Background. Uncertainties regarding influenza disease impact and benefits of vaccination may contribute to low vaccination rates among adults aged 50–64 years.
Methods. This prospective cohort study assessed the burden of influenza-like illness (ILI) among working adults aged 50–64 years and the effectiveness of influenza vaccination in reducing the rate of ILI and productivity losses. Employees of the University of Minnesota (Minneapolis) were invited via e-mail to participate in the study during October 2006. The study data were collected using internet-based surveys at baseline (October 2006) and during the follow-up period (from November 2006 through April 2007). Months included in the 2006–2007 influenza season were identified retrospectively from Minnesota Department of Health surveillance data. Vaccine effectiveness for reducing the rate of ILI, ILI-associated health care use, the number of days of illness, work loss, and reduced on-the-job productivity during the influenza season were assessed using multivariable regression models after controlling for important confounders.
Results. Four hundred ninety-seven persons were included in the study, 85 (17.1%) of whom experienced an ILI. Among unvaccinated participants, ILI was responsible for 45% of all days of illness during the influenza season, 39% of all illness-related work days lost, and 49% of all days with illness-related reduced on-the-job productivity. In the multivariable regression analyses, vaccination was associated with a significant reduction in the rate of ILI (adjusted odds ratio, 0.48; 95% confidence interval, 0.27–0.86) and fewer days of illness, absenteeism, and impaired on-the-job performance.
Conclusion. ILIs were common among our study participants, accounting for a large portion of illness, work loss, and impaired work performance during the influenza season. Vaccination was associated with substantial health and productivity benefits. Vaccine delivery should be improved for this high-priority group.
Influenza continues to be an important cause of morbidity and mortality. Each year, 5%–20% of the US population becomes ill [1, 2]. In the elderly population, the serious complications of influenza include excess hospitalizations  and deaths . In the younger adult population, important manifestations of influenza include morbidity that results in restricted activity, work absenteeism, impaired work performance, and increased health care use .
Beginning with the 2000–2001 influenza season, all persons aged 50–64 years have been included as a group targeted for annual vaccination in the United States [6, 7]. By 2006, however, the vaccination rate among this group was only 33.1% (44.4% for persons aged 50–64 years who were at high risk for serious complications from influenza and 28.2% for persons aged 50–64 years who were at low risk for serious complications from influenza), a rate well below the 2010 national health goal of 60% . Few previous studies have assessed influenza-associated morbidity and the effectiveness of vaccination in this population, and uncertainties regarding the impact of illness and benefits of vaccination may, therefore, contribute to low vaccination rates, especially in light of recent seasons with vaccine shortages and delays, recommendations for tiered vaccine delivery, and difficulties faced by providers and their patients with regard to ensuring priority vaccination of groups at high risk for serious complications from influenza [9, 10]. We conducted this prospective cohort study to clarify the burden of influenza-like illness (ILI) and the benefits of vaccination among adults aged 50–64 years.
During October 2006, e-mail invitations to participate in this study were sent to University of Minnesota (Minneapolis) employees. Interested employees were directed to a secure Web site to learn more about the study. Consenting employees then completed a baseline questionnaire that asked about participants' demographic and health characteristics. Follow-up information was obtained about occurrences of illness, associated symptoms, impact on work, and health care use for each month from November 2006 through April 2007. E-mail reminders for the participants to complete the internet-based follow-up surveys were sent during the week after each study month. Information on influenza vaccination status was obtained for the 2006–2007 season in the final follow-up survey to ensure complete capture of vaccination status for the 2006–2007 season (e.g., for persons who might have been vaccinated in December or after). Persons who completed baseline and all follow-up surveys were eligible for a random drawing for a $50 book store gift certificate per every ∼250 participants. The study was reviewed and approved by the human subjects committees of the University of Minnesota and the Minneapolis Veterans Affairs Medical Center.
Outcomes that occurred during months for which influenza activity was local, regional, or widespread according to influenza surveillance data from the Minnesota Department of Health were included in the influenza season analyses.
