Immunogenicity, Efficacy, and Effectiveness of Influenza Vaccines

Estimates of vaccine efficacy (i.e., prevention of illness among vaccinated persons enrolled in controlled clinical trials) and vaccine effectiveness (i.e., prevention of illness in vaccinated populations) of influenza vaccines depend on many factors, including the age and immunocompetence of the vaccine recipient, the degree of similarity between the viruses in the vaccine and those in circulation, study design, diagnostic testing measures, and the outcome being measured. Studies of influenza vaccine efficacy and effectiveness have used a variety of outcome measures, including the prevention of ILI, medically attended acute respiratory illness (MAARI), LCI, P&I-associated hospitalizations or deaths, and prevention of seroconversion to circulating influenza virus strains. Efficacy or effectiveness estimates for more specific outcomes such as LCI typically are higher than for less specific outcomes such as MAARI because the causes of MAARI include infections with other pathogens that influenza vaccination would not be expected to prevent (105).

Randomized controlled trials that measure LCI virus infections (by viral culture or reverse transcription polymerase chain reaction [RT-PCR]) as the outcome provide the best and most persuasive evidence of vaccine efficacy, but such data are not available for all populations. Such studies are difficult to perform in populations for which influenza vaccination is already recommended. Observational studies, particularly those that compare non-influenza-specific outcomes among vaccinated populations to those among unvaccinated populations, are more subject to biases than studies using laboratory-confirmed outcomes. For example, an observational study that finds that influenza vaccination reduces overall mortality among elderly persons might be biased if healthier persons in the study are more likely to be vaccinated and thus less likely to die for any reason (106, 107). Bias due to frailty (a characteristic which can be associated with both a lower likelihood of vaccination and increased likelihood of severe illness) is also a concern in observational studies. Observational studies that use a case-positive, control test-negative study design (in which all participants present with illness, and case/control status is assigned on the basis of influenza testing) might be less subject to frailty bias (108).

For studies assessing laboratory-confirmed outcomes, estimates of vaccine efficacy and effectiveness also might be affected by the specificity of the diagnostic tests used. A 2012 simulation study found that for each percentage point decrease in diagnostic test specificity for influenza virus infection, vaccine effectiveness would be underestimated by approximately 4% in classic case-control studies (109). In a simulation study which evaluated the effects of different values of influenza diagnostic test sensitivity and specificity on vaccine effectiveness estimates from cohort, classic case-control, and test-negative designs, it was concluded that misclassification resulted in slightly more biased VE estimates for test-negative studies than for other designs. However, the degree of bias was not thought to be meaningful when realistic combinations of attack rates, sensitivity, and specificity were considered (110).

A study of data from the National Inpatient Sample (a large database of hospital discharge data comprising approximately 8 million records annually from approximately 1,000 hospitals, representing 46 states as of 2011) noted a decrease of 295,000 hospitalizations associated with P&I of (95% CI = 139,000–451,000) and a of 13,600 P&I-associated inpatient deaths (95% CI = 2,700–24,400) for October 2008 through December 2011, compared with what would have been expected on the basis of previous rates (111). This time period correlates with that of expansion of the target groups for annual influenza vaccination to include all persons aged ≥6 months. However, it is not possible to definitively attribute these decreases directly to increased vaccination.

Humoral and cell-mediated responses to influenza vaccination among children and adults have been studied. Serum antibodies against hemagglutinin are considered to be correlates of vaccine-induced protection for inactivated influenza vaccines (IIVs) (2). Increased levels of antibody induced by vaccination decrease the risk for illness caused by strains that are antigenically similar to those strains of the same type or subtype included in the vaccine (3, 112-114). Most healthy children and adults have high titers of strain-specific antibody after IIV vaccination (113, 115). However, although immune correlates such as achievement of certain antibody titers after vaccination correlate well with immunity on a population level, reaching a certain antibody threshold (typically defined as a hemagglutination inhibition antibody (HAI) titer of 32 or 40) might not predict protection from infection on the individual level.

Compared with IIV, live attenuated influenza vaccine (LAIV) induces lower levels of serum antibodies but induces cellular immune responses more effectively. The magnitude of this effect differs among adults and children. One study of children aged 6 months–9 years and adults aged 22–49 years noted a significant increase in influenza A-specific interferon γ-producing CD4+ and CD8+ T cells among children following receipt of LAIV but not following receipt of IIV. No significant increase in these parameters was noted among adults following receipt of either vaccine (116).

Immune responses elicited by influenza vaccines are generally strain-specific. Antibody against one influenza virus type or subtype generally confers limited or no protection against another type or subtype, nor does it typically confer protection against antigenic variants of the same virus that arise by antigenic drift. However, among adults, vaccination can cause a “back boost” of antibody titers against influenza A(H3N2) viruses that have been encountered previously either by vaccination or natural infection (117).

Studies using a serological definition of influenza virus infection have raised concerns that dependence on a serological diagnosis of influenza in clinical trials might lead to overestimation of vaccine efficacy because of an “antibody ceiling” effect in adult participants with historic exposures to both natural infections and vaccination (118). This could result in the decreased likelihood that antibody increases can be observed in vaccinated participants after influenza infection with circulating viruses, as compared with adult participants in control arms of trials. Thus, vaccinated participants might be less likely to show a fourfold increase in antibody levels after influenza infection with circulating viruses compared with unvaccinated participants in such studies. Whether there is a substantial antibody ceiling effect in children, particularly younger children without extensive experience with influenza antigens, is not known.

The viral composition of influenza vaccines must be determined months in advance of the start of each season, to allow enough time for manufacture and distribution of vaccine. Selection of viruses is based on consideration of global influenza surveillance data, from which decisions are made regarding the viruses most likely to circulate during the upcoming season. During some seasons, because of antigenic drift among influenza A viruses or change in predominant lineage among B viruses, circulating viruses might differ from those included in the vaccine. Seasonal influenza vaccine effectiveness can be influenced by mismatches to circulating influenza viruses. Good match between vaccine and circulating viruses was associated with increased protection against MAARI-related ED visits and hospitalizations among older persons (119), ILI in younger working adults (47), and LCI (120) in observational studies. Results from other investigations suggest that influenza vaccine can still provide some protection against influenza and outcomes such as influenza-associated hospitalizations, even in seasons when match is suboptimal (121, 122). In addition to antigenic drift of circulating influenza viruses, vaccine viruses might undergo adaptive mutations during propagation in eggs that also can contribute to an antigenic differences between vaccine virus and circulating viruses, which in some cases, has been suggested to contribute to reduced vaccine effectiveness (123).

Inactivated influenza vaccines (IIVs) comprise the largest category of vaccines currently available. IIVs are administered by intramuscular or intradermal injection and contain nonreplicating virus. Immunogenicity, effectiveness, and efficacy have been evaluated in children and adults, although fewer data from randomized studies are available for certain age groups (e.g., persons aged ≥65 years).

Since the introduction of quadrivalent IIV (IIV4) in the United States during the 2013–14 season, both trivalent (IIV3) and quadrivalent IIVs have been available. Both IIV3s and IIV4s contain an A(H1N1) virus, an A(H3N2) virus, and a B virus. IIV4s contain the viruses selected for IIV3s, and in addition contain a fourth virus, which is a B virus selected from the opposite lineage of that selected for IIV3s.

In general, prelicensure studies of immunogenicity of the currently licensed IIV4s compared with corresponding IIV3 products from the same manufacturer have demonstrated superior immunogenicity for IIV4 for the added influenza B virus without interfering with immune responses to the remaining three vaccine viruses (124-132). Effectiveness studies conducted during some seasons have demonstrated that IIV3 provided similar protection against circulating influenza B viruses of both lineages. For example, U.S. Influenza Vaccine Effectiveness Network found that IIV3 provided statistically significant protection against both the included B lineage (66%; 95% CI = 58–73) and the nonincluded B lineage (51%; 95% CI = 36–63) during the 2012–13 season when both lineages co-circulated (133). Similarly, in an observational study conducted during the 2011-12 season, in which both B lineages co-circulated, effectiveness was similar for both (52%, 95%CI 8 to 75% for B/Victoria; and 66%, 95%CI 38 to 81% for B/Yamagata) (134). Cross-lineage protection was observed for IIV3 and ccIIV3 in a randomized trial (135); in another randomized trial of IIV3 there was no cross lineage protection (136).

