High-dose influenza vaccine in nursing home residents: not to be sneezed at

In this issue of The Lancet Respiratory Medicine, Stefan Gravenstein and colleagues¹ present data from a large cluster-randomised trial done to identify whether high-dose, inactivated influenza vaccine can provide better protection against severe influenza disease than standard-dose vaccine in nursing home residents. This question is important since the elderly population 65 years and over, especially those in nursing homes, account for almost 80% of the 12 000 to 56 000 annual influenza-associated deaths in the USA^2,3 and the risk of severe disease rises rapidly with increasing age. Yet standard-dose inactivated influenza vaccines, which have been administered since the 1950s in the USA, are known to be less immunogenic in older adults than in healthier, younger adults.⁴ Furthermore, recent studies⁵ have found that high-dose influenza vaccines are more efficacious than standard-dose vaccines among community-dwelling elderly people.

Gravenstein and colleagues found that of the elderly population residing in nursing homes, the high-dose vaccine was significantly more effective in preventing the primary study outcome of respiratory-related hospital admissions (adjusted relative risk [ARR]=0·873, 95% CI 0·776–0·982, p=0·02) and two of the secondary outcomes, pneumonia hospitalisations (ARR=0·791, 0·267–0·953, p=0·013) and all-cause hospital admissions, (ARR=0·915, 95% CI 0·863–0·970, p=0·0028). No significant effect of high-dose vaccine was detected on mortality or frailty. The relative reduction of 12·7% in all seasonal respiratory-related hospital admissions in high-dose vaccine recipients is striking, and the significant reduction of 8·5% in all-cause hospital admissions in the same analysis was even more so. This finding points to a notable reduction in absolute risk of all-cause hospital admissions in this at-risk population who had previously had about 1 hospital admission per 1000 patient days during the study period. Indeed, the authors estimate that vaccinating only 68·7 residents with high-dose vaccine would have prevented one hospital admission of any cause that season, on top of any hospital admissions prevented by the standard-dose vaccine.

Although these results are strong evidence that the high-dose vaccine did better in the elderly nursing home residents, a little caution is needed in the interpretation of these findings. It comes as no surprise that influenza vaccine would affect all-cause hospital admissions as well as those caused by respiratory complications, since influenza is associated with a substantial burden of hospitalisations and death for non-respiratory causes (eg, circulatory)^3,6 however, the scale of the reduction in all-cause hospitalisations in this study is unexpectedly large for two reasons. First, since the respiratory-coded hospitalisations represented only one in five all-cause hospitalisations in the study, the 8·5% reduction in seasonal all-cause hospitalisations would have comprised predominantly of averted non-respiratory hospitalisations. Second, on average, no more than 10% of all-cause seasonal mortality in elderly people has been attributed to influenza (even in years of mismatch and therefore little population protection afforded).⁷ Since the matched standard-dose comparator vaccine in this study would have prevented some influenza-associated severe disease⁸ and that the high-dose vaccine was unlikely to be 100% effective, the additional reduction of up to 8·5% of all-cause hospital admission attributed to the high-dose vaccine in this study suggests that more than 10% of all-cause seasonal hospital admissions might have been due to influenza in the study season. This is especially notable because the season was characterised by circulation of the A(H1N1) subtype which generally has a smaller impact on the elderly than in seasons in which A(H3N2) viruses predominate.³

Most previous studies of influenza vaccination in the elderly population^9,10 have been, by necessity, observational in design because people in this age group could not be randomly assigned to a placebo group for ethical reasons. The challenge for these observational studies has been the need to effectively deal with possible bias and confounders, especially when low specificity, non-laboratory confirmed outcomes have been assessed. The large, novel, and bold randomised design by Gravenstein and colleagues should prevent biased measures of vaccine effect, and moreover the cluster randomisation approximates how any vaccine would be actually used. The large sample size of a population with a high number of hospital admissions aims to be suitably powered to assess non-laboratory confirmed outcomes (as reported by discharge codes). This approach also has the advantage of potentially measuring extended benefits of influenza vaccines on other health events, which is not possible with only laboratory-confirmed outcomes because only a fraction of severe influenza outcomes would have been detected through laboratory diagnosis.

Although the long-term care facilities and their residents, had similar baseline characteristics, imbalances in more than 85% of nursing home residents who received standard-dose or high-dose vaccine did occur, which led to the potential of confounding. For instance, those nursing home residents that were more likely to receive one of the vaccines might also have been more (or less) likely to be admitted to hospital for some other reason. Indeed, the adjustment for a range of pre-specified confounders in the crude analysis increased all the point estimates in the adjusted analysis, and several became significant when they had not been before. It is hard to imagine, however, that residual unaccounted confounding would explain the amount of protection of high-dose vaccine reported in this study. Nonetheless, it would have been reassuring to see subgroup analyses done by age, comorbidity, and specific non-respiratory outcomes to assess the homogeneity of effect. Also, the few influenza diagnoses results in a small leap of faith that the observed effect is in fact due to prevention of influenza-confirmed outcomes: a nested vaccine effectiveness study with laboratory-confirmed outcomes could have addressed this concern. The study investigators chose sites close to influenza surveillance sites—a nice design feature, but one perhaps that was not taken full advantage of. Since influenza circulation timing can vary across the USA, stratifying outcomes from each site into those that occurred during local influenza circulation and those that occurred outside that period might have shown that the effect of the vaccine was highest when influenza circulation was at its peak.

In summary, this study presents strong evidence that influenza has a major role in seasonal hospital admissions of older nursing home residents and, taken together with other similar studies of high-dose versus standard-dose vaccines, it seems that high-dose influenza vaccine can reduce this burden substantially better than a standard-dose vaccine. With the recent US Food and Drug Administration approval of adjuvanted influenza vaccine for people aged 65 years and over and a recombinant protein vaccine also licensed for this age group, the public health community has a growing range of options to better reduce the burden of influenza-associated severe disease and death in this highest risk population for the first time in more than 60 years.

Lancet Respir Med 2017