Association between the first influenza vaccination (1st vac) or revaccination (revac) and lower respiratory tract infection during influenza epidemic periods, stratified by year. Adjustment has been made for sex, underlying chronic diseases (cardiovascular and respiratory diseases; neurological and psychiatric disorders; hypertension; malignancies; renal insufficiency; and diabetes mellitus), smoking, number of antibiotic prescriptions, and number of visits to a general practitioner. On the x-axis, the first vaccination, any revaccination, and revaccination vs first vaccination are all fully adjusted. On the y-axis, the hazard ratio (HR) indicates the risk of lower respiratory tract infection following first vaccination or revaccination vs no vaccination, or following revaccination vs first vaccination. Error bars indicate 95% confidence intervals (CIs).
Voordouw BCG, Sturkenboom MCJM, Dieleman JP, Stijnen T, van der Lei J, Stricker BHC. Annual Influenza Vaccination in Community-Dwelling Elderly Individuals and the Risk of Lower Respiratory Tract Infections or Pneumonia. Arch Intern Med. 2006;166(18):1980-1985. doi:10.1001/archinte.166.18.1980
Influenza vaccination has been associated with a reduction in the number of hospitalizations for respiratory conditions in elderly persons over the period from 1996 to 2002. Little is known, however, about the effect of influenza vaccination on the whole range of severity of respiratory tract infections.
We investigated the effect of annual influenza vaccination on the occurrence of lower respiratory tract infections (LRTIs) in community-dwelling elderly individuals. From 1996 to 2002, we performed a population-based cohort study, using the computerized Integrated Primary Care Information database in the Netherlands, of community-dwelling subjects who were 65 years or older on January 1 of the year of study entry. For each year, the individual cumulative exposure to influenza vaccination since study entry was computed. We compared the risk of LRTI after a first vaccination or revaccination with the risk for no vaccination using a time-varying multivariate Cox proportional hazard model, adjusted for age, sex, smoking, and underlying disease.
In the study population of 26 071 subjects, 3412 developed LRTIs during follow-up. During the influenza epidemic periods, a first vaccination did not reduce risk for LRTI. In the total population, the hazard ratio following a first vaccination was 0.86 (95% confidence interval [CI], 0.71 to 1.05); in the population without or with comorbidity, these ratios were 0.90 (95% CI, 0.56 to 1.45) and 0.83 (95% CI, 0.66 to 1.04), respectively. During epidemic periods, revaccination reduced risk of LRTI by 33% (95% CI, 8% to 52%) in individuals without comorbidity. In individuals with comorbidity, the risk reduction of 5% was nonsignificant (95% CI, −10% to 18%).
In this study, annual influenza revaccination was associated with a reduction in LRTI in community-dwelling elderly individuals.
During winter periods, up to 30% of elderly individuals (those ≥65 years)experience acute respiratory infections, of which up to 20% are attributed to the influenza virus.1- 4 Respiratory complications following infection with influenza virus include acute bronchitis, pneumonia, and exacerbations of chronic bronchitis or asthma.5 Accordingly, rates of hospitalization for pneumonia and influenza, acute bronchitis, and chronic respiratory disease are significantly higher during influenza epidemic periods.6,7 Acute bronchitis is the most common complication seen in primary care following influenza infection. The risk increases with age and in the presence of underlying conditions.8 In a meta-analysis9 covering 9 cohort studies (including 6 in nursing homes), influenza vaccination reduced the risk of hospitalization for pneumonia on average by 53% but with a considerable range.10,11 In a more recent meta-analysis12 of community-dwelling elderly individuals, the risk reduction of hospitalization for respiratory conditions after influenza vaccination was less than 40%. Other recent cohort studies of elderly patients confirm the results from the meta-analyses13,14 but indicate strong effect modification by risk profile.14- 17
To date, to our knowledge, studies have addressed only LRTIs leading to hospital admission, whereas most LRTIs are dealt with in primary care. To investigate the extent to which annual influenza vaccination is associated with the risk of community-dwelling elderly individuals developing LRTI that requires or does not require hospitalization, we conducted a population-based cohort study.
In the Dutch health care system, all persons have their own general practitioner (GP) who files all relevant medical details on their patients from primary care visits, hospital admissions, and visits to outpatient clinics. Since 1997, GPs have executed a nationwide influenza vaccination program for which they are reimbursed by the public budget. During annual mass vaccination days in October and November, all individuals with predefined risk factors and those who are 65 years or older are invited to participate in the vaccination campaign free of charge. The GPs register the vaccinations in their patient records.