Studies involving seasonal IIV among young children have demonstrated that 2 vaccine doses provide better protection than 1 dose during the first season a child is vaccinated. In a study during the 2004–05 season of children aged 5–8 years who received IIV3 for the first time, the proportion of children with protective antibody responses was significantly higher after 2 doses than after 1 dose of IIV3 for each antigen (p = 0.001 for influenza A[H1N1]; p = 0.01 for influenza A[H3N2]; and p = 0 0.001 for influenza B) (137). Vaccine effectiveness is lower among children aged <5 years who have never received influenza vaccine previously or who received only 1 dose in their first year of vaccination than it is among children who received 2 doses in their first year of being vaccinated. A retrospective study of billing and registry data among children aged 6–21 months conducted during the 2003–04 season found that although receipt of 2 doses of IIV3 was protective against office visits for ILI, receipt of 1 dose was not (138). Another retrospective cohort study of children aged 6 months through 8 years, the majority of whom received IIV3 (0.8% received LAIV3), also conducted during the 2003–04 season, found no effectiveness against ILI among children who had received only 1 dose (139). In a case-control study of approximately 2,500 children aged 6–59 months conducted during the 2003–04 and 2004–05 seasons, being fully vaccinated (having received the recommended number of doses) was associated with 57% effectiveness (95% CI = 28–74) against LCI for the 2004–05 season; a single dose was not significantly effective (too few children in the study population were fully vaccinated during the 2003–04 season to draw conclusions) (140). The results of these studies support the recommendation that all children aged 6 months–8 years who are being vaccinated for the first time should receive 2 doses separated by at least 4 weeks (see Children Aged 6 Months through 8 Years).

Estimates of the efficacy of IIV among children aged ≥6 months vary by season and study design. Limited efficacy data are available for children from studies that used culture- or RT-PCR–confirmed influenza virus infections as the primary outcome. A large randomized trial compared rates of RT-PCR–confirmed influenza virus infections among 4,707 children aged 6–71 months who received IIV3, IIV3 with MF59 adjuvant (aIIV3; not currently licensed for children in the United States), or a control vaccine (meningococcal conjugate vaccine or tickborne encephalitis vaccine). During the two seasons of the study (2007–08 and 2008–09), efficacy of IIV3 versus control vaccine was 43% (95% CI = 15–61). Efficacy of aIIV3 versus control was 86% (95% CI = 74–93) (141). In a randomized trial conducted during five influenza seasons (1985–90) in the United States among children aged 1–15 years, receipt of IIV3 reduced culture-positive influenza by 77% (95% CI = 20–93) during A(H3N2) years and 91% (95% CI = 64–98) during A(H1N1) years (113). A single-season placebo-controlled study that enrolled 192 healthy children aged 3–19 years found the efficacy of IIV3 was 56% (p<0.05) among those aged 3–9 years and 100% among those aged 10–18 years (142); influenza infection was defined either by viral culture or by a postseason antibody rise in HI titer among symptomatic children from whom no other virus was isolated. In a randomized, double-blind, placebo-controlled trial conducted during two influenza seasons among 786 children aged 6–24 months, estimated efficacy was 66% (95% CI = 34–82) against culture-confirmed influenza illness during 1999–2000. However, vaccination did not reduce culture-confirmed influenza illness significantly during 2000–2001, when influenza attack rates were lower (3% versus 16% during 1999–2000 season) (143).

Receipt of IIV was associated with a reduction in acute otitis media in some studies but not in others. Two studies reported that IIV3 decreases the risk for otitis media among children (144, 145). However, a randomized, placebo-controlled trial conducted among 786 children aged 6 through 24 months (mean age: 14 months) indicated that IIV3 did not reduce the proportion of children who developed acute otitis media during the study (143). Influenza vaccine effectiveness against a nonspecific clinical outcome such as acute otitis media, which is caused by a variety of pathogens and typically is not diagnosed by use of influenza virus detection methods, would be expected to be lower than effectiveness against LCI.

A 2012 meta-analysis found a pooled IIV3 efficacy against RT-PCR or culture-confirmed influenza of 59% (95% CI = 51–67) among adults aged 18–65 years for eight of twelve seasons analyzed in 10 randomized controlled trials (146). Vaccination of healthy adults was associated with decreased work absenteeism and use of health care resources in some studies, when the vaccine and circulating viruses are well-matched (47, 49). In another study of healthy working adults conducted during the 2012–13 season, no significant difference in missed work hours between vaccinated and unvaccinated subjects was noted (147).

Older adults have long been recognized as a high-risk group for severe influenza illness, and have been recommended to receive annual influenza vaccination since the 1960s (80). Historically, most effectiveness data in this population pertain to standard-dose IIVs, which contain 15 µg of HA of each vaccine virus per dose. Discussion of the more recently licensed high-dose IIV3 (HD-IIV3), adjuvanted IIV3 (aIIV3), and quadrivalent recombinant influenza vaccine (RIV4) in this age group occurs below.

Studies suggest that antibody responses to influenza vaccination are decreased in older adults. It is likely that increasing dysregulation of the immune system with aging contributes to the increased likelihood of serious complications of influenza infection (148). A review of HAI antibody responses to IIV3 in 31 studies found that 42%, 51%, and 35% of older adults (aged ≥58 years) seroconverted to A(H1N1), A(H3N2), and B vaccine antigens, respectively, compared with 60%, 62%, and 58% of younger persons (aged <58 years). When seroprotection (defined as an HAI titer ≥40) was the outcome, 69%, 74%, and 67% of older adults versus 83%, 84%, and 78% of younger adults achieved protective titers to A(H1N1), A(H3N2), and B antigens, respectively (149). Although an HAI titer ≥40 is considered to be associated with approximately 50% clinical protection from infection, this standard was established in young healthy adults (3), and few data suggest that such antibody titers represent a correlate of protection among elderly adults. An analysis of serologic data from a randomized controlled efficacy trial of high-dose IIV among the elderly found that an HAI titer of ≥40 corresponded to 50% protection (similar to the recognized threshold for younger adults) when the assay virus was well-matched to the circulating virus but higher titers were required with poor match (150). Limited or no increase in antibody response is reported among elderly adults when a second dose is administered during the same season (151-153).

Most data concerning vaccine effectiveness among community-dwelling older adults comes from observational studies. One randomized controlled trial conducted among community-dwelling persons aged ≥60 years found IIV3 to be 58% effective (95% CI = 26–77) against serologically-confirmed influenza illness during the 1991–92 season, during which vaccine viruses were considered to be well-matched to circulating strains (154). The outcome used for measuring the efficacy estimate was seroconversion to a circulating influenza virus and symptomatic illness compatible with clinical influenza infection, rather than viral culture or PCR-confirmed influenza infection. Use of such outcomes raises concern that seroconversion after symptomatic illness will be less likely among vaccinated persons who have higher levels of pre-existing anti-HA antibody than among those not vaccinated, leading to an overestimate of the true vaccine efficacy. This phenomenon was demonstrated in a clinical trial conducted among healthy adults aged 18 through 49 years (118).

Other evidence of effectiveness of influenza vaccines among older adults is derived from observational studies and from analyses of health care system data. A 2010 Cochrane review of influenza vaccine effectiveness studies among community-dwelling persons aged ≥65 years pooled data from 75 studies (randomized, quasi-randomized, cohort, and case-control studies) to assess efficacy against laboratory-confirmed influenza (LCI) or influenza like illness (ILI) (155). IIV3 was not significantly effective against LCI, ILI, or pneumonia. The quality of the pooled evidence was rated as generally low because of the paucity of randomized clinical trials. A different team of investigators subsequently performed a meta-analysis of these data, but using a different stratification method and examining a smaller number of clinically relevant outcomes. Using these methods, the authors estimated vaccine effectiveness for LCI of approximately 49% (95% CI = 33–62), and for ILI of 39% (95% CI = 35–43) (156). A more recent systematic review, published in 2014, included pooled data from 35 test-negative design case-control studies involving community-dwelling elderly. This review concluded that although influenza vaccine was not significantly effective during periods of localized influenza activity (defined as cases limited to one administrative unit of a country or reported from a single site), influenza vaccine was effective against LCI irrespective of vaccine match or mismatch to the circulating viruses during regional (OR: 0.42; 95% CI = 0.30–0.60 when matched; OR 0.57; 95% CI = 0.41–0.79 when not matched) and widespread outbreaks (OR: 0.54; 95% CI = 0.46–0.62 when matched; OR 0.72; 95% CI = 0.60–0.85 when not matched), although the effect was stronger when the vaccine viruses matched circulating viruses. Vaccine was effective during sporadic activity, but only when vaccine matched (OR: 0.69; 95% CI = 0.48–0.99) (157).