Baseline characteristics of vaccinated and unvaccinated participants were compared using χ2 and Student's t tests. Multivariable regression models were constructed to assess vaccine effectiveness for reducing outcomes after controlling for important covariates. The outcomes included occurrences of ILI (defined as fever or feverish with cough or sore throat) and ILI-associated provider visits, as well as the numbers of days of illness, disability, work absenteeism, and working while ill.
To assess for evidence of bias, we compared the risk of non–upper respiratory illness during the influenza season among vaccinated participants with that among unvaccinated participants; we also calculated the risk of ILI during the non–influenza months included in the study. For these outcomes, we did not expect vaccination to be associated with a reduced risk of illness. Vaccine effectiveness analyses were weighted for the number of months of follow-up data available for each participant during the influenza season and during the non–influenza period, because some participants did not complete all 6 of the follow-up surveys.
Of the 2151 university employees who volunteered for the study, 563 were aged 50–64 years; 497 of these persons completed baseline questionnaires, provided information on their vaccination status, and were included in this study. These participants completed 92% of the follow-up surveys.
Overall, 404 participants (81%) reported receiving influenza vaccine for the 2006–2007 season. At baseline, vaccinated persons showed a trend toward having poorer health, being more likely to have functional limitations and significant chronic medical conditions, and being more likely to take prescription medications, compared with unvaccinated persons, although none of the differences were statistically significant (table 1). Unvaccinated persons were somewhat more likely to be current cigarette smokers than were vaccinated persons, but the difference was not statistically significant.
The 2006–2007 influenza season in Minnesota extended from January 2007 through March 2007, months during which consecutive weeks demonstrated local, regional, or widespread influenza activity. These dates also corresponded with the peaks seen with school outbreaks and hospitalizations during the influenza season in Minnesota. The predominant circulating viruses for that season were type A, subtype H1N1 viruses that were well matched to the vaccine .
Of the 497 study participants, 85 (17.1%) reported an ILI during the influenza season. Morbidity associated with these illnesses is shown in table 2. Persons with ILI were sick for ∼8 days, missed ∼1.5 days of work because of illness, and worked for >4 days while still symptomatic. Twenty-six (31%) of participants with ILI visited a health care provider, and 20 (24%) received prescription antibiotics. ILI in unvaccinated persons appeared to be more severe than ILI in vaccinated persons, according to most of the parameters studied, but the differences were not always statistically significant (table 2). The median level of work effectiveness reported by participants for the days that they worked while ill was 70%–75% (25th percentile, 50%; 75th percentile, 80%). When the 85 participants who had an ILI were asked from whom they thought they contracted their illness, 38 (45%) indicated someone at work, 14 (16%) indicated someone in their household, 5 (6%) indicated someone at a social or public gathering, and 28 (33%) indicated an other or unknown source.
In figure 1, the contributions of ILI to total days of illness, work loss because of illness, and working while ill during the influenza season are shown for unvaccinated persons. ILI was responsible for 45% of all days of illness, 39% of all work days lost because of illness, and 49% of days of working while ill.
Forward and backward selection procedures were used to identify variables to include in the multivariable regression models. Selection criteria included overall contribution to the model, parsimony, and consistency across outcomes. The final regression models included sex, cigarette smoking status, general health level (fair to poor vs. good to excellent), having ⩾1 high-risk medical condition, impaired functional status (defined as at least some difficulty walking one-quarter mile; standing for 2 h; stooping, bending, or kneeling; or lifting or carrying something weighing 10 lb [4.5 kg]), number of activity-limited days during the previous month, and influenza vaccination status for the 2006–2007 season. Logistic regression and general linear regression analyses were performed using SPSS, version 13.0, for Windows (SPSS). Influenza vaccination was associated with a significant reduction in the risk of ILI (adjusted OR, 0.48; 95% CI, 0.27–0.86) (table 3). Because the rate of ILI was >10%, this OR underestimates the corresponding relative risk. Based on the method of Zhang and Yu , we estimated a corresponding relative risk of 0.55 (95% CI, 0.33–0.89) and a vaccine effectiveness (calculated as 1 minus the relative risk) of 45%. Vaccination was also associated with statistically significant reductions in the numbers of days of illness, days in bed, days with significant decreases in ability to do usual activities, days of work lost, and days of working while ill (table 4).