Data for this study were derived from the Integrated Primary Care Information (IPCI) database at the Department of Medical Informatics of the Erasmus Medical Center in Rotterdam, the Netherlands. The IPCI database is a general practice research database that contains electronic patient records on a cumulative population of 530 000 patients from approximately 150 GPs and includes demographic information, patient complaints and symptoms, diagnoses, results of laboratory tests, referral notes from consultants, hospital admissions, and prescriptions. Prescriptions include drug name, Anatomical Therapeutical Chemical code, dosage form, dose, prescribed quantity, and indication. Symptoms and diagnoses are recorded using the International Classification for Primary Care as the coding system18 and also as free text. Because the IPCI is the sole repository of medical records, the participating GPs do not keep additional paper records. To ensure anonymity, patient and practice identifiers are altered. The system complies with European Union guidelines on the use of data for medical research and has been shown to be valid for pharmacoepidemiologic research.19- 21 The IPCI internal review board approved the study.
For this study, we selected all persons who were 65 years or older with a permanent status in one of the IPCI GP practices. Eligible persons had at least 1 year of recorded database medical history, which was required to adequately determine prior health status and vaccination status. To limit misclassification of exposure, we excluded all practices that did not consistently register influenza vaccination over the study years. Inconsistent registration included recorded vaccination coverage rates of less than 25% or a variation of 50% or more over the years. In the remaining population, we conducted a cohort study that included the period of October 1, 1996, to September 30, 2002. The end of follow-up was defined as the first episode of LRTI, death, moving out of the GP practice, or the end of the study period, whichever came first.
The cumulative number of influenza vaccinations since study entry was determined between October 1 and December 31 of each calendar year. Exposure was classified into 9 mutually exclusive categories: nonexposed; first, second, third, fourth, fifth, and sixth or seventh vaccination; vaccination interruption; and restart. Subjects without a record of vaccination in their medical history or during follow-up were considered to be nonexposed until their first recorded vaccination. Nonexposed subjects could be flagged as vaccine refusers in the GP database. These persons were not automatically invited for the annual influenza vaccination until they indicated that an annual invitation would be appreciated. These vaccine refusers were categorized as nonexposed for the main analyses. The vaccination history of the subjects was assessed for 1 year prior to study entry. Therefore, subjects with a recorded vaccination in the year before the study started were listed as having their second vaccination in the first year of study participation, although they may have had multiple vaccinations in the years prior to the period in which vaccination history was recorded. At each additional consecutive vaccination during the study period, the cumulative number recorded was increased by 1. Once subjects were in the interruption category, they stayed in it until vaccination restart, as described previously.21 Vaccination interruption and restart were 2 separate exposure categories and did not overlap with nonexposed or cumulative vaccination categories. In this time-varying approach to exposure analysis, individuals were able to contribute information to multiple exposure categories during their follow-up.
The primary outcome in this study was a first episode of LRTI, which was defined as pneumonia, acute bronchitis, or an exacerbation of chronic bronchitis. We identified LRTIs from the medical chart of the patient, and LRTIs were considered as an outcome only if they resulted in antibiotic drug therapy or were confirmed by radiography and/or microbiologic findings. To be able to compare our results with other studies, we also assessed hospitalizations for pneumonia as a secondary outcome. Events were allocated to the vaccination status defined in the preceding vaccination period.
Age, sex, smoking, antibiotic use, and the number of visits to a GP were considered as covariates. Furthermore, we identified 7 disease clusters as potential confounders: chronic respiratory tract disease (chronic obstructive pulmonary disease, emphysema, chronic bronchitis, and asthma), chronic cardiovascular disease (heart failure, angina pectoris, history of myocardial infarction or stroke, aortic aneurysm, and chronic arterial dysfunction), diabetes mellitus, hypertension, chronic renal insufficiency, malignancies, and neurological or psychiatric disorders (parkinsonism, dementia, polyneuropathy, multiple sclerosis, epilepsy, alcoholism, depression, psychosis, schizophrenia, and Ménière's disease). Presence of these conditions at study entry or their development during follow-up was verified from the medical charts through automatic screening and manual validation. The population without comorbidity had no comorbidity at baseline and did not develop any of the predefined conditions during the follow-up period. Information on the characteristics of each influenza season was obtained from J. C. de Jong, PhD, of the National Influenza Center, Erasmus Medical Center.
For estimation of the univariate association between vaccination, covariates, and the development of LRTI or pneumonia resulting in hospitalization, we used a Cox proportional hazards model. Multivariate time-varying Cox proportional hazard models were developed to estimate the hazard ratios for different vaccination states while adjusting for all other risk factors. For all analyses, age expressed in days was taken as the time axis to fully adjust for age.