Influenza vaccination might reduce the frequency of secondary complications and risk for influenza-related hospitalization and death among community-dwelling adults aged ≥65 years with and without high-risk medical conditions (158-163). However, these studies have been conducted using medical record databases and did not use reductions in LCI illness as an outcome of interest. Such methods have been challenged because results might not be adjusted adequately to control for the possibility that healthier persons might be more likely to be vaccinated than less healthy persons (106, 107, 164-167). Several studies that have used methods to account for unmeasured confounding have reported effectiveness estimates for nonspecific serious outcomes such as P&I hospitalizations or all-cause mortality among community-dwelling older persons of ~10% or less, which is more plausible than higher estimates from earlier studies (168-170). A test-negative case-control study of community-dwelling adults aged ≥65 years noted that receipt of 2010–11 seasonal influenza vaccine was associated with a 42% reduction (95% CI = 29–53) in hospitalizations for LCI. When analyzed by type/subtype, the reduction was 40% (95% CI = 26–52) for influenza A(H3N2) and 90% (95% CI = 51–98) for influenza A(H1N1); no benefit was seen against influenza B (13%; 95% CI = -77–58) (171). In a study covering the 2007-08 through 2010-11 seasons, among outpatients aged ≥65 years presenting with ARI with RT-PCR-confirmed influenza, self-rated symptom severity was less for those who had been vaccinated than for those who had not (172). An analysis of data from the Influenza Hospitalization Surveillance Network (FluSurv-NET) for the 2012-13 season found no difference in symptom severity in vaccinated vs unvaccinated adults, but length of ICU stay was shorter for those aged 50 through 64 years who had been vaccinated (173). A subsequent study from the same network for 2013-14 found vaccination to be associated with reduced length of hospital and ICU stay among persons aged 50-64 years ≥65 years, as well as lower odds of in-hospital death in these age groups (174).

Influenza infection is a common cause of morbidity and death among institutionalized older adults. Influenza vaccine effectiveness in preventing respiratory illness among elderly persons residing in nursing homes has been estimated at 20%–40% (175, 176). A Cochrane review of 64 studies demonstrated that vaccination was more effective for persons living in institutional settings than for community-dwellers (177). However, documented outbreaks among well-vaccinated nursing-home populations suggest that vaccination might not have discernable effectiveness, particularly when circulating strains are drifted from vaccine strains (178, 179).

The desire to improve immune response and vaccine effectiveness among adults aged ≥65 years has led to the development and licensure of vaccines intended to promote a better immune response in this population. Currently, both a high-dose IIV3 and an aIIV3 are licensed specifically for this age group, in addition to standard-dose unadjuvanted IIVs and RIVs. Specific discussion of HD-IIV3, aIIV3, and RIV4 for older adults is discussed below (see HD-IIV3, aIIV3, and RIV4 for Older Adults).

Passive transfer of anti-influenza antibodies from vaccinated women to neonates has been documented (180-182). Protection of infants though maternal vaccination has been observed in several studies. In a randomized controlled trial conducted in Bangladesh, vaccination of pregnant women during the third trimester resulted in a 36% reduction in respiratory illness with fever among these women, as compared with women who received pneumococcal polysaccharide vaccine. In addition, influenza vaccination of mothers was 63% effective (95% CI = 5 to 85) in preventing LCI in their breastfed infants during the first 6 months of life (183). A randomized placebo-controlled trial of IIV3 among HIV-infected and uninfected women in South Africa reported efficacy against RT-PCR–confirmed influenza of 50.4% (95% CI = 14.5 to 71.2) among the HIV-uninfected mothers and 48.8% (95% CI = 11.6 to 70.4) among their infants (184). In a study conducted in Mali in which pregnant women were randomized to receive either IIV3 or quadrivalent meningococcal vaccine during the third trimester and infants were followed to detect LCI through 6 months of age, vaccine effectiveness against LCI among the infants was 67.9% (95%CI 35.2 to 85.3) through 4 months and 57.3% (95%CI 30.6 to 74.4) through 5 months; by six months of follow up effectiveness was 33.1% (95%CI 3.7 to 53.9) (185). A randomized placebo-controlled trial of year-round influenza vaccination in Nepal (where influenza circulates throughout the year, rather than seasonally), vaccine effectiveness against LCI among infants 0–6 months of age was 30% (95%CI = 5 to 48) for the full study period. Vaccines with two different compositions were used during this period; vaccine effectiveness for the vaccine used during the first period was 16% (95%CI -19 to 41) while that for the latter was 60% (95%CI = 26 to 88) (186).

Among observational studies, in a matched case-control study of infants admitted to a large urban hospital in the United States during 2000–2009, investigators found that maternal vaccination was associated with significantly lower likelihood of hospitalization for LCI among infants aged <6 months (91.5%; 95% CI = 61.7 to 98.1) (187). A prospective cohort study among Native Americans reported that infants aged <6 months of vaccinated mothers had a 41% reduction of the risk for LCI in the inpatient and outpatient settings (RR: 0.59; 95% CI = 0.37 to 0.93) and a 39% reduction in risk for ILI-associated hospitalization (RR: 0.61; 95% CI = 0.45 to 0.84) (188). In a study of 1,510 infants aged <6 months, those of vaccinated mothers were less likely to be hospitalized with LCI than those of nonvaccinated mothers (aOR: 0.55; 95% CI = 0.32 to 0.95) (189). In a case control study covering the 2010-11 and 2011-12 influenza seasons, vaccination of pregnant women reduced their risk of laboratory-confirmed influenza by approximately half (190).

In a nonrandomized controlled trial during the 1992–93 season involving 137 children who had moderate to severe asthma, vaccine efficacy against laboratory-confirmed influenza A(H3N2) infection was 54% among children aged 2 through 6 years and 78% among children aged ≥7 through 14 years; vaccine efficacy against laboratory-confirmed influenza B infection was 60% among children aged ≥7 through 14 years, but nonsignificant for the younger age group (191). In a two-season study of 349 asthmatic children, IIV3 vaccine was associated with a 55% reduction in the occurrence of ARI in children aged <6 years (95% CI = 20 to 75; p = 0.01), but no association was noted among children aged 6 through 12 years (192).

The association between vaccination and prevention of asthma exacerbations is unclear. A retrospective uncontrolled cohort study based on medical and vaccination records during three seasons (1993–94 through 1995–96) found that asthmatic children aged 1 through 6 years showed an association between receipt of IIV3 and reduced rates of exacerbations in two out of three seasons (193). In a study of 80 asthmatic children aged 3–18 years, current influenza vaccination status was associated with a significant reduction (OR: 0.29, 95% CI = 0.10 to 0.84) in oral steroid use in the 12 months before the survey (194). Other studies have failed to show any benefit against asthma exacerbation (195, 196).

A small study evaluated immune response to IIV3 among asthmatic children who were receiving prednisone for asthma exacerbation symptoms. Among 109 children aged 6 months through 18 years, 59 of whom had no asthma symptoms and 50 of whom were symptomatic and required prednisone, no difference was noted in antibody response to A(H1N1) and A(H3N2) following receipt of IIV3. Response to the B component of the vaccine was significantly better in the prednisone group (197).

There is some evidence to suggest that vaccine effectiveness among adults aged <65 years with chronic medical conditions might be lower than that reported for healthy adults. In a case-control study conducted during the 2003–04 influenza season, when the vaccine was a suboptimal antigenic match to many circulating virus strains, effectiveness for prevention of LCI (tests used were not specified) illness among adults aged 50–64 years with high-risk conditions was 48% (95% CI = 21 to 66) compared with 60% (95% CI = 43 to 72) for healthy adults. For influenza-related hospitalizations, effectiveness varied more substantially by risk status: among those with high-risk conditions, vaccine effectiveness was 36% (95% CI = 0 to 63) whereas it was 90% (95% CI = 68 to 97) among healthy adults (198).

