key: cord-252293-8286lsof authors: Suzuki, Motoi; Katsurada, Naoko; Le, Minh Nhat; Kaneko, Norihiro; Yaegashi, Makito; Hosokawa, Naoto; Otsuka, Yoshihito; Aoshima, Masahiro; Yoshida, Lay Myint; Morimoto, Konosuke title: Effectiveness of inactivated influenza vaccine against laboratory-confirmed influenza pneumonia among adults aged ≥65 years in Japan date: 2018-05-17 journal: Vaccine DOI: 10.1016/j.vaccine.2018.04.037 sha: doc_id: 252293 cord_uid: 8286lsof BACKGROUND: The effectiveness of inactivated influenza vaccine (IIV) against laboratory-confirmed influenza pneumonia in older adults remains to be established. METHODS: Pneumonia patients aged ≥65 years who visited a study hospital in Chiba, Japan, were prospectively enrolled from February 2012 to January 2014. Sputum samples were collected from participants and tested for influenza virus by polymerase chain reaction assays. Influenza vaccine effectiveness (IVE) against laboratory-confirmed influenza pneumonia was estimated by a test-negative design. RESULTS: Among a total of 814 pneumonia patients, 42 (5.2%) tested positive for influenza: 40 were positive for influenza A virus, and two were positive for influenza B virus. The IVE against laboratory-confirmed influenza pneumonia was 58.3% (95% confidence interval, 28.8–75.6%). The IVE against influenza pneumonia hospital admission, severe pneumonia, and death was 60.2% (95% CI, 22.8–79.4%), 65.5% (95% CI, 44.3–78.7%), and 71% (95% CI, −62.9% to 94.8%), respectively. In the subgroup analyses, the IVE against influenza pneumonia was higher for patients with immunosuppressive conditions (85.9%; 95% CI, 67.4–93.9%) than for those without (48.7%; 95% CI, 2.7–73%) but did not differ by patients’ statin use status. CONCLUSION: IIV effectively reduces the risk of laboratory-confirmed influenza pneumonia in older adults. Influenza is a major public health concern for older adults. Influenza infections generally cause self-limited illnesses but can result in severe disease such as pneumonia in older adults with and without underlying conditions. Older age is associated with a higher risk of pneumonia and mortality in influenza patients [1] . Based on our recent estimates, the incidence of influenza pneumonia and its related mortality among people aged !65 years in Japan were 210 and 24 per 100,000 persons/year, respectively [2] . Cumulative evidence has suggested that influenza vaccines are effective at reducing the risk of medically attended influenza in children and adults [3, 4] . Currently, seasonal influenza vaccination is recommended for older adults in more than 90 countries [5] . However, its clinical benefit has long been discussed because vaccine responses are reduced by an age-related decline in adaptive immunity [6, 7] . Positive results have been reported from recent meta-analyses: influenza vaccines reduce medically attended influenza by 20-44% [8] and influenza-associated hospitalization by 37% in older adults [9] . However, evidence is lacking for the protective effect of influenza vaccination on influenza pneumonia, including primary influenza pneumonia and secondary bacterial pneumonia. In a study by Grijalva influenza pneumonia by 56.7%, although the majority of their patients were people aged <65 years [10] . Therefore, the influenza vaccine effectiveness (IVE) against laboratory-confirmed influenza pneumonia in older adults remains to be established. We conducted this study to investigate the effectiveness of the trivalent inactivated influenza vaccine (IIV) against laboratoryconfirmed influenza pneumonia and its related outcomes in adults aged !65 years. We also conducted subgroup analyses to explore differences in IVE by patient characteristics, particularly those related to immunosuppressive status. This single-center prospective study was conducted at Kameda Medical Center (KMC), Kamogawa, Chiba, Japan, as part of the Adult Pneumonia Study Group-Japan (APSG-J) Study [2, [11] [12] [13] . The APSG-J Study was a multicenter prospective study of adult pneumonia conducted at four community-based hospitals in Japan from September 2011 to August 2014. To investigate IVE, influenza vaccination history was systematically collected at KMC. In this study, pneumonia patients aged !65 years who visited KMC from February 2012 to January 2014 were included. The diagnosis of pneumonia was made by staff physicians according to clinical signs, symptoms, and radiological findings. Demographic and clinical information was collected from patients and medical charts. Sputum samples were collected from patients at the time of enrollment. If the patient was unable to cough up sputum, sputum was induced with the inhalation of hypertonic saline solution. Details of study settings and designs have been described previously [2, 13] . Gram staining and sputum culture were performed on site. Sputum samples were transferred to the Institute of Tropical Medicine, Nagasaki University, and tested by in-house multiplex polymerase chain reaction (PCR) assays to identify the influenza virus (A and B) and 11 other viral pathogens (respiratory syncytial virus [RSV], human metapneumovirus, human parainfluenza virus types 1-4, human rhinovirus [HRV], human coronavirus 