The effect of vaccine exposure on the risk of the outcome was evaluated for a first vaccination or revaccination vs no vaccination. Subsequently, the outcomes were stratified for presence of comorbidity and age at study entry (65-69 years, 70-79 years, or >79 years). We first investigated these effects during the full follow-up period and subsequently during the influenza epidemic period. This epidemic period started on the first day of the first week of the recorded epidemic and ended on the last day of the last week of the recorded epidemic. Hence, the duration of the influenza epidemic period varied per year. As a comparison, the association of influenza vaccination and LTRI was assessed also in the summer period (June, July, and August) during which there is almost no circulation of the influenza virus in the Netherlands. In each analysis, the reference group comprised all nonvaccinated subjects. All results were expressed as hazard ratios with 95% confidence intervals. For all analyses, we used the PROC PHREG program of SAS statistical software (version 8.2; SAS Inc, Cary, NC).
The study population comprised 26 071 elderly individuals, of whom 58% were women (Table 1). The mean (SD) age at study entry was 73.1 (7.4) years. In this population, 3412 subjects developed a first episode of LRTI. In 1295 patients this first LRTI episode was classified as pneumonia, and 455 of these patients were hospitalized with this condition. The risk of developing LRTI was higher in men, in the presence of comorbidity, and in smokers, and the risk increased with age. The presence of chronic respiratory diseases at baseline was especially associated with an increased risk of LRTI.
Influenza vaccination coverage and epidemic characteristics of this population have been published previously.21 In short, vaccination coverage varied from 64% to 74% and was slightly higher in the subpopulation with comorbidity (72%-79%). Most vaccinations (96%) were given in October or November, and the remainder in December. During the study, 59 111 influenza vaccinations were administered to 20 976 persons; 5095 persons never received any influenza vaccination, and 3063 subjects were listed as vaccine refusers for 1 or more seasons. All influenza epidemics were predominated by influenza A(H3N2) strains, which showed a good match between circulating and vaccine strain(s) except for the 1997-1998 season (mismatch). True epidemic activity was absent in the 2000-2001 season; in addition, this activity was caused by an influenza B strain. Furthermore, epidemic activity was mild to moderate in the 1996-1997 and 1999-2000 seasons and mild in the other seasons. The peak activity of influenzalike illness ranged from 7 cases per 10 000 persons (2000-2001 season) to 32 cases per 10 000 persons per week (1999-2000 season) and was observed between weeks 2 and 13 of the epidemic (J. C. de Jong, PhD, written communication, August 20, 2004).
Table 2 summarizes the association between influenza vaccination and LRTI. If we considered the full follow-up period, a first influenza vaccination or revaccination was not associated with a reduction in the risk of LRTI. If we restricted follow-up to the epidemic period, the risk of LRTI following any revaccination decreased by 33% but only in subjects without comorbidity (Table 2). Influenza vaccination did not protect against LRTI in persons with comorbidity or with chronic respiratory disease. Stratification by age suggested a tendency toward a lower risk of LRTI following revaccination in elderly individuals, but this was not seen for first vaccinations.
We were able to differentiate those refusing vaccination from the rest of the nonexposed control group. In total, 344 first episodes of LRTI were recorded in subjects who refused vaccination for 1 or more seasons. Persons with underlying comorbidity listed as vaccine refusers were at a lower risk of LRTI compared with the rest of the nonvaccinated persons with comorbidity (hazard ratio, 0.75; 95% confidence interval, 0.59-0.95). By contrast, persons without comorbidity listed as vaccine refusers were at a significantly higher risk of LRTI than the rest of the nonvaccinated control group without comorbidity (hazard ratio, 1.78; 95% confidence interval, 1.11-2.86). Subjects who interrupted the vaccination series and/or restarted had no benefit from the vaccination history. Their risk was comparable with the nonexposed control group (data not shown).
Considering the full follow-up period, there was no benefit of first or revaccination on the risk of pneumonia (Table 3). During the influenza epidemic period, revaccination was associated with a 50% risk reduction of pneumonia but only in the subpopulation without comorbidity. A first influenza vaccination was associated with a 71% risk reduction (95% confidence interval, 4%-90%) of hospitalized pneumonia, but no significant association was seen for revaccination. Influenza vaccination was not associated with all cases of LRTI or pneumonia during the summer period (Table 2 and Table 3). Only in the seasons with mild to moderate epidemic activity was revaccination associated with a reduced risk of LRTI compared with a first vaccination (Figure). During seasons with mild or no epidemic activity, subjects with a first vaccination consistently had a lower risk of LRTI than revaccinated subjects.