Some observational studies have found large reductions in hospitalizations or deaths for adults with chronic medical conditions. For example, in a case-control study conducted during 1999–2000 in the Netherlands among 24,928 persons aged 18 through 64 years with underlying medical conditions, vaccination was reported to reduce deaths attributable to any cause by 78% and reduce hospitalizations attributable to acute respiratory or cardiovascular diseases by 87%. (199). Among patients with diabetes mellitus, vaccination was associated with a 56% reduction in any complication, a 54% reduction in hospitalizations, and a 58% reduction in deaths (200). Effects of this magnitude on nonspecific outcomes are likely to have been caused by confounding from unmeasured factors (e.g., dementia and difficulties with self-care) that are associated strongly with the measured outcomes (106, 107).

A randomized controlled trial conducted among 125 adults in Thailand with chronic obstructive pulmonary disease (COPD) observed that vaccine efficacy was 76% (95% CI = 32 to 93) in preventing influenza-associated acute respiratory infection (defined as respiratory illness associated with HAI titer increase and/or positive influenza antigen on indirect immunofluorescence testing) during a season when circulating influenza viruses were well-matched to vaccine viruses (201). A systematic review of studies of influenza vaccine among COPD patients identified evidence of reduced risk for exacerbation from vaccination (202). Eleven trials were included but only six of these were specifically performed in COPD patients. The others were conducted on elderly and high-risk persons, some of whom had chronic lung disease. However, a systematic review that focused on studies of adults and children with asthma concluded that evidence was insufficient to demonstrate benefit of vaccination in this population (203).

Evidence suggests that acute respiratory infections might trigger atherosclerosis-related acute vascular events (204). Some studies have attempted to evaluate the impact of vaccination on such events. Several randomized controlled trials have suggested protective efficacy of influenza vaccination against vascular events. The FLUVACS study randomized participants with known coronary artery disease to IIV3 or placebo and followed up at 6 months, 1 year and 2 years. Vaccination was associated with lower cardiovascular mortality (RR: 0.25; 95% CI = 0.07-0.86 at 6 months and RR: 0.34; 95% CI = 0.17–0.71 at 1 year) and lower risk for a composite endpoint including cardiovascular death, nonfatal myocardial infarction, or severe ischemia (RR: 0.50; 95% CI = 0.29-0.85 at 6 months and 0.59; 95% CI = 0.40–0.86 at 1 year) compared with controls (205, 206). In the FLUCAD study, a randomized trial of 658 participants with coronary artery disease, rates of coronary ischemic events at 12 months were significantly lower in the vaccinated group (hazard ratio [HR]: 0.54; 95% CI = 0.29–0.99) (207). Another composite endpoint, major CV events (including cardiovascular death, myocardial infarction, or coronary revascularization) was not significantly different between vaccinated and placebo groups. In a trial of 439 participants with acute coronary syndrome, influenza vaccination resulted in a significant reduction of major coronary adverse events (adjusted HR [aHR]: 0.67; 95% CI = 0.51–0.86), but not cardiovascular death (0.62; 95% CI = 0.34–1.12) (208). A pooled analysis of these data with those of the FLUVACS study showed a significant reduction of major cardiovascular events (pooled effectiveness 44%; 95% CI = 25 to 58), cardiovascular deaths (pooled effectiveness: 60%; 95% CI = 29 to 78); and hospitalization (pooled effectiveness 51%; 95% CI = 16 to 72) in vaccinated participants at one-year follow up (209). A self-controlled case series study conducted through medical record review of over 17,000 persons aged ≥18 years who had experienced a stroke found a reduction of 55% in the risk for stroke in the first 1–3 days after vaccination; subsequent reductions were 36% at 4–7 days, 30% at 8–14 days, 24% at 15–28 days, and 17% at 29–59 days (210).

Statin medications, a class of drugs commonly used among persons with vascular disease, are known to have immunomodulatory effects. A posthoc analysis of data from a randomized clinical trial comparing MF59-adjuvanted IIV3 and unadjuvanted IIV3 among persons aged ≥65 years demonstrated lower geometric mean titers following vaccination among persons receiving chronic statin therapy (by 38% [95% CI = 27 to 50] for A(H1N1), by 67% [95% CI = 54 to 80] for A(H3N2), and by 38% [95% CI = 28 to 49] for B). The effect was more pronounced among those receiving synthetic statin drugs (fluvastatin, atorvastatin, and rosuvastatin) relative to those receiving fermentation-derived statins (pravastatin, simvastatin, lovastatin, and Advicor) (211). A retrospective cohort study covering nine influenza seasons found reduced effectiveness of influenza vaccine against MAARI among statin users (212); however, this study did not evaluate confirmed influenza illness. In a population-based study of 3,285 adults aged 45 years and over covering the 2004-5 through 2014-15 influenza seasons, statin use was associated with lower vaccine effectiveness against LCI due to H3N2 viruses (vaccine effectiveness 45% [95%CI 27 to 59] for statin nonusers vs. -21% [95%CI -84 to 20] for statin users); statin use was not associated with lower vaccine effectiveness against H1N1pdm09 or B viruses (213).

Vaccination might be beneficial for persons with chronic liver disease. A prospective study of 311 persons with cirrhosis, 198 of whom received IIV3 and the remainder of whom were unvaccinated, noted reduction in the rates of ILI (14% versus 23%; p = 0.064) and of culture-positive influenza (2.3% versus 8.8%; p = 0.009) in the vaccinated group (214). Review of data from Taiwan’s National Health Insurance program from 2000 through 2009 noted a lower hospitalization rate among persons with chronic hepatitis B infection who had been vaccinated compared with those who had not (16.29 versus 24.02 per 1,000 person-years) (215).

Studies of the immunogenicity and effectiveness of seasonal influenza vaccine among persons with obesity have shown conflicting results. An evaluation of immunogenicity of influenza vaccine conducted among pregnant and postpartum women reported that seroconversion rates among obese women were lower than those among normal-weight participants, but the difference was not statistically significant (216). Two other observational studies focused on the impact of obesity on postvaccination immune response. One study comparing 1-month and 12-month postvaccination immune response showed that obese persons mounted a vigorous initial antibody response to IIV3 (217). However, higher BMI was associated with a decline in influenza antibody titers after 12 months postvaccination. A second study of older adults reported that immunogenicity of IIV3 was similar in obese and normal-weight older adults, with a slight increase in seroconversion for the A/H3N2 virus among those who were obese, but not for the other viruses (218). In a small study involving 51 children aged 3–14 years with varying BMI measurements (219), seroprotection rates at 4 weeks postvaccination were significantly higher against influenza A(H1N1)pdm09 strain in the overweight/obese group (p<0.05) when compared with the normal-weight group. This difference diminished over time, with the antibody response similar or slightly higher in overweight/obese children when measured 4 months postvaccination. A test-negative case-control study of hospitalized adult patients reported an unadjusted vaccine effectiveness against LCI hospitalizations of 79% (95% CI = -6– 96); after adjusting for obesity, the vaccine effectiveness estimate increased to 86% (95% CI = 19–97); the presence of obesity increased the odds of laboratory-confirmed influenza by 2.8 times (220).

In general, HIV-infected persons with minimal AIDS-related symptoms and normal or near-normal CD4+ T lymphocyte cell counts who receive IIV develop adequate antibody responses (221-223). Among persons who have advanced HIV disease and low CD4+ T-lymphocyte cell counts, IIV might not induce protective antibody titers (223, 224); a second dose of vaccine does not improve immune response (224, 225). In an investigation of an influenza A outbreak at a residential facility for HIV-infected persons, vaccine was most effective at preventing ILI among persons with >100 CD4+ cells and among those with <30,000 viral copies of HIV type-1/mL (226). In a randomized placebo-controlled trial conducted in South Africa among 506 HIV-infected adults, including 349 persons on antiretroviral treatment and 157 who were antiretroviral treatment-naïve, efficacy of IIV3 for prevention of culture- or RT-PCR–confirmed influenza illness was 75% (95% CI = 9–96) (227). In a randomized study of a two-dose regimen of IIV3 vs. placebo conducted among 410 children aged 6-59 months (92% of whom were receiving antiretroviral therapy) in South Africa during 2009, vaccine efficacy was 17.7% (95%CI = 0 to 62.4). It was suggested that poor immunogenicity and drift of the circulating H3N2 virus contributed to the poor efficacy (228).