In our analyses to explore for evidence of bias in our models, we found that vaccination was not associated with a decreased risk of non–upper respiratory tract illness during the influenza season or a reduced risk of ILI during the non–influenza periods (table 5). Thus, we did not find evidence for selection bias in our data.
In this study, we revealed that ILI occurred frequently and was associated with substantial morbidity among working adults aged 50–64 years who participated in our study. Among unvaccinated participants, ILI was associated with 45% of days of illness due to all causes, 39% of work days lost due to all causes, and 49% of days of working while ill due to all causes during the influenza season. With vaccination, we demonstrated a substantial reduction in the risk of ILI of ∼45% and reductions of ⩾60% in the numbers of days of illness, days of work lost, days of working while ill, and days in bed because of ILI.
A systematic review of influenza vaccine for healthy adults generally aged <65 years found that the effectiveness of inactivated influenza vaccine during years with a good vaccine strain–circulating strain match was 30% (relative risk, 0.70; 95% CI, 0.59–0.83) against clinical ILI without laboratory confirmation and 42% (relative risk, 0.58; 95% CI, 0.37–0.91) against ILI-associated physician visits . Vaccination was also associated with a mean reduction in the number of days of illness of 0.48 days per person vaccinated (95% CI, 0.34–0.62) and a mean reduction in the number of working days lost of 0.21 days per person vaccinated (95% CI, 0.36–0.05). Another meta-analysis assessed vaccine efficacy in studies that used a clinical case definition of febrile respiratory illness during the peak of the influenza season—a case definition that is similar to the one used in our study—and found that vaccination was associated with a 54% reduction in the rate of clinical ILI . Most of the studies included in the meta-analyses involved working adults aged <50 years. Our study results are similar to the results in those studies and extend previous findings by demonstrating that working adults aged 50–64 years (both healthy adults and those at high risk for serious complications from influenza) can experience levels of benefit from vaccination that are consistent with those previously seen in healthy working adult populations in general.
A recent study estimated the annual economic burden of influenza in the United States to be $87 billion; lost productivity because of work absenteeism and premature mortality comprised the bulk of the economic burden . However, the costs associated with impaired on-the-job productivity (known as presenteeism) are other important contributors to the total economic burden caused by illness. Presenteeism accounted for 18%–60% of all costs for the top 10 health conditions affecting US employers  and for approximately two-thirds of the total lost productivity costs related to the common cold . Presenteeism, in addition to absenteeism, is undoubtedly a major contributor to the economic burden associated with influenza. Influenza B virus infection in healthy adults was found to impair their ability to perform certain tasks to a level similar to that seen with sleep deprivation or alcohol consumption . ILI has also been associated with impaired school performance among college and university students  and impaired productivity while at work among healthy working adults [20, 21]. In our study, unvaccinated persons experienced a mean of 2.02 days of working while ill (i.e., presenteeism days) and a mean of 0.57 days of work lost because of ILI over the course of the influenza season. Vaccination prevented a mean of 0.37 days of work lost and a mean of 1.45 days of working while ill per each person vaccinated. With a self-reported level of effectiveness of only 75% of normal on those days spent working with an ILI, the number of presenteeism days avoided among vaccinated persons represents an approximately equal productivity gain to the 0.37 absenteeism days avoided (i.e., [100%-75%]×1.45=0.36 days at 100% effectiveness). Studies that fail to account for presenteeism in their estimates of the economic burden of influenza may substantially underestimate the impact of illness and the productivity benefits associated with vaccination.