229E/OC43, human adenovirus, and human bocavirus) [14] . The detection limits of the multiplex PCR assays were 10 -100 copies per reaction as reported previously [14] . Influenza virus subtyping was performed for influenza A-positive samples via RT-PCR of the influenza HA genes using previously published methods [15, 16] . Patients were defined as having laboratory-confirmed influenza pneumonia if their sputum sample tested positive for influenza A or B virus by PCR. Influenza pneumonia patients were classified as having influenza-associated bacterial pneumonia if their sputum samples were microscopically purulent (i.e., Geckler's classification groups 4 and 5) and tested positive for bacterial pathogens by culture or PCR; otherwise, they were classified as having primary influenza pneumonia. A test-negative design (TND) case-control study was applied to estimate IVE [17] . Unlike the conventional case-control design, the TND does not require non-disease controls; instead, in this study design, researchers collect clinical samples from patients with a specific condition (eg, influenza like illnesses) and classify the patients into cases (i.e., influenza tested positive patients) and controls (i.e., influenza tested negative patients) according to the influ-enza test results. The TND is less susceptible to bias due to differences in health care-seeking behavior among cases and controls and provides reliable IVE estimates [18, 19] . Recently, TND studies have been widely used to estimate IVE against medically attended influenza and influenza-associated hospitalization [8, 9] . In the current study, our primary outcome was laboratoryconfirmed influenza pneumonia. Cases were pneumonia patients who tested positive for influenza A or B, and controls were pneumonia patients who tested negative for both influenza A and B. The odds of vaccination were compared between cases and controls, and IVE was expressed as (1-odds ratio) Â 100%. Our secondary outcomes were (1) primary influenza pneumonia, (2) influenza-associated bacterial pneumonia, influenza pneumonia-related hospital admission, (4) severe influenza pneumonia, and (5) influenza pneumonia death. In Japan, all adults aged !65 years are recommended by the Ministry of Health, Labor and Welfare to receive one dose of the seasonal influenza vaccine [20] . The trivalent IIV vaccine was used during the study period (2011-12, 2012-13, and 2013-14 seasons); the quadrivalent IIV vaccine was introduced in the 2015-16 season. High-dose or adjuvanted IIVs have not been licensed in Japan. The compositions of the trivalent IIV vaccines used during the study seasons and their antigenic match status are summarized in Supplementary Table 1 . Influenza vaccination histories were collected from medical records and confirmed by patients and/or their guardians. Patients were considered vaccinated for influenza if they had received at least one dose of influenza vaccine in the 12 months before the hospital visit. Because the duration from influenza vaccination to the hospital visit was recorded as a monthly data, all patients who had been vaccinated within a month were considered vaccinated in our primary analysis. Patients were considered as having unknown influenza vaccination statuses if their influenza vaccination histories were not recorded in medical charts and could not be confirmed by the patients or their guardians; this group was excluded from our primary analysis. Patients were categorized into three age groups: 65-74 years, 75-84 years, and 85 years or older. Patient disability status was evaluated using the Eastern Cooperative Oncology Group Performance Status score [21] . Body mass index (BMI, kg/m 2 ) was classified as underweight (<18.5), normal (18.5-24.9), or overweight (!25.0). Chronic conditions included diabetes mellitus, heart failure, ischemic heart disease, cerebrovascular disease, liver disease, renal disease, neurological disease, cancer, chronic obstructive pulmonary disease, bronchial asthma, and previous tuberculosis disease. Immunosuppressive status included cancer, oral steroid use, and immunosuppressive drug use. Patients were considered to have severe pneumonia if they required oxygen therapy, mechanical ventilation, or a vasopressor after admission. The period from November to April was considered the influenza season. The characteristics of patients were compared according to influenza infection status (i.e., influenza pneumonia vs. noninfluenza pneumonia) and influenza vaccination status (i.e., vaccinated vs. unvaccinated) using chi-square tests and Fisher's exact tests for categorical variables and Wilcoxon rank sum tests for numerical variables. IVE was estimated using logistic regression models. Pre-specified confounding factors were sex, age, the pres-ence of chronic conditions, the presence of immunosuppression, smoking status, the duration from onset to hospital visit, and the period of the study, and all these variables were included in the final multivariable logistic regression models. We also considered the performance status score and BMI category as potential confounders and examined if IVE estimates changed after adjusting for these variables. Confidence intervals (CIs) were