Is the influenza vaccine effective in decreasing infection, hospitalization, pneumonia, and mortality in healthy adults?

Abstract

The annual influenza vaccine is universally recommended by the CDC for individuals greater than six months of age. However, unlike the other 16 routinely recommended vaccines, influenza vaccination is annual and developed based on circulating strains that are constantly modifying the primary immune targets, the hemagglutinin and neuraminidase receptors. Given the variation in these critical receptors, the need to predict changes in the protein structure, and the necessity of such frequent immunizations it is reasonable to question the annual vaccinations efficacy, especially in otherwise healthy adults. This article will review literature on inactivated influenza vaccine efficacy on preventing infection, decreasing hospitalizations and pneumonia, and mortality.

Summary of the Issues:

Influenza (flu) is a droplet-transmitted virus that causes seasonal epidemics globally in addition to four pandemics since the start of the 20^th century. The Center for Disease Control (CDC) estimates that since 2010 in the United States, between 2% and 12% of the population annually develops flu infection resulting in 12,000 −56,000 deaths per year and a staggering economic toll of $87 billion.¹ This large population and economic burden has driven research into flu vaccination since the discovery of the virus in the 1930’s.² The two types of influenza virus that commonly infect humans, influenza A and B, are classified based on two key receptors - hemagglutinin (HA or H) and neuraminidase (NA or N). Their subtypes are also added to the naming system with numerical abbreviation (ie. H1N1 or H3N2). The HA and NA receptor combinations dictate which strains can infect humans or animals and are critical to viral entry and exit from cells in the respiratory tract. Also, the HA and NA receptors are important antigens in initiation of immune response to influenza and as a result have become central targets of medications and the vaccines. However, vaccine development for influenza has proved more challenging than for the 16 other infectious agents for which routine vaccinations are available. The challenge results from antigenic drifts, or spontaneous changes in the rapidly replicating viral genes that result in small but immunologically significant changes in the receptors. These receptor changes are able to deceive the immune system that has been previously exposed to influenza. The need to predict antigenic drifts and rely on annual vaccination for immunity calls into question the vaccine’s effectiveness. The areas that will be addressed in this paper include the effectiveness of the inactivated flu vaccine in healthy adults 16 to 65 years of age in preventing influenza and decreasing infection severity, hospitalization, and mortality.

Summary of the Evidence:

The 2018 Cochrane review by Demicheli et al. included 22 studies that addressed the effectiveness of the parenteral flu vaccine in preventing infection based on whether patient was diagnosed clinically by a provider with influenza-like-illness (ILI) or empirically with laboratory confirmed illness (LCI) by reverse transcriptase PCR (RT-PCR).³ Overall, the flu vaccine effectiveness in preventing ILI was found to be 16% (95% CI 5–25%; RR 0.84) and number needed to vaccinate (NNV) to be 29 while in the effectiveness in LCI was 59% (95% CI 53–64%; RR 0.41) with a NNV of 71. The review authors did clarify that the reason for their seemingly inconsistent NNV between ILI and LCI studies was because the incidence of clinical versus lab confirmed infection are very different and incidence is factored into calculation of the NNV. The 2014 Cochrane review further had very similar results but also did note from their panel of studies that the NNV decreased from 40 to 17 in the ILI studies if the flu vaccine was antigenically mismatched then no statistically significant efficacy was found.⁴ In the LCI studies, matched vaccines were 62% effective (95% CI 52%−69%) with a NNV of 52 and even unmatched antigen vaccines had a 55% efficacy (95% CI 41% −66%) with a NNV of 60.

The systematic review by Tricco et al used LCI data including studies with RT-PCR, viral culture, or four-fold elevation in HA antibody titers to compare the efficacy of matched and mismatched vaccines.⁵ The review showed that mismatched vaccines still had efficacy of 52% (95% CI 37%−63% from 9 studies) compared to a matched vaccine with efficacy of 65% (95% CI 54%−73% in 8 studies). The authors admitted that though the vaccine efficacy was slightly higher with the matched vaccine compared to the mismatched the confidence intervals were large and overlapping.

An observational case-control study including adults and children by Grijalva et al showed that those who had laboratory confirmed influenza pneumonia had lower odds, 0.43 (95^& CI 0.28– 0.68), of vaccination compared to those with non-influenza pneumonia.⁶ However in sub-group analysis of adults the odds ratio was 0.59 with a wide confidence interval crossing one (95% CI 0.34–1.02) suggesting the possibility of no true difference in vaccination odds between adults with influenza and non-influenza pneumonia. This was consistent with the 2018 Cochrane review which showed a 14.1–14.7% decrease in hospitalization, once again with a wide confidence interval (RR 0.96, 95% CI 0.85 to 1.08).³ No studies have addressed impact on mortality in the healthy adult patient population.

Systematic reviews pose unique challenges in identifying different methodology from studies that may not be easily compared or may lead to conflicting results. One concern is the classification of ILI as an outcome for validating vaccine efficacy because the results are not easily comparable to studies with LCI and have too many potential confounders. The authors of the Cochrane reviews did report the definition of influenza-like illness from each study they reviewed and the symptoms included were various combinations of fever, sore throat, coryza, myalgia, headache and fatigue. ^3,4 All of these symptoms, the authors admit, are consistent with both influenza and non-influenza infections. This is likely reflected in the variation in prevalence between ILI and LCI and the two fold difference in their respective NNV. The 2014 Cochrane review showed that while the antigenically mismatches vaccines were still efficacious in LCI studies, the mismatched vaccines were not efficacious in ILI studies in decreasing the NNV.⁴ This also supports the view that these ILI studies captured too many infections that were not truly due to influenza because of the broad definition of ILI, ultimately resulting in underestimation of the vaccine’s efficacy and therefore a type II error. On the other hand, the LCI results from the 2014 Cochrane review and that of Tricco et al. came to the reasonable and consistent conclusion that even mismatched vaccines were efficacious at influenza prevention. ^4,5

Conclusion:

The evidence from four systematic reviews indicates inactivated influenza vaccination efficacy of between 16–59% in healthy adults based on whether the results are clinically derived or empirically derived. The literature that did address hospitalization and pneumonia both indicated at best a small benefit of vaccination with low certainty of evidence. Finally, no evidence exists regarding the impact of the influenza vaccine on mortality. Based on this evidence, healthy individuals can be educated that the influenza vaccine decreases the likelihood of developing the influenza infection but it is less certain if there is a benefit on the severity or complications of the infection.