To the Editor: In their article about the effectiveness of influenzavaccination in the community-dwelling elderly (Oct. 4 issue),1Nichol and colleagues define the influenza season as the timefrom the reporting of the first isolate to the reporting ofthe last isolate, for each region and season. We are concernedthat their analysis included deaths early and late in the seasonthat were almost certainly unrelated to influenza and that couldnot have been prevented by vaccination. The actual period ofelevated risk for influenza-related death is substantially shorterthan the interval between the reporting of the first isolateand that of the last isolate. In two West Coast cities, theperiod of predominant influenzavirus transmission accountedfor less than half of the total time between the reporting ofthe first and last seasonal isolates.2 The same is true in Wisconsin,where we have prospectively tested patients in a population-basedcohort over three seasons. Nichol et al. should consider additionalanalyses to determine whether the benefit of vaccination withrespect to mortality varies over time within each influenzaseason. If the effect is real, the benefit should be greatestduring the 4-to-8-week period of maximum virus circulation andshould be much lower at the beginning and end of the season.
Edward A. Belongia, M.D. Laura A. Coleman, Ph.D. James G. Donahue, D.V.M., Ph.D. Marshfield Clinic Research Foundation Marshfield, WI 54449 belongia.edward{at}marshfieldclinic.org
References
Nichol KL, Nordin JD, Nelson DB, Mullooly JP, Hak E. Effectiveness of influenza vaccine in the community-dwelling elderly. N Engl J Med 2007;357:1373-1381. [Free Full Text]
Izurieta HS, Thompson WW, Kramarz P, et al. Influenza and the rates of hospitalization for respiratory disease among infants and young children. N Engl J Med 2000;342:232-239. [Free Full Text]
To the Editor: Nichol et al. report that elderly persons receivinginfluenza vaccine have lower risks of death and hospitalizationfor pneumonia or influenza than nonvaccinated elderly personsduring the influenza season. In an 8-year study of a similarpopulation of members of a health maintenance organization,we found risk reductions among vaccinated elderly persons duringthe influenza season to be essentially identical to those reportedby Nichol et al. (Table 1).1 However, we also found even greaterreductions before the influenza season. During that period,vaccination is not expected to have an effect, so any apparentvaccine benefit represents bias due to the preferential useof vaccine by healthier elderly persons. Instead of using ahypothetical-unmeasured-confounder model, which in their studyincluded restricted and weak assumptions about the strengthof confounding2,3 and did not consider multiple confounders,4Nichol et al. should have evaluated the actual influence ofbias in their study by calculating the relative risks duringall the noninfluenza control periods, thus filling in the blanksin Table 1. In the absence of such analyses, and given the existingevidence of extensive bias,1 the results reported by Nicholet al. should be interpreted cautiously.
Table 1. Relative Risk of Death or Hospitalization among Vaccinated Elderly Persons vs. Unvaccinated Elderly Persons in the Studies by Nichol et al. and Jackson et al., According to Period.
Jennifer C. Nelson, Ph.D. Michael L. Jackson, Ph.D., M.P.H. Lisa A. Jackson, M.D., Ph.D. Group Health Center for Health Studies Seattle, WA 98101 nelson.jl{at}ghc.org
Dr. Jackson reports receiving grant support from GlaxoSmithKline,Sanofi Pasteur, and Novartis; consulting fees from Sanofi Pasteur;lecture fees from MedImmune; and advisory fees from Novartis.No other potential conflict of interest relevant to this letterwas reported.
References
Jackson LA, Jackson ML, Nelson JC, Neuzil KM, Weiss NS. Evidence of bias in estimates of influenza vaccine effectiveness in seniors. Int J Epidemiol 2006;35:337-344. [Free Full Text]
Walter LC, Brand RJ, Counsell SR, et al. Development and validation of a prognostic index for 1-year mortality in older adults after hospitalization. JAMA 2001;285:2987-2994. [Free Full Text]
Jackson LA, Nelson JC, Benson P, et al. Functional status is a confounder of the association of influenza vaccine and risk of all cause mortality in seniors. Int J Epidemiol 2006;35:345-352. [Free Full Text]
Fewell Z, Davey Smith G, Sterne JAC. The impact of residual and unmeasured confounding in epidemiologic studies: a simulation study. Am J Epidemiol 2007;166:646-655. [Free Full Text]
To the Editor: Nichol et al. did not succeed in eliminatingfrailty bias in their observational studies. As argued previously,1,2given that only approximately 5% of all wintertime deaths amongelderly persons are attributable to influenza, the conclusionthat influenza vaccination can prevent half of all wintertimedeaths is simply not plausible. Frailty bias is the most likelyexplanation of this result.