In a randomized study comparing the immunogenicity of high-dose versus standard-dose IIV3 among 195 HIV-infected adults aged ≥18 years (10% of whom had CD4 counts under 200 cells/µL), seroprotection rates were higher in the high-dose group for A(H1N1) (96% versus 87%; p = 0.029) and influenza B (91% versus 80%; p = 0.030). Both vaccines were well-tolerated (229). However, in a comparative study of 41 children and young adults aged 3–21 years with cancer or HIV infection, high-dose IIV3 was no more immunogenic than standard-dose IIV3 among the HIV-infected recipients (230). It should be noted that high-dose IIV3 is currently licensed only for adults aged ≥65 years.

Observational studies suggest that immunogenicity among persons with solid organ transplants varies according to factors such as transplant type, time from transplant, and immunosuppressive regimen. In one review, overall seroprotective and seroconversion responses ranged from 15% to 93%, with lower responses seen in lung transplant and greater responses several years after kidney transplant (231). Among persons who have undergone kidney transplantation, seroresponse rates have been observed that were similar or slightly reduced compared with healthy persons (232-236). Response may be dependent upon time post-transplant. Antibody response among persons who were 6 months post kidney transplant were lower than observed for healthy controls in one prospective study (232). In another study, among kidney transplant recipients who were 3–10 years posttransplant, a 93% seroprotection rate to A(H1N1) antigen after vaccination was noted (233). In a study of persons with a history of kidney transplant found that influenza vaccination in the first year after transplant was associated with a lower rate of transplant rejection (aHR: 0.77; 95% CI = 0.69–0.85; p<0.001) and death (0.82; 95% CI = 0.76–0.89; p<0.001) (237). A small study involving participants with liver transplants indicated a reduced immunologic response to influenza vaccinations (238); another study noted rates were lowest if vaccination occurred within the four months after the transplant procedure (239). In a randomized controlled trial among persons who had received various solid organ transplants (kidney, liver, heart, and lung) comparing one dose of IIV3 with two doses spaced 5 weeks apart, seroprotection rates were higher at 10 weeks postvaccination among those who received two doses; there was no significant difference at one year postvaccination. Prevalence of microbiologically diagnosed influenza was similar in the two groups (2/252, or 0.8%, in the one-dose group compared with 3/247, or 1.2%, in the two-dose group) (240).

RIV was initially licensed as the trivalent vaccine, Flublok (RIV3, Protein Sciences, Meriden, Connecticut), A quadrivalent formulation, Flublok Quadrivalent (RIV4; Protein Sciences, Meriden, Connecticut) was licensed in 2016. For the 2017-18 season, it is anticipated that both RIV3 and RIV4 will be available.

RIVs contain 45 µg of purified HA protein per virus (135 µg total for RIV3; 180 µg total for RIV4). The HA proteins are produced via the introduction of the genetic sequence for the HA into an insect cell line (Spodoptera frugiperda) via a Baculovirus viral vector. This process uses neither live influenza viruses nor eggs (241, 242).

As a relatively new type of influenza vaccine, fewer postmarketing effectiveness data are available for RIVs than IIVs. Initial licensure of RIV3 was for persons aged 18 through 49 years. In prelicensure studies comparing RIV3 versus placebo among persons aged 18 through 49 years, serum antibody responses were induced to all three vaccine components (243). In a randomized placebo-controlled study conducted among healthy persons aged 18 through 49 years during the 2007–08 influenza season (242, 244), estimated vaccine effectiveness for CDC-defined ILI with a positive culture for influenza virus was 75.4% (95% CI = -148.0–99.5) against matched strains. Of note, more precise estimation of vaccine effectiveness against matched strains was not possible because 96% of isolates in this study did not antigenically match the strains represented in the vaccine (242). Estimated vaccine effectiveness without regard to match was 44.6% (95% CI = 18.8–62.6) (244).

In October 2014, the approved age indication for Flublok was expanded to ≥18 years on the basis of data from randomized trials demonstrating adequate immunogenicity among persons aged ≥50 years (245, 246). More recently, a pre-licensure randomized controlled trial of Flublok Quadrivalent vs. a licensed comparator IIV4 was performed among persons aged ≥50 years during the 2014-15 season (247, 248). This study is discussed in a later section (see HD-IIV3, aIIV3, and RIV4 for Older Adults). When evaluated in children 6 through 59 months of age, Flublok was found be safe but less immunogenic than comparable volumes of IIV3, particularly amongst children <36 months of age (249). Flublok is not licensed for children <18 years of age.

Given the high risk of severe influenza illness for older adults, and the lesser benefit of vaccination in this age group noted in many studies, substantial efforts have gone toward the development and study of new influenza vaccines intended to provide better immunity in this age group. Vaccines recently licensed specifically for persons aged ≥65 years include high-dose IIV3 (HD-IIV3) and adjuvanted IIV3 (aIIV3). In recent years, studies have been conducted comparing the benefits for older adults of these vaccines, as well as for RIV4, with those conferred by standard-dose, unadjuvanted IIVs (SD-IIVs); a few have been studies of LCI-related outcomes (Table 3 in Recommendations document). For each of these vaccines, there is at least some evidence of benefit as compared with SD-IIVs. However, no studies directly comparing these three vaccines to one another have been reported.

The only high-dose IIV, Fluzone High-Dose (Sanofi Pasteur, Swiftwater, Pennsylvania), is licensed for persons aged ≥65 years and has been available since the 2010–11 influenza season. It is a trivalent formulation containing 60 µg of HA of each vaccine virus per dose (180 µg total), compared with 15 µg of each vaccine virus per dose in standard-dose IIVs (250). Licensure was based on superior immunogenicity compared with standard-dose IIV in this age group. Immunogenicity data from three studies of high-dose IIV3 among persons aged ≥65 years indicated that vaccine with four times the HA antigen content of standard-dose vaccine elicited substantially higher HAI titers (251-253). Prespecified criteria for superiority in one clinical trial study were defined by a lower bound of the 95% CI for the ratio of geometric mean HI titers of >1.5, and a lower bound of the 95% CI for the difference in seroconversion rates (fourfold rise of HI titers) of >10%. These criteria were met for influenza A(H1N1) and influenza A(H3N2) virus antigens, but not for the influenza B virus antigen (for which criteria for noninferiority were met)(252, 254).

The largest study to date documenting benefit of high-dose vaccine for older adults is a randomized comparative efficacy trial of high-dose versus standard-dose IIV3 conducted among nearly 32,000 persons aged ≥65 years over the 2011–12 and 2012–13 influenza seasons. The primary endpoint of this study was efficacy of HD-IIV3 relative to SD-IIV3 in preventing culture- or RT-PCR-confirmed influenza caused by any influenza viral types or subtypes, and associated with protocol-defined ILI. Protocol-defined ILI was specified as occurrence of at least one respiratory symptom (sore throat, cough, sputum production, wheezing, or difficulty breathing) concurrent with at least one systemic symptom (temperature >99.0°F, chills, tiredness, headaches or myalgia). For this outcome, the study reported 24.2% (95%CI 9.7 to 36.5) greater relative efficacy of the high-dose IIV3 compared to standard-dose IIV3 for protection against LCI caused by any viral type or subtype (255). In this study, superior efficacy of Fluzone High-Dose compared to SD-IIV3 was demonstrated. The pre-specified statistical superiority criterion for the primary endpoint (lower limit of the 2-sided 95% CI of vaccine efficacy of Fluzone High-Dose relative to Fluzone >9.1%) was met (250). For a secondary outcome, prevention of culture-confirmed influenza caused by viral types/subtypes similar to those contained in the vaccine and associated with modified CDC-defined ILI (temperature >99°F with cough or sore throat), the relative efficacy of HD-IIV3 vs. SD-IIV3 was 51.1% (95%CI 16.8 to 72.0). While this study did not examine health care utilization, pneumonia, and deaths confirmed to be due to influenza, all-cause hospitalizations, deaths, and pneumonia cases were examined. A subsequent analysis of data from this trial, in which SAEs were evaluated for possible relatedness to influenza by blinded physician reviewers, reported that compared to SD-IIV3, HD-IIV3 was associated with a relative vaccine efficacy of 39.8% (95%CI 19.3 to 55.1) for serious pneumonia and 17.7% (95%CI 6.6 to 27.4) for serious cardiopulmonary events possibly related to influenza; relative efficacy against all-cause hospitalizations was lower (6.9%; 95%CI 0.5 to 12.8) (256).

