Felix I. Zemel, MPH DrPH Student Tufts University School of Medicine
Influenza-like illness (ILI) has a high burden on society 7%-15% of ILI is caused by influenza Median duration of illness is 3 days Prevention efforts to reduce the burden of ILI increased during the second half of the 20 th century. Considerations for prevention Seasonal variation in influenza Lack of carryover of vaccine effectiveness across flu seasons
“To identify, retrieve and assess all studies evaluating the effects (efficacy, effectiveness and harm) of vaccines against influenza in healthy adults we defined” 1. Efficacy: “the capacity of the vaccines to prevent influenza A or B and its complications”; 2. Effectiveness: “the capacity of the vaccines to prevent ILI and its consequences”; and 3. Harm: “any harmful event potentially associated with exposure to influenza vaccines
Study type Randomized studies Randomized controlled trials (RCT) Quasi-RCT Influenza vaccine in humans with: Placebo No intervention Types, doses, or schedules of vaccination “Assess protection of participants from naturally-occurring influenza” Non-randomized studies “Reported evidence on the association between influenza vaccines and serious adverse effects”
Participants Healthy individuals Aged 16 to 65 years Irrespective of influenza immune status Interventions Inactivated parenteral vaccines Live aerosol vaccines Inactivated aerosol vaccines
Clinical Number of symptomatic influenza and ILI cases occurring Seriousness of symptoms Harms Number and seriousness of adverse effects Systemic effects Local effects
Systematic review Cochrane Central Register of Controlled Trials EMBASE 1990 – June 2010 MEDLINE (Pubmed) Jan 1966 – June 2010 Bibliography review Vaccine Journal First issue – end of 2009
Grey literature Letters to manufacturers Letters to first and corresponding authors of studies in review Data collection / extraction One author performed initial screening of retrieved citations Two other authors independently applied inclusion criteria to all identified and retrieved articles One author supervised and arbitrated data extraction
Utilized Cochrane Handbook for Systematic Reviews of Interventions Randomization Generation of the allocation sequence Allocation concealment Blinding Follow-up Non-randomized studies Presence of potential confounders Newcastle-Ottawa Scale (NOS)
Assessed at the analysis stage Based upon number of NOS items judged inadequate in each study Low = up to one inadequate item Medium = up to three inadequate items High = more than three inadequate items Very high = no description of methods
Unit of analysis Varying definitions of ‘epidemic period’ Case definition for ILI Different methods of laboratory confirmation of influenza cases Missing data not assessed Heterogeneity Reporting bias Positive effects = greater risk of bias Industry funded = greater risk of bias Industry funded = greater odds of publication in high impact factor journal
Efficacy (against influenza) and effectiveness (against ILI) Risk Ratio (RR) Main findings Risk Difference (RD) with 95% confidence intervals Absolute vaccine efficacy (VE) 1-RR whenever statistically significant
Table 1. Study characteristics of studies included for analysis of inactivated aerosol vaccine versus placebo Study DesignBlinding# of ParticipantsPopulationIntervention (n)Outcome Type Boyce, 2000 Open label RCT Single- blinded 74 Healthy adults (10-40 years) who did not receive vaccination during 6 months preceding trial 1) M-59 trivalent flu vaccine (31) 2) Unadjuvated vaccine (30) 3) Placebo (13) 48 hour exam 7 day diary Local Systemic reactions Langley, 2005 RCTnr 1 Healthy adults (18-50 years) 1) 3-non covalent vaccines (nr) 2) Placebo (nr) Local (7 day) Systemic (7 day) Systemic (48 hrs) Waldman, 1969c RCT Double- blind 597 School teachers during epidemic period (age not reported) 1) Monovalent inactivated aerosol vaccine (479) 2) Placebo (118) Clinical cases Side effects Waldman, 1969d RCT Double- blind 590 School teachers during epidemic period (age not reported) 1) 471 Polyvalent inactivated aerosol vaccine (471) 2) Placebo (119) Clinical cases Side effects Waldman, 1972a RCT Double- blind 244 Students and Staff members (age not reported) 1) Monovalent A aerosol vaccine (195) 2) Placebo (49) Clinical cases Adverse effects Waldman 1972c RCT Double- blind 243 Students and Staff members (age not reported) 1) Bivalent AB aerosol vaccine (194) 2) Placebo (49) Clinical cases Adverse effects 1 nr = not reported
Table 2. Risk-of-bias analysis among studies included for analysis of effectiveness / efficacy / harm associated with inactivated aerosol vaccine administration versus placebo by Di Pietrantonj et al (2010). Author, yearType of biasEvaluation of riskSupport for judgment Boyce, 2000Selection (allocation concealment) Unclear from description Langley, 2005Selection bias (allocation concealment) High, based upon description Inadequate data presented in description Waldman, 1969c,dSelection bias (allocation concealment) Unclear from description Waldman, 1972a,cSelection bias (allocation concealment) LowClear description in methods
Table 3. Results of meta-analysis of the effectiveness and harms associated with inactivated aerosol flu vaccine administration versus placebo and GRADE evidence profile Outcome No. of studies No. of participants Risk Ratio (95% CI) Quality 2 Inactivated aerosol vaccine versus placebo or do-nothing ILI (0.40, 0.83)●●○○ Low Local Harms (0.85, 1.40)●●●○ Moderate Systemic Harms 41, (0.77, 1.31)●●●○ Moderate 1968 to 1969 pandemic: inactivated polyvalent aerosol vaccine versus placebo ILI 21, (0.46, 0.95)●●○○ Low 1 ILI = Influenza-like illness 2 Quality of estimate, based upon GRADE criteria
Figure 1. Summary of recommendation for use of inactivated aerosol vaccine for treatment of Influenza-Like Illness, based upon GRADE criteria PopulationRecommendation Grade Healthy adults aged years Current evidence is insufficient to assess the balance of benefits and harms of the treatment. Evidence is lacking, and some existing evidence is of poor quality, or conflicting, and the balance of benefits and harms cannot be determined. I Rationale: 1.Insufficient number of studies 2.Conflicting case definitions 3.Insufficient participant descriptions