Recent studies have highlighted the potential role of school children in the transmission of influenza within communities [22–24] and the role of health care workers in the transmission of influenza in health care settings [25, 26], whereas the role of coworkers in the transmission of influenza in the workplace has not been well described. In 1 national consumer survey from 2006, 30% of working adults reported that they had contracted the influenza virus from a coworker . In our study, 45% of participants with an ILI thought that they had contracted their illness at work, whereas only 16% thought their illness was the result of a household contact. These findings suggest that transmission of influenza in the workplace is likely to be important and should be studied further.
More than one-half of the vaccinated participants in our study were vaccinated at their workplace. This is consistent with national trends and the role of nontraditional settings in vaccine administration . Many employers have workplace-based influenza vaccination programs . Such programs are often well received and provide a safe and convenient venue for vaccine administration . Because of the low national vaccination rates among persons aged 50–64 years, vaccination in both traditional and nontraditional settings should be strongly encouraged.
Functional status may be a potential confounder in observational studies of influenza vaccination effectiveness in the elderly population . We included several measures of health-related quality-of-life and functional status in our study. These generally included health questions and questions about activity-limited days during the previous month that were adapted from the Centers for Disease Control and Prevention's health-related quality-of-life questions used in the Behavioral Risk Factor Surveillance System surveys ; we also included questions about difficulties with selected physical activities (e.g., walking one-quarter mile, standing 2 h, stooping or bending, and lifting of carrying something weighing 10 lb) that were adapted from the Centers for Disease Control and Prevention's National Health Interview surveys . Therefore, we were able to include information from these questions in our multivariable models. Of interest, vaccinated persons tended to have poorer health status and functional status, compared with unvaccinated persons.
Our study has several limitations. Results from observational studies, including cohort studies such as this one, may be misleading because of bias and residual confounding. We attempted to measure important variables and adjusted for them in our multivariable models. We also tested for and did not find evidence of a selection bias between vaccinated and unvaccinated participants. Our estimates of vaccine effectiveness are also similar to previously published results. Nevertheless, our findings should be interpreted with caution. In addition, our participants may not be representative of all working adults aged 50–64 years. Our participants were University of Minnesota employees. The vaccination rate among our participants was higher than nationally reported vaccination rates among adults aged 50–64 years. In addition, our participants may have been somewhat healthier than adults aged 50–64 years in general. For example, our participants were less likely to be current smokers (3.4% vs. 14.2% ) and to report fair-to-poor health (5.3% vs. 11.0%), compared with Minnesota adults aged 55–64 years . Thus, it is not clear how our findings might translate to other types of workers or workers in other settings. In addition, our data were obtained by self-report through monthly internet-based surveys. Self-report of influenza vaccination status [35, 36] and symptoms of infectious illness, including respiratory illnesses , has been shown to be acceptably accurate and valid, and monthly follow-up intervals have been used in other studies assessing occurrences of ILI. Monthly follow-up intervals may, however, result in the underestimation of actual absenteeism rates . We also used a simple, self-reported measure of work effectiveness for those days that participants worked while still ill that was selected for simplicity and brevity. Although other authors have used similar methods , it is unclear how our results might correlate with more-objective measures of impaired productivity. Finally, our study was a single-season study with a good match between vaccine strains and circulating viruses. Because of the variability from year to year with regard to circulating viruses and the degree of similarity between vaccine and circulating strains, vaccine effectiveness will undoubtedly vary from year to year.
Recent health economic studies have suggested that influenza vaccination of persons aged 50–64 years may be cost effective in the United States , the United Kingdom , and other countries [41, 42]. Our study provides additional evidence of the substantial disease burden associated with ILI among working adults in this age group and the substantial benefits that can be achieved with vaccination. Efforts should be enhanced to improve vaccination rates in this high-priority group.
Financial support. GlaxoSmithKline Biologicals to the Minnesota Veterans Research Foundation.
Potential conflicts of interest. K.L.N. has received research funding from Sanofi Pasteur and has served on medical advisory boards or as a consultant for GlaxosmithKline Biologicals, Sanofi Pasteur, MedImmune, Novartis, and CSL Behring. M.E.G. is an employee of GlaxoSmithKline Biologicals. S.J.D'H. and E.E.: no conflicts.