In addition to the clinical outcomes evaluated in the study described above, healthcare-consumption data derived from this trial were used to perform a cost-effectiveness analysis (257). Mean participant medical costs in the study were lower among those who received HD-IIV3 ($1376.52) than those who received SD-IIV3 ($1492.64; difference=-115.62, 95%CI -264.18 to 35.48). Mean societal costs were also lower among the HD-IIV3 participants ($1506.48 vs. $1634.50; difference=-128.02, 95%CI -286.89 to 33.30). A probabilistic sensitivity analysis indicated that the HD-IIV3 is 93% likely to be cost saving.

A cluster-randomized trial conducted during the 2013-14 season among residents of 823 U.S. nursing homes (409 facilities in which residents received HD-IIV3 and 414 in which they received SD-IIV3) evaluated risk of hospital admissions related to pulmonary or influenza-like illnesses (258). The included facilities included 75,917 residents aged 65 years and older, 53,008 of whom were considered long-stay residents. Outcomes were identified via Medicare hospital claims data, which were matched to 38,256 residents. The incidence of respiratory-related admissions was significantly lower among the facilities randomized to HD-IIV3 (adjusted relative risk [aRR] 0.873, 95%CI 0.776 to 0.982). Also significantly lower were rates for pneumonia admissions (aRR 0.791, 95%CI 0.267 to 0.953), and all-cause hospital admissions (aRR 0.915, 95%CI 0.863 to 0.970).

An observational study conducted during the 2010-11 season among patients aged ≥65 years receiving primary care at Veterans Health Administration medical centers noted no significant differences in effectiveness of HD-IIV3 vs. SD-IIV3 for hospitalizations with a discharge diagnosis for influenza or pneumonia. Receipt of HD-IIV3 was also not associated with lower rates of all-cause hospitalization. However, for the subset of participants aged ≥85 years, receipt of HD-IIV3 was associated with lower risk of hospitalization for pneumonia and influenza (risk ratio 0.52; 95%CI 0.29 to 0.9) (259).

HD-IIV3 has also been evaluated through analysis of Medicare data. Among 929,730 recipients aged ≥65 years of HD-IIV3 and 1,615, 545 recipients of SD-IIV3 during the 2012-13 season, receipt of HD-IIVs was associated with fewer non-laboratory confirmed but probable influenza infections (defined as receipt of a rapid influenza diagnostic test followed by a prescription for oseltamivir, relative VE 22%, 95%CI 15 to 29) and hospital admissions with a billing code for influenza (relative VE 22%, 95%CI 16 to 27) (260). In an analysis of Medicare data from the 2012-13 and 2013-14 seasons (including 1,039,645 recipients of HD-IIV and 1,683,264 recipients of SD-IIV during 2012–13, and 1,508,176 HD-IIV and 1,877,327 SD-IIV recipients during 2013–14), receipt of HD-IIV3 was associated with reduced risk of death relative to SD-IIV3 during the 2012-13 season (36.4%; 95% CI, 9.0% to 56%), when A(H3N2) viruses predominated; but not during the 2013-14 season (2.5%; 95% CI, –47% to 35%], in which A(H1N1) viruses predominated (261).

The only adjuvanted influenza vaccine in the U.S., Fluad (Seqirus, Holly Springs, North Carolina), was initially licensed in the U.S. in November 2015. It contains the oil-in-water adjuvant, MF59. Similarly to HD-IIV3, it is licensed specifically for persons aged ≥65 years. Several studies have compared aIIV3 with SD-IIV3; however, fewer data are available than for HD-IIV3, and there have been no randomized trials of relative efficacy against LCI among older adults. In a comparison of immunogenicity of the two vaccines, Fluad met criteria for noninferiority for all three vaccine viruses based on predefined thresholds for seroconversion rate differences and GMT ratios; criteria for superiority were not met (262, 263). In a Canadian observational study of 282 persons aged ≥65 years (165 of whom received aIIV3, 62 of whom received SD-IIV3, and 55 of whom were unvaccinated) conducted during the 2011–12 season that compared Fluad with unadjuvanted IIV3, the relative effectiveness of Fluad against LCI among the 227 vaccinated participants was reported to be 63% (95% CI = 4 to 86) (264). Some differences in the populations receiving each vaccine were described (in two of three health authorities participating, persons aged 75 years and older and those in long-term care facilities were preferentially given aIIV3; in the third, those in long term care facilities received aIIV3 and all others received SD-IIV3). A prospective study of 107,661 medical records covering 170,988 person-seasons during the 2006-07 through 2008-09 influenza seasons reported lower relative risk of hospitalizations coded for influenza and pneumonia among persons aged 65 years and older who received aIIV3 as compared with IIV3 (relative risk 0.75, 95%CI=0.57 to 0.98) (265). An observational study conducted in Italy during the 2010-11 and 2011-12 seasons, in which unadjuvanted SD-IIV3 was used during the first season and aIIV3 during the second season, reported that aIIV3 was more effective in preventing hospitalizations coded for pneumonia and influenza (not LCI) among recipients aged ≥75 years (adjusted VE 53%, 95%CI=33 to 68 for aIIV3 vs. adjusted VE 46%, 95%CI=24 to 62 for IIV3), while unadjuvanted SD-IIV3 was more protective than aIIV3 for recipients aged 65 through 74 years (adjusted VE 53%; 95%CI 3 to 78 for IIV3 vs. adjusted VE 34%, 95%CI=24 to 65) (266). That the two vaccines were not compared during the same season may be a limitation of this study.

Two RIVs are currently licensed in the U.S.: Flublok (RIV3) and Flublok Quadrivalent (RIV4; Protein Sciences, Meriden, Connecticut). Both are licensed for persons aged 18 years and older. Fewer data are available concerning the relative effectiveness of RIV4 compared with other licensed vaccines for this age group than is currently the case for HD-IIV3. In prelicensure studies comparing RIV3 versus placebo among persons aged 18 through 49 years, serum antibody responses were induced to all three vaccine components (243). In a study comparing RIV3 with IIV3 among persons aged ≥65 years, seroconversion rates against influenza A(H1N1) and A(H3N2) were higher in the RIV3 group. Response was inferior for influenza B; however, this result is difficult to interpret as the B antigens were different in the two vaccines (245). In a pre-licensure randomized controlled trial of Flublok Quadrivalent vs. IIV4 among 8,604 persons aged ≥50 years during the 2014-15 season, RIV4 was more effective in prevention of LCI than IIV4, with a relative efficacy of 30% (95%CI 10 to 47). This season was characterized by a predominance of drifted A(H3N2) viruses, and consequent poor match between vaccine and circulating viruses (247, 248). While the study was not powered for statistical significance for relative efficacy by influenza virus type or subtype, results showed a trend towards non-inferior relative efficacy for Flublok Quadrivalent against influenza A, but not against influenza B (for which there were fewer cases). Relative efficacy for all A(H3N2) was 36% (95% CI: 14 to 53) and for influenza B was 4% (95% CI: -72 to 46). The RIV4 Influenza B antigens were well matched to circulating strains. In a subanalysis of data from those aged ≥65 years against all influenza A and B, RIV4 was not significantly more effective than IIV4 against RT-PCR-confirmed protocol-defined ILI (relative efficacy 17%, 95%CI=-20 to 43), but was more effective than IIV4 against culture-confirmed protocol-defined ILI (relative efficacy 42%, 95%CI=9 to 65).

LAIV virus strains replicate in nasopharyngeal epithelial cells. The protective mechanisms induced by vaccination with LAIV are not understood completely but appear to involve both serum and nasal secretory antibodies, as well as cell-mediated immune responses. The immunogenicity of LAIV3 has been assessed in multiple studies (267-269).

The single LAIV licensed in the United States was originally a trivalent vaccine (FluMist; MedImmune, Gaithersburg, Maryland). FluMist Quadrivalent was licensed by FDA in 2012, and replaced the trivalent formulation beginning with the 2013–14 season. Prelicensure studies comparing LAIV4 to LAIV3 demonstrated that HAI antibody responses to LAIV4 were noninferior to responses to LAIV3 among healthy children and adults ≤49 years (270, 271).

