Pediatr Adolesc Med
  2004 Nov158(11)1031-1036.
• Hambidge SJ, Glanz JM, McClure D, et al. Safety
  of the trivalent influenza vaccine in infants and
  young children ages 6-23 months A
  population-based study. JAMA 2006 Oct
  25296(16)1990-7.
• Glanz JM, Mcclure DL, Xu S, Hambidge SJ, Lee M,
  Kolczak MS, et al. Four different study designs
  to evaluate vaccine safety were equally valid
  with contrasting limitations. J Clin Epidemiol
  2006 59808-18.

21
Negative Odds Ratios

• In both studies (I II), there were several
  significant ORs between 0.40 and 0.90 for common
  acute illnesses
• URI
• Cough
• Otitis Media
• Pneumonia
• Sinusitis
• Bronchitis

22
Why The Negative Odds Ratios?

• Healthy Vaccinee Effect
• Expected Symptoms Effect
• Sick of the Clinic Effect
• Vaccinated When Sick Effect

23
Risk and Control Periods
Risk Period Days 1-14
Flu Vaccination Day 0
Control Group 2 Days 15 - 28
Control Group 1 Days -28 -15
24
Why The Negative Odds Ratios?

• Healthy Vaccinee Effect
• Expected Symptoms Effect
• Sick of the Clinic Effect
• Vaccinated When Sick Effect

25
Why The Negative Odds Ratios?

• Healthy Vaccinee Effect
• Expected Symptoms Effect
• Sick of the Clinic Effect
• Vaccinated When Sick Effect

26
Why The Negative Odds Ratios?

• Healthy Vaccinee Effect
• Expected Symptoms Effect
• Sick of the Clinic Effect
• Vaccinated When Sick Effect