Should we question the benefits of influenza vaccination for the elderly?

August 2005

Lone Simonsen, PhD, and colleagues at the National Institutes of Health published a provocative article questioning the benefits of annual influenza vaccination of elderly people in the United States. She based her conclusions on two apparently conflicting observations. First, the influenza vaccination coverage rate in the elderly increased from about 15% in 1980 to about 65% in 2001, and second, despite reports that influenza vaccination reduces all-cause winter season mortality by 30% to 50%, influenza-related mortality among the elderly actually increased during the same period. What is one to make of these observations? Can they be reconciled? Are her conclusions justified?

Simonsen and colleagues used national mortality data and a modified Serfling-type regression model to estimate annual winter season excess mortality during the period from 1968 to 2001. She adjusted the age of her study population to the U.S. population in 1970 and separately estimated the excess mortality rate for seasons dominated or not dominated by influenza A (H3N2) viruses. She found that excess influenza-related all-cause mortality accounted for only 5% to 10% of all winter season deaths. In people 65 to 74 years of age, the excess mortality rate in the H3N2 seasons declined from 1968 (the year of the Hong Kong H3N2 pandemic) to 1980 and remained stable, thereafter. She and her colleagues concluded that this was due to persisting immunity following earlier natural H3N2 infections during the 1968 to 1980 period when people in this age group were healthier and middle-aged.

In contrast, among people older than 85 years of age, the excess mortality rate remained constant during the decade following the 1968 pandemic year and then tended to increase over time. In this much older age group, long-lasting anti-H3N2 immunity acquired following childhood H3N2 infection before 1892 was protective during the initial years after 1968. In this age group, researchers concluded the overall response to vaccination throughout the entire study period was poor.

It is important to recognize the several strengths of researchers’ analysis. Nonetheless, how should the results of her analysis be interpreted? Simonsen and colleagues observed, “there are not enough influenza-related deaths to suggest the conclusion that vaccination can reduce total winter mortality among the U.S. elderly population by as much as half.” She added, “if vaccination reduces influenza-related mortality by 70% to 80%, then a 50-percentage point increase in vaccination coverage among the elderly after 1980 should have reduced both excess probe and irrigation and excess all-cause mortality by 35% to 40%. We found no evidence (that this) had occurred.” She concluded, “observational studies must overstate the mortality benefits of the vaccine” and “this vast disconnect between conclusions from different studies must be sorted out.”

Simonsen and colleagues’ conclusions are open to question for at least three reasons: she did not consider the “ecologic fallacy;” she did not acknowledge that in most elderly people, influenza vaccination effectiveness in reducing mortality does not decrease with increasing age; and she did not consider the differences between her study populations and those in which vaccination effectiveness has been evaluated.

Ecologic fallacy

Epidemiologists recognize that ecologic studies evaluate data primarily in the aggregate and cannot provide estimates of effects in individuals, especially when there are heterogeneity of exposure and other covariates. This bias is widely known as the ecologic fallacy. Simonsen and colleagues mention one such bias, namely that people who are discharged from hospital during the fall vaccination season are at greatly increased risk of being readmitted or dying of an influenza-related illness during the subsequent winter season compared with nondischarged people, yet they may be less likely to be vaccinated. During the years of the study, however, there were undoubtedly other substantial differences in the vaccination rate by geography, race and socioeconomic status. Recently, these differences have been especially well documented.

Simonsen and colleagues argued that a declining response to influenza vaccination with increasing age might explain her results, and she cited several studies showing decreased serologic responsiveness and reduced protection against influenza illness following vaccination. However, a direct measure of the influence of age on mortality reduction following vaccination has been reported for people 65 to 74, 75 to 84 and older than 85 during four successive winter seasons. Researchers did not observe any differences in vaccination effectiveness among the three age groups (table 1). Researchers also noted an almost 25-fold increased risk of dying among unvaccinated high-risk people older than 85 compared with healthy people 65 to 74. The authors of this study concluded, “as the proportion of persons 85 years of age and older increases, we can expect to see an increasing absolute number of deaths due to influenza. This increase is due at least in part to an increasing risk of dying influenced both by advancing age and increasing burden of underlying illness and, not primarily, due to a decrease in influenza vaccine effectiveness.” These results were obtained in a study of community-dwelling elderly people and not in a population that included all elderly people.

Different study populations

Simonsen and colleagues studied winter season mortality for the entire elderly population, including those living in nursing homes.

Virtually all of the observational studies that have shown reductions in influenza-related mortality following vaccination have been conducted among the community-dwelling elderly and have excluded nursing home residents. This difference is of critical importance. For example, in 1997, nursing home residents represented less than 5% of the U.S. elderly population, and yet they accounted for approximately one-third of all deaths that occurred that year (table 2). Among people who were older than 85, a group that in Simonsen and colleagues’ study accounted for 44% of influenza-related deaths among the elderly in the 1990s, approximately half of all deaths occurred among nursing home residents. In all likelihood, the proportion of deaths occurring in people who did not live in the community was even greater, because many nursing home residents who were discharged to hospitals probably died in hospital without returning to live in the community. Whether the results of observational studies conducted among community-dwelling elderly “overstate the mortality benefits of the vaccine” among all elderly people (including nursing home residents) may be unclear. What is clear, however, is that there is no “vast disconnect between conclusions from different studies” on the effectiveness of influenza vaccination. The “vast disconnect” that Simonsen and colleagues speak of is not in the estimates of vaccination effectiveness itself, but in the different populations researchers have studied.

Greater efforts to improve the vaccination rate for the elderly, including eliminating disparities in the vaccination rate among different groups, will help prevent more influenza-related hospitalizations and deaths. Nonetheless, whatever the “obvious implications for influenza vaccination policy” of Simonsen’s results might be, we should not doubt the benefits of current policy to vaccinate all elderly people, over 95% of whom still live in the community.

For more information:

• Cohen J. Influenza. Study questions the benefits of vaccinating the elderly. Science. 2005;307(5712):1026.
• Fedson DS, Wadja A, Nicol JP, et al. Disparities between influenza vaccination rates and risk for influenza-associated hospital discharge and death in Manitoba in 1982-1983. Ann Intern Med. 1992;116(7):550-555.
• Gabrel CS. Characteristics of elderly nursing home current residents and discharges: data from the 1997 National Nursing Home Survey. Adv Data. 2000;25(312):1-15.
• Greenland S. Ecologic versus individual-level sources of bias in ecologic estimates of contextual health effects. Int J Epidemiol. 2001;30(6):1343-1350.
• Hoyert DL, Kochanek KD, Murphy SL. Deaths: final data for 1997. Natl Vital Stat Rep. 1999;47(19):1-104.
• Lemon SC, Rakowski W, Clark MA, et al. Variations in influenza vaccination among the elderly. Am J Health Behav. 2004;28(4):352-360.
• Nichol KL, Nordin J, Mullooly J, et al. Influence of advancing age on influenza vaccination effectiveness among community dwelling elderly. In: Kawaoka Y, ed. Options for the control of influenza V. Amsterdam: Elsevier; 2004: 98-100.
• Reichert TA, Sugaya N, Fedson DS, et al. The Japanese experience with vaccinating schoolchildren against influenza. N Engl J Med. 2001;344(12):889–896.
• Simonsen L, Reichert TA, Viboud C, et al. Impact of influenza vaccination on seasonal mortality in the U.S. elderly population. Arch Intern Med. 2005;165(3):265-272.