A large randomized, double-blind, placebo-controlled trial among 1,602 healthy children aged 15–71 months assessed the efficacy of LAIV3 against culture-confirmed influenza during two seasons (1996-97 and 1997-98) (272, 273). During the first season, when vaccine and circulating virus strains were well-matched, efficacy against culture-confirmed influenza was 94% (95% CI = 88–97) for participants who received 2 doses of LAIV3 separated by >6 weeks, and 89% (95% CI = 65–96) for those who received 1 dose (272). During the second season, when the A(H3N2) component in the vaccine was not well-matched with circulating virus strains, efficacy for 1 dose was 86% (95% CI = 75–92) for this virus. The overall efficacy for any influenza during the two seasons was 92% (95% CI = 88–94) (273). In a randomized placebo-controlled trial comparing 1 dose versus 2 doses of LAIV3 in 3,200 vaccine-naïve children aged 6–35 months in South Africa, Brazil, and Argentina during the 2001 and 2002 seasons, efficacy was 57.7% (95% CI = 44.7–67.9) after 1 dose of LAIV3 and 73.5% (95% CI = 63.6–81) after 2 doses (274) during the first year of the study. Other two-season, randomized, placebo-controlled trials have demonstrated similar efficacy rates of LAIV3 among young children, ranging from 85% to 89% among children in childcare (275) to 64% to 70% for children in eight regions in Asia (276).

Effectiveness studies have demonstrated that LAIV3 use among healthy children was associated with reduced risk of outcomes other than LCI. In one community-based, nonrandomized open-label study, reductions in MAARI were observed during the 2000–01 season among children who received 1 dose of LAIV3 during 1999–2000 or 2000–2001), even though antigenically drifted influenza A(H1N1) and B viruses were circulating during the latter season (277). Receipt of LAIV3 resulted in 21% fewer febrile illnesses (95% CI = 11–30) and 30% fewer febrile otitis media diagnoses (95% CI = 18–45) (272). A meta-analysis of six placebo-controlled studies concluded that the effectiveness of LAIV3 against acute otitis media associated with culture-confirmed influenza among children aged 6–83 months was 85% (95% CI = 78–90) (278).

A randomized, double-blind, placebo-controlled trial of LAIV3 effectiveness among 4,561 healthy working adults aged 18 through 64 years assessed multiple endpoints, including reductions in self-reported respiratory tract illness without laboratory confirmation, work loss, health care visits, and medication use during influenza outbreak periods. The study was conducted during the 1997–98 influenza season, when the vaccine and circulating A(H3N2) viruses were not well-matched. The frequency of febrile illnesses was not significantly decreased among LAIV3 recipients compared with those who received placebo. However, vaccine recipients had significantly fewer severe febrile illnesses (19% reduction) and febrile upper respiratory tract illnesses (24% reduction); and significant reductions in days of illness, days of work lost, days with health care provider visits, and use of prescription antibiotics and over-the-counter medications (279). Estimated efficacy of LAIV3 against influenza confirmed by either culture or RT-PCR in a randomized, placebo-controlled study among approximately 1,200-2,000 young adults was 48% (95% CI = -7–74) in the 2004–05 influenza season, 8% (95% CI = -194–67) in the 2005–06 influenza season, and 36% (95% CI = 0–59) in the 2007–08 influenza season; efficacy in the 2004–05 and 2005–06 seasons was not significant (280-282).

Studies comparing the efficacy of IIV3 to that of LAIV3 among adults have been conducted in a variety of settings and populations using several different outcomes. Among adults, most comparative studies demonstrated that LAIV3 and IIV3 have similar efficacy, or that IIV3 was more efficacious (280-285). In a retrospective cohort study comparing LAIV3 and IIV3 among 701,753 nonrecruit military personnel and 70,325 new recruits, among new recruits, incidence of ILI was lower among those who received LAIV3 than IIV3. The previous vaccination status of the recruits was not stated; it is possible that this population was relatively naïve to vaccination compared with previous service members who were more likely to have been vaccinated routinely each year (286).

Several studies comparing LAIV3 with IIV3 prior to the 2009 pandemic demonstrated superior efficacy of LAIV3 among young children (283, 287-290). A randomized controlled trial conducted among 7,852 children aged 6–59 months during the 2004–05 influenza season demonstrated a 55% reduction in cases of culture-confirmed influenza among children who received LAIV3 compared with those who received IIV3 (288). In this study, LAIV3 efficacy was higher compared with IIV3 against antigenically drifted viruses and well-matched viruses. An open-label, nonrandomized, community-based influenza vaccine trial conducted among 7,609 children aged 5–18 years during an influenza season when circulating A(H3N2) strains were poorly matched with strains contained in the vaccine also indicated that LAIV3, but not IIV3, was effective against antigenically drifted A(H3N2) viruses. In this study, children who received LAIV3 had significant protection against LCI (37%) and P&I events (50%) (290). LAIV3 provided 32% increased protection in preventing culture-confirmed influenza compared with IIV3 in one study conducted among children aged ≥6 years and adolescents with asthma (289) and 52% increased protection compared with IIV3 among children aged 6–71 months with recurrent respiratory tract infections (287).

On the basis of these data, in June 2014, ACIP recommended that when immediately available, LAIV should be used for healthy children aged 2 through 8 years who have no contraindications or precautions (291). However, subsequent analysis of data from three observational studies of LAIV4 vaccine effectiveness for the 2013–14 season (the first season in which LAIV4 was available) revealed no statistically significant effectiveness of LAIV4 against influenza A(H1N1)pdm09 among children aged 2 through 17 years (292-294). Analysis of data from the U.S. Influenza Vaccine Effectiveness Network for the 2010–11 through 2013–14 seasons noted that children aged 2 through 17 years who received LAIV had similar odds of influenza regardless of receipt of LAIV3 or IIV3 during 2010–11 through 2012–13; however, during the 2013–14 season odds of influenza were significantly higher for those who received LAIV4 (OR: 5.36; 95% CI = 2.37–12.13 for children aged 2 through 8 years; OR: 2.88; 95% CI = 1.62–5.12 for children aged 2 through 17 years) (295). During this season, the H1N1pdm09 virus predominated for the first time since the 2009 pandemic. During the 2014-15 season, when antigenically drifted H3N2 viruses predominated, neither LAIV4 nor IIV provided significant protection among U.S. children aged 2 through 17 years; LAIV4 did not offer greater protection than IIV for these viruses (296-298), in contrast to earlier studies in which LAIV3 provided better protection than LAIV against drifted H3N2 viruses (288, 290). LAIV4 exhibited significant effectiveness against circulating influenza B viruses in these U.S. studies. Based on these influenza vaccine effectiveness data for the 2013–14 and 2014–15 seasons, ACIP concluded that a preference of LAIV4 over IIV was no longer warranted (299).

The diminished effectiveness against H1N1pdm09 during the 2013-14 season was hypothesized to be attributable to reduced stability and infectivity of the A/California/2009/(H1N1) vaccine virus, conferred by a single amino acid mutation in the stalk region of the HA protein (300), and possibly associated with exposure of some LAIV lots to temperatures above those recommended for storage during U.S. distribution (301). For the 2015–16 season, to address stability concerns surrounding the A/California/7/2009(H1N1) HA, a different influenza A(H1N1) virus was included in LAIV4 (A/Bolivia/559/2013[H1N1]) (302). During the 2015-16 season, in which A(H1N1)pdm09 viruses were again predominant, data from the U.S. Influenza Vaccine Effectiveness Network, U.S. Department of Defense, and MedImmune demonstrated no statistically significant effectiveness of LAIV4 among children aged 2 through 17 years against H1N1pdm09, although point estimates varied (303). Conversely, estimated effectiveness of IIV against these viruses among children aged 2 through 17 years was significant across all three studies. Following review of this information in June 2016, ACIP made the interim recommendation that LAIV4 should not be used for the 2016–17 influenza season (304).

