The Confusion Surrounding High-Dose Influenza Vaccine’s Effectiveness in Reducing Hospitalizations in Older Adults
So you are the editor of a prominent medical journal, and you receive two seemingly well-executed studies involving large patient numbers and addressing the same issue. Most editors love these rare occurrences because, as long as the reported results are similar, two independent studies reaching the same conclusions establish the issue of this journal as a reference for many years to come ( and every journal, of course, strives to be indexed as much as possible).
But what are you supposed to do if you receive two distinct manuscripts of supposedly well-executed studies as proposed above, but the reported outcomes significantly differ? According to the editors of The New England Journal of Medicine, when recently confronted with exactly this kind of problem, you invite a third party to write an editorial.
So here are the specifics: Based on several studies in the literature, most countries nowadays recommend for their older citizens (above age 65)—in contrast to younger individuals—a “high(er)-dose” inactivated influenza vaccine because several studies have demonstrated in these older patients improved protection against laboratory-confirmed influenza. However, whether these high(er)-dose vaccines also protect against severe clinical outcomes (which includes the need for hospitalization) was apparently not equally well established because, at the same time, Spanish (GALFLU trial) and Danish (DANFLU-2 trial) groups of investigators independently studied this question in large patient populations in their respective countries. And, as the following verbatim conclusions of both studies (in bold) will demonstrate differed at least to a minor, yet clinically relevant, degree:
GALFLU TRIAL:1 Among adults 65 to 79 years of age, there appeared to be fewer hospitalizations for influenza or pneumonia with a high-dose inactivated influenza vaccine than with standard dose.
Here, 103,169 participants were randomized: 31,307 participants were enrolled in two flu seasons and were counted for each during 2023–2024 and 2024–2025, representing 59,490 and 74,986 participants, respectively. Mean (±SD) age was 72.3±4.3 years, and 53.6% were men.
A primary end-point event occurred in 174 of 67,093 participants (absolute risk, 0.26%) in the high-dose group and in 227 of 66,789 (absolute risk, 0.34%) in the standard-dose group (relative vaccine effectiveness, 23.7%; 95% confidence interval [CI], 6.6 to 37.7). Hospitalization for influenza occurred in 63 of 67,093 participants (absolute risk, 0.09%) with high-dose and in 92 of 66,789 (absolute risk, 0.14%) with standard-dose group (relative vaccine effectiveness, 31.8%; 95% CI, 5.0 to 51.3). The incidence of serious adverse events appeared, thus, in contrast to the author’s own conclusions (see above in bold), appeared similar.
DANFLU-2 TRIAL:2 In this trial, a high-dose inactivated influenza vaccine did not result in a significantly lower incidence of hospitalization for influenza or pneumonia than a standard dose among older adults.
In this pragmatic, open-label, randomized, controlled trial during 2022–2023, 2023–2024, and 2024–2025 influenza seasons, older adults (≥65 years of age) received either a high or standard dose of the inactivated vaccine. The primary endpoint was hospitalization for influenza or pneumonia. Of 332,438 participants, 166,218 were assigned to the high-dose vaccine, 166,220 to the standard-dose vaccine [mean (±SD) age of the participants was 73.7±5.8 years]. and 161,538 participants (48.6%) were women. A primary end-point event occurred in 1138 high-dose (0.68%) and in 1210 (0.73%) standard-dose participants (relative vaccine effectiveness, 5.9%; 95.2% confidence interval [CI], −2.1 to 13.4; P=0.14). Hospitalization for influenza occurred in 0.06% in high-dose and in 0.11% of standard-dose patients (relative vaccine effectiveness, 43.6%; 95.2% CI, 27.5 to 56.3); hospitalization for pneumonia occurred in 0.63% and 0.63%, respectively (relative effectiveness, 0.5%; 95.2% CI, −8.6 to 8.8); for cardiorespiratory disease in 2.25% and 2.38% (relative effectiveness, 5.7%; 95.2% CI, 1.4 to 9.9); and for any cause in 9.38% and 9.58% (relative effectiveness, 2.1%; 95.2% CI, –0.1 to 4.3). Death from any cause occurred in 0.67% and 0.66% (relative effectiveness, −2.5%; 95.2% CI, −11.6 to 5.9). The incidence of serious adverse events, therefore, was clearly similar in the two groups.
So to summarize, the GALFUL trial claimed “fewer hospitalizations” with higher dose vaccines in its headline, yet at the same time noted that serious events did not differ between the two dosages. The DANFLU-2 TRIAL really found absolutely no difference in outcomes. Combined, the answer appears quite clear: Somewhat counterintuitively,“high-dose” vaccines in older people appear to be a waste!
But hold on for a moment, because then there was, of course, also the editorial.3 And—besides the fact that neither the author of the editorial nor the editors of The Journal, of course, noticed the contradictory conclusions in the GALFUL trial—the editorial suddenly disclosed that the main purpose of these two papers—or why these papers (of not great quality as we will be discussing in a moment) were likely accepted was not the purpose that both had advertised (to determine whether high-dose flu vaccines are better in older people than a standard dose), even though that real purpose was never really noted by the authors of the two papers.
The real purpose for publication of these two papers was apparently the recognition that even randomized, placebo-controlled trials, correctly, still considered the gold standard in determining causal relationships between an intervention and its outcome, however, also have shortcomings and, for a variety of reasons (including at times cost), are often simply not feasible. Alternative methods of study design, therefore, can be important adjuncts to traditional trials, a point we have been making innumerable times over decades in these pages.
The editorial then noted that both trials were examples of less costly alternative designs to traditional placebo-controlled studies. The central argument the editorial basically made was that large scale real life studies, even, for example, if not controlled for co-existing medical conditions, can still offer valuable data. To a degree that is, of course, correct, and medical practice is, indeed, in most medical specialties, in a large majority, dependent on exactly such study data; but to claim that such studies may come even close to non-inferiority in comparison to prospectively, randomized, well-controlled studies is pure nonsense.
The two studies discussed here, indeed, demonstrate this fact well: It is almost impossible to purposefully design two almost identical studies of such large patient populations as these two studies represented. Moreover, both study populations were Western European (Spain and Denmark) and, therefore, not only likely more genetically homogenous than such studies, for example, would be in the U.S., yet even within Europe, it matters what the genetic background of a study population is. Danes are clearly different from Spaniards.
We, therefore, would suggest that we take the two original papers for what they were meant to be, studies to determine whether high-dose flu vaccines in older people after age 65 make sense. It appears that they don’t. But if we really want to find out for sure, we still need a prospectively randomized study!
References
1. Pardo-Seco et al., N Engl J Med 2025; 393:2303-2312
2. Johansen et al., N Engl J Med 2025;393:2291-2302
3. Neuzil KM. N Engl J Med 2025;393(23):2361-2362