Application of four criteria from our recently published bias-detectionframework2 (vaccine match, season severity, age, and outcomespecificity) indicates profound bias in the data presented byNichol et al. (Table 1). The authors do not present the dataneeded to apply our crucial fifth criterion, seasonality.2,4If vaccination is found to be effective in the weeks beforeeach influenza epidemic starts, bias must be present.2,4 Theauthors' postepidemic analysis is inadequate, because the biasdecreases over time. We urge the authors to report vaccine-effectivenessestimates for the period before the influenza season or forthe very beginning of the season, which would clarify the levelof bias in their studies.
Table 1. Application of a Bias-Detection Framework to Estimates of Vaccine Effectiveness Presented by Nichol et al.
Lone Simonsen, Ph.D. George Washington School of Public Health Washington, DC 20037 lone{at}gwu.edu
Cecile Viboud, Ph.D. Fogarty International Center Bethesda, MD 20892
Robert J. Taylor, Ph.D. SAGE Analytica Bethesda, MD 20814
References
Simonsen L, Reichert TA, Viboud C, Blackwelder WC, Taylor RJ, Miller MA. Impact of influenza vaccination on seasonal mortality in the US elderly population. Arch Intern Med 2005;165:265-272. [Free Full Text]
Simonsen L, Taylor RJ, Viboud C, Miller MA, Jackson LA. Mortality benefits of influenza vaccination in elderly people: an ongoing controversy. Lancet Infect Dis 2007;7:658-666. [CrossRef][Web of Science][Medline]
Thompson WW, Shay DK, Weintraub E, et al. Mortality associated with influenza and respiratory syncytial virus in the United States. JAMA 2003;289:179-186. [Free Full Text]
Jackson LA, Jackson ML, Nelson JC, Neuzil KM, Weiss NS. Evidence of bias in estimates of influenza vaccine effectiveness in seniors. Int J Epidemiol 2006;35:337-344. [Free Full Text]
To the Editor: Influenza vaccination has been found to reducethe risk of influenza in the elderly.1 Nichol et al. reportthat influenza vaccination was associated with a 48% reductionin the risk of death from any cause among community-dwellingelderly persons. However, seasonal influenza has been reportedto result in a substantially smaller percentage of deaths fromany cause among the elderly.2,3,4 How can influenza vaccineprevent a much larger percentage of deaths than are caused bythe disease that the vaccine is supposed to prevent?
M. Miles Braun, M.D., M.P.H. Hector S. Izurieta, M.D., M.P.H. Robert Ball, M.D., M.P.H. Food and Drug Administration Rockville, MD 20852 miles.braun{at}fda.hhs.gov
References
Govaert TM, Thijs CT, Masurel N, Sprenger MJ, Dinant GJ, Knottnerus JA. The efficacy of influenza vaccination in elderly individuals: a randomized double-blind placebo-controlled trial. JAMA 1994;272:1661-1665. [Free Full Text]
Simonsen L, Taylor RJ, Viboud C, Miller MA, Jackson LA. Mortality benefits of influenza vaccination in elderly people: an ongoing controversy. Lancet Infect Dis 2007;7:658-666. [CrossRef][Web of Science][Medline]
Thompson WW, Shay DK, Weintraub E, et al. Mortality associated with influenza and respiratory syncytial virus in the United States. JAMA 2003;289:179-186. [Free Full Text]
Miniño AM, Arias E, Kochanek KD, Murphy SL, Smith BL. Deaths: final data for 2000. Natl Vital Stat Rep 2002;50:1-119. [Medline]
The authors reply: Belongia et al. recommend analysis of peripeakinfluenza data. Such analyses, even if possible, might be misleadingfor bias detection. We primarily relied not on community-specificsurveillance but on regional surveillance that was passive withuneven coverage. Within regions, peaks vary widely. Also, severecomplications can occur weeks after the peak. Peak analysesexclude some outcomes and will result in lower absolute riskreductions but not always higher relative risk reductions, ascompared with other periods. In a study of healthy adults, theeffectiveness of influenza vaccine, measured as the relativerisk reduction, for absenteeism due to respiratory illness wasnearly identical for the peak influenza period, the total influenzaperiod, and the total outcome period.1
Nelson et al. and Simonsen et al. suggest the analysis of datafrom the preinfluenza season. However, analyses of mortalitybefore the influenza season can introduce biases.2 To minimizebias, our subjects had to be alive on day 1 of the influenzaseason. A 2-year longitudinal analysis of more than 120,000subjects showed significant vaccine effectiveness against hospitalizationfor pneumonia or influenza during the 1998–1999 and 1999–2000influenza seasons but not the 1999 and 2000 summer seasons (unpublisheddata), findings that confirm the implications of our originalsummer analyses and argue against bias.