Estimates of effectiveness of LAIV against H1N1pdm09 have not been consistent in all studies and all countries. Point estimates of vaccine effectiveness have varied. In the United Kingdom, where a phased rollout of routine use of LAIV for healthy children begin during the 2013-14 season, effectiveness of LAIV4 among 2 through 17 year olds during the 2015-16 season was estimated to be 57.6% (95%CI 25.1—76.0) for all influenza, 41.5% (95%CI -8.5—68.5) for H1N1pdm09, and 81.4% (95%CI 39.6—94.3) against influenza B (305). In Finland during the 2015-16 season, effectiveness of LAIV4 against H1N1pdm09 among 2-year-olds was 50.7% (95%CI 28.4—66.1) against all influenza, 47.9% (95%CI 21.6—65.4) for influenza A (presumably predominantly H1N1pdm09), and 57.2% (95%CI 0.0—81.7) for influenza B (306). In addition to the different age group under study (2 year olds vs. 2 through 17 year olds), these results contrast with those of the U.S. and the United Kingdom, in that the estimate for H1N1 is statistically significant, whereas that for influenza B is not (and has a lower point estimate). In both the United Kingdom and Finland, as in the U.S., the point estimates for effectives of LAIV against H1N1pdm09 were lower for LAIV than for IIV. In Canada, data collected with the Sentinel Provider Site Surveillance Network (SPSN) for both 2013-14 and 2015-16 showed similar point estimates for effectiveness against H1N1pdm09 for LAIV (LAIV3 in 2013-14 and LAIV4 in 2015-16) and IIV; however, the estimate for LAIV in each case was not statistically significant (likely due to the small sample size in these analyses) (307). The Canadian National Advisory Committee on Immunization (NACI) concluded that for the 2016-17 season, the Canadian preference of the use of LAIV for 2 through 17 year olds was no longer supported by the available data.

The cause(s) of the low effectiveness of LAIV observed against H1N1pdm09, and for the differences in reported effectiveness among the various populations and countries studied, are not completely understood. Multiple factors may be involved. In addition to the reduced replicative fitness of the A/California/7/2009 and A/Bolivia/559/2013 H1N1 viral constructs (currently believed to be the root cause (308)), a potential contributory factor might be interference associated with the introduction of the second influenza B strain in LAIV4 (which was introduced in the U.S. in 2013-14). However, reduced effectiveness of LAIV against influenza A(H1N1)pdm09 was also observed with LAIV3 in the U.S. during the 2010-11 season (295). It has also been hypothesized that differences in prior vaccine coverage among children may contribute to differences in replicative fitness in different populations, leading to differences in effectiveness. However, analyses of U.S. data from the US Flu VE Network revealed no significant effect of prior vaccination (303). ACIP will continue to review data relevant to LAIV as it becomes available.

The composition of influenza vaccines is changed in most seasons, with one or more vaccine strains replaced annually to provide protection against viruses that are anticipated to circulate. Even in seasons in which vaccine composition does not change, annual vaccination has been recommended because of decline in protective antibodies over time postvaccination (309-311); however, the rate and degree of decline observed has varied. One study of HA and neuraminidase antibody levels following vaccination of adults noted a slow decline, with a 2-fold decrease in titer estimated to take >600 days (312). A review of studies reporting postvaccination seroprotection rates among adults aged ≥60 years noted that seroprotection levels meeting Committee of Proprietary Medicinal Products standards were maintained for ≥4 months for the H3N2 component in all 8 studies and for the H1N1 and B components in 5 of 7 studies (313).

Nonetheless, concerns have arisen regarding waning of protection within the course of a single influenza season, particularly among adults. Several more recent observational studies have attempted to evaluate changes in influenza vaccine effectiveness over the course of a single influenza season. Some of these studies have noted a decrease in vaccine effectiveness, particularly against influenza A(H3N2) viruses, most markedly among older adults (314-317). A test negative case-control study of children and adults conducted in Navarre, Spain during the 2011–12 season noted a decline in vaccine effectiveness, from 61% (95% CI = 5–84) in the first 100 days after vaccination to 42% (95% CI = -39–75) between days 100–119 and then to -35% (95% CI = -211–41) after ≥120 days. Persons vaccinated >120 days before diagnosis were at an increased risk for contracting influenza, when compared with those vaccinated <100 days (OR: 3.45; 95% CI = 1.10–10.85; p = 0.034) (315). This decline primarily affected persons aged ≥65 years, among whom the OR for influenza was 20.81 (95% CI = 2.14–202.71; p = 0.009) for persons vaccinated >120 days before diagnosis versus those vaccinated <100 days before diagnosis. A similar study conducted in the United Kingdom, also during the 2011–12 season, estimated an overall vaccine effectiveness against A(H3N2) of 53% (95% CI = 0–78) among those vaccinated <3 months prior, and 12% (95% CI = -31–41) for those vaccinated ≥3 months prior. The proportion of older participants was too small to detect a substantial difference in vaccine effectiveness in this age group (317). An additional case-control analysis from the 2007–08 season revealed a modest but significant increase in the OR for A(H3N2) influenza every 14 days after vaccination among young children (OR for influenza increasing 1.2 for each 14-day interval for children aged 2 years) and older adults (1.3 for each 14-day interval for adults aged 75 years); the same pattern was not observed among older children and younger adults (314). A recent multiseason analysis from the U.S. Influenza Vaccine Effectiveness (U.S. Flu VE) Network found that VE declined by about 7% per month for H3N2 and influenza B, and 6—11% per month for H1N1pdm09 (318). VE remained greater than zero for at least five to six months after vaccination. Similar waning effects have not been observed consistently across age groups and virus subtypes in different populations, and the observed decline in protection could be attributable to bias, unmeasured confounding, or the late season emergence of antigenic drift variants that are less well-matched to the vaccine strain.

Observations of a potential negative effect of repeat vaccination on vaccine effectiveness were initially made during the 1970s (319-322). A number of recent studies have indicated that response to, and effectiveness of, influenza vaccine during any given season may be modified by receipt of vaccine in prior seasons. In a study conducted among healthy 30- through 60-year olds during the 1983-84 through 1987-88 seasons during which whole-virus seasonal trivalent inactivated vaccines were used (the one exception being the addition of a monovalent split-virus A(H1N1) recommended to supplement the trivalent vaccine in 1986), moderate reductions in serum antibody response were associated with increased prior exposure to influenza vaccine during the last seasons of the study, but no decrease in protection against infection was noted (323).

Some more recent studies have noted decreased effectiveness associated with vaccination in the prior season. In a community-based study in Michigan conducted in 2010-11 (during which H3N2 viruses predominated), overall vaccine effectiveness was low and not significant (31%, 95% CI -7 to 55%). When stratified by whether vaccine had been received the previous season, substantially lower effectiveness was noted among those who had been vaccinated during both 2010-11 and 2009-10 (-45%, 95% CI -226 to 35), as compared with those who received vaccine during only the latter season (62%, 95%CI 17 to 82%) (324). In a similarly designed study in the same community conducted during the 2013-14 season, when H1N1pdm09 predominated, no negative effect of prior season vaccination was observed (325). A study in Australia conducted over the 2010-11 through 2014-15 seasons noted no significant difference in effectiveness of hospitalization for influenza illness between those vaccinated in the current season only (35%, 95%CI 21 to 46) vs the prior season only (33%, 95%CI 17 to 47); effectiveness was highest among those who had received vaccine during both seasons (51%, 95%CI 45 to 57) (326).

Other studies have evaluated vaccination history over more than one prior season. In a case-control study conducted in a healthcare system in Wisconsin, covering eight seasons between 2004-05 and 2012-13, which examined effectiveness against (H3N2) and B viruses, and in which participants were classified as frequent vaccinees (had received IIV during 4 or 5 of the previous 5 seasons), infrequent vaccinees (received IIV during 1 to 3 of the previous 5 seasons) or nonvaccinees (received no IIV during the previous 5 seasons), current season vaccination was effective regardless of previous vaccination history. Considering vaccination history for only current and prior seasons, effectiveness was similar for those who were vaccinated during the current season only, the previous season only, or both seasons. However, in an analysis using 5 seasons of vaccination history, there were significant differences in vaccine effectiveness among frequent vaccinees as compared with nonvaccinees (327). In a Spanish study which evaluated the effectiveness of vaccination against H1N1pdm09 from the 2010-11 through 2015-16 seasons, compared with those who had never been vaccinated, effectiveness was not reduced among those who were vaccinated in the current season only, or among those who were unvaccinated in the current season but had previously received >2 doses of H1N1pdm09-containing vaccine. However, effectiveness was lower among those vaccinated in the current season after >2 prior doses, and among those currently unvaccinated who had 1-2 prior doses (328).

The effects of prior vaccination have not been observed consistently across all studies and seasons, and may differ by influenza virus type or subtype. A better understanding of these effects is needed in order to guide recommendations. Importantly, most studies in which a negative effect of prior vaccination was observed, vaccination during the current season (with or without prior season vaccination) was more protective than being unvaccinated in the current season.