In the study described herein, we show that influenza vaccination reduces the risk of LRTI, but only in specific subgroups and during the epidemic period. The observed protection against hospitalization for pneumonia is in line with other studies.11- 13 In our study, revaccination was beneficial only in years with mild to moderate epidemic activity. The only season in which both first vaccination and revaccination were associated with a significantly lower risk of LRTI was the 2000-2001 season, which was predominated by the influenza B strain. This emphasized the need to analyze these morbidity outcomes by epidemic year. Others14,16,17 reported that the potential impact of influenza vaccination on the risk of developing pneumonia depends on an individual's risk profile. With increasing age, a tendency toward a stronger effect of revaccination was noted in our study, although the estimates did not reach statistical significance. One explanation might be the gradual impairment of health status with age, resulting in a higher disease burden, which allows for a stronger impact of vaccination. The tendency toward a reduced beneficial effect of a first vaccination might reflect an impaired immune response in elderly individuals. Similar observations, although not distinguished between first vaccination and revaccination, led to the conclusion that the benefit of influenza vaccination in elderly individuals was limited to those younger than 70 years.22
Like all observational studies, this study may suffer from selection bias, information bias, and confounding factors. Selection bias was discarded because the data were obtained from computerized GP patient records that are population-based and independent of morbidity. Information bias may occur on exposure and outcome. Misclassification of the cumulative number of influenza vaccinations is possible, but the incidence rate is likely to be low because all data are recorded in the computer at the time of the visit. The vaccinations are supplied almost exclusively by GPs, and we included only practices with consistent vaccination coverage over the follow-up period. Misclassification of LRTI in a GP setting may be a problem. To some extent, the validation process in free text is valuable to control for possible variations in the application of ICPC codes by GPs. However, to enhance sensitivity of the clinical criteria, confirmation by microbiologic findings is important.23 Respiratory tract infections are caused by a variety of viruses and bacteria and are common in elderly individuals, especially during winter months,23 but are often not confirmed.3 Although the influenza virus is the most commonly isolated specimen in respiratory infections in adults,24 other viruses (eg, respiratory syncytial virus, rhinoviruses, and corona viruses) are more frequent in elderly individuals.23 Also, clinical symptoms of LRTI can be misleading in community-dwelling elderly individuals.3 Misclassification of LRTI may have been differential if GPs were less likely to diagnose LRTI in vaccinated persons in view of the assumed effectiveness. Another potential problem is confounding by indication. Although the national recommendation is to vaccinate all elderly individuals who are 65 years or older, generally healthy individuals do not always follow such advice.25 On the contrary, critically ill patients may refuse vaccination or may not be offered the vaccination in view of their poor prognosis. However, this seemed not to be the case in our study. Persons with underlying comorbidity listed as vaccine refusers seemed to be in relatively better health than the rest of the nonvaccinated control group whereas those without comorbidity seemed to be relatively less healthy. Apparently, however, vaccine refusers do not have a consistent risk profile. We suspected potential confounding by indication because preexisting chronic respiratory tract disease, cardiovascular tract disease, diabetes mellitus, malignancy, chronic renal insufficiency, and neurological and psychiatric disorders were independent risk factors for increased influenza-associated morbidity. Therefore, we adjusted for these risk factors. However, some residual confounding is possible and may have led to an underestimation of the true effect of the influenza vaccination.23
The strength of this study is that the effect of repeated vaccinations on LRTI or pneumonia could be estimated in each of 6 consecutive seasons in which influenza epidemic and vaccine strain characteristics varied. The possible influence of vaccine strain selection on effectiveness of vaccination in repeated vaccines has already been reported.26,27 In contrast to the impact of interruption and restart of vaccination on mortality, in the present study, interruption and restart of the vaccination series had no obvious impact on vaccination effectiveness.21
In summary, this study indicates that in a population of community-dwelling elderly individuals, repeated influenza vaccinations may reduce the risk of developing LRTI or pneumonia in years with higher influenza epidemic activity. Although the protective effect is modest, influenza vaccination should be advised in view of the high background incidence of LRTI and because of the observed reduced mortality in elderly individuals.21
Correspondence: Bettie C. G. Voordouw, MD, PhD, MPH, Medicines Evaluation Board, PO Box 16229, 2500 BC, The Hague, the Netherlands (email@example.com).
Accepted for Publication: June 20, 2006.
Author Contributions:Study concept and design: Voordouw, Sturkenboom, Stijnen, and Stricker. Acquisition of data: Voordouw, Sturkenboom, Dieleman, and Stricker. Analysis and interpretation of data: Voordouw, Sturkenboom, Dieleman, Stijnen, and Stricker. Drafting of the manuscript: Voordouw. Critical revision of the manuscript for important intellectual content: Sturkenboom, Dieleman, Stijnen, van der Lei, and Stricker. Statistical analysis: Voordouw, Sturkenboom, Stijnen, Dieleman, and Stricker. Obtained funding: Stricker, Sturkenboom, and van der Lei. Administrative, technical, and material support: Sturkenboom, Dieleman, and van der Lei. Study supervision: Sturkenboom, Stricker, and van der Lei.
Financial Disclosure: None reported.
Acknowledgment: We thank Jan C. de Jong, PhD, Department of Virology, National Influenza Centre, Erasmus Medical Center, Rotterdam, for the use of the influenza epidemiological data.