In their table, Simonsen et al. report the results of the applicationof their bias-detection framework3 to our data. The source ofthe effectiveness estimates for hospitalization for pneumoniaor influenza for good-match influenza vaccine as compared withpoor-match vaccine that they ascribe to our study is unclear.The actual vaccine-effectiveness estimate for good-match vaccineas compared with poor-match vaccine (29% [95% confidence interval,24 to 34] vs. 25% [15 to 34]) does not support their conclusion.In addition, our summer analyses (relevant to their seasonalitycriterion) do not support the claim of bias. Finally, limitationsin the assumptions restrict their framework's usefulness (Table 1).
Table 1. Limitations of the Bias-Detection Framework.
Braun et al. ask about reconciling our vaccine-effectivenessestimates with excess-death estimates. Excess-death estimatesdepend on the models used. These estimates are imprecise andindirectly measured and may exclude deaths from influenza thatcontribute to the "baseline" death rates. The total number ofinfluenza-associated deaths is unknown. In addition, excess-deathestimates apply to the entire U.S. population. Our results maynot be generalizable to all elderly. For instance, nursing-homeresidents, who make up less than 5% of the elderly in the UnitedStates but who account for more than 30% of the deaths, werenot included in our study.
Concerns about residual confounding are important. Our sensitivityanalyses modeled a prevalent confounder associated with an increasedrisk of an outcome that is at least as high as that seen withhigh-risk status or previous hospitalization. As expected, theresulting vaccine-effectiveness estimates were lower than inthe overall analysis, but they were still significant. An additionalmultiple-confounder sensitivity analysis confirmed that onlya large and unlikely combination of effects would eliminatevaccine effectiveness with respect to death: multiple independentconfounders, combined to create a single normal confounder moderatelyassociated with vaccination (1/2 SD between the vaccinated andunvaccinated groups), would have to have an association withdeath that was more than 2.35 times that with age for vaccineeffectiveness to be reduced to 0%. Since age is generally thestrongest predictor of death, together with the other predictorsalready included in our models, this is a large associationto be required for missing confounders. Despite uncertaintiesabout precise levels of benefit, our conclusions are robust:influenza vaccination in community-dwelling elderly personscan prevent hospitalization and death.
Kristin L. Nichol, M.D., M.P.H. Veterans Affairs Medical Center Minneapolis, MN 55417 nicho014{at}umn.edu
James D. Nordin, M.D., M.P.H. HealthPartners Research Foundation Minneapolis, MN 55440
Eelko Hak, Ph.D. University Medical Center Utrecht 3508 AB Utrecht, the Netherlands
References
Nichol KL, Mendelman P. Influence of clinical case definitions with differing levels of sensitivity and specificity on estimates of the relative and absolute health benefits of influenza vaccination among healthy working adults and implications for economic analyses. Virus Res 2004;103:3-8. [CrossRef][Web of Science][Medline]
Hak E, Hoes AW, Nordin J, Nichol KL. Benefits of influenza vaccine in US elderly -- appreciating issues of confounding bias and precision. Int J Epidemiol 2006;35:800-802. [Free Full Text]
Simonsen L, Taylor RJ, Viboud C, Miller MA, Jackson LA. Mortality benefits of influenza vaccination in elderly people: an ongoing controversy. Lancet Infect Dis 2007;7:658-666. [CrossRef][Web of Science][Medline]
Ohmit SE, Victor JC, Rotthoff JR, et al. Prevention of antigenically drifted influenza by inactivated and live attenuated vaccines. N Engl J Med 2006;355:2513-2522. [Free Full Text]
Hannoun C, Megas F, Piercy J. Immunogenicity and protective efficacy of influenza vaccination. Virus Res 2004;103:133-138. [CrossRef][Web of Science][Medline]
Previous