adjusted for the residential area level clustering of patients using robust standard errors. We conducted sensitivity analyses as follows: (1) restricting the analysis to patients who visited during influenza seasons; (2) excluding patients vaccinated <1 month prior to hospital visit; (3) excluding patients vaccinated >6 months prior to hospital visit; (4) using patients who were negative for influenza virus but positive for non-influenza respiratory viruses as controls; (5) using patients who were negative for all viruses as controls [22] ; (6) using propensity scores for adjustment; and (7) including patients with unknown vaccination status using multiple imputation. Stratified analyses were conducted to investigate the potential effect modifications by patient characteristics (i.e., sex, age group, underlying condition, immunosuppressive status, and statin use status). Stratum-specific IVE estimates were compared using a likelihood ratio test (test for interaction). This study was approved by the institutional review board (IRB) of the Institute of Tropical Medicine, Nagasaki University, Nagasaki, Japan and the IRB of Kameda Medical Center, Chiba, Japan. Anonymized data were used in this study. During the study period, a total of 1494 pneumonia patients aged !65 years were enrolled in the study. Among them, sputum samples were obtained from 1044 patients (70%). After excluding 230 patients whose influenza vaccination history were unavailable (22% of patients with sputum samples), a total of 814 patients were eligible for our analyses (Fig. 1) . Among them, 42 (5%) tested positive for influenza virus by PCR: 40 were positive for the influenza A virus, and the other two were positive for the influenza B virus. Among the 26 influenza A-positive samples that were subtyped (65% of all influenza A-positive samples), all were positive for the H3N2 strain. Non-influenza viruses were detected in 178 patients: HRV was the leading virus detected (n = 77, 9%), followed by RSV (n = 36, 4%). Non-influenza viruses were co-detected in 6 of the 42 influenza-positive patients (14%) and detected in 172 of the 772 influenza-negative patients (22%). Bacterial pathogens were co-detected in 26 of the 42 influenza-positive patients (62%). Demographic and clinical characteristics were compared between influenza pneumonia patients (i.e., cases) and noninfluenza pneumonia patients (i.e., controls) (Tables 1 and 2 ). Cases were more frequently found in winter seasons than controls, but other characteristics were similar between cases and controls. Among 814 patients, 525 (65%) had been vaccinated for influenza. Vaccinated patients more frequently had received home oxygen therapy and had been diagnosed with chronic respiratory obstructive disease than unvaccinated patients, while other characteristics were similar between two groups (Tables 1 and 2) . After adjusting for confounders, the IVE against laboratoryconfirmed influenza pneumonia was 58.3% (95% CI, 28. ( Table 3 ). The change in IVE estimates was marginal after additional adjustment for performance status (58.9%; 95% CI, 30.6-75.7%) or BMI category (58.0%; 27.6-75.6%); therefore, these variables were not included in the final models. The sensitivity analyses showed similar results. IVE was relatively higher (68.9%; 95% CI, 46.4-81.9%) when we used patients who were negative for influenza but positive for non-influenza viruses as controls, but the value was almost identical to the primary analysis when we used patients who were negative for all viruses (57.8%; 95% CI, 26.9-75.7%). For the secondary outcomes, the IVE against primary influenza pneumonia (70.1%; 95% CI, 19.8-88.9%) was higher than that Table 4 . The IVE against influenza pneumonia was higher in patients with immunosuppressive conditions (85.9%; 95% CI, 67.4-93.9%) than in those without these conditions (48.7%; 95% CI, 2.7-73%; test for interaction, p = 0.001). IVE did not differ by sex. The point estimate of IVE decreased with increased age, but the difference did not reach a statistically significant level (test for interaction, p = 0.17). Patients' chronic conditions and statin use status did not modify IVE. IIV effectively reduced the risk of laboratory-confirmed influenza pneumonia in adults aged !65 years. IVE was higher among patients with immunosuppressive conditions, while statins did not modify IVE. To our knowledge, this is the first study that confirmed the beneficial effect of seasonal influenza vaccination against laboratory-confirmed influenza pneumonia in older adults. The benefit of seasonal influenza vaccination in older adults is still debated [3, 23] . In this age group, the age-related decline in adaptive immunity results in reduced responses to influenza vaccination [6, 7] ; moreover, multiple chronic conditions and frailty may also contribute to weak immune responses [24] . However, despite an observed lower antibody response compared with that of younger adults [25] , recent evidence supports the protective effect of influenza vaccination against medically attended influenza in older adults. According to a systematic review by Belongia et al, the pooled IVE was 24% (95% CI, À6% to 45%) for the H3N2 strain, 63% (95% CI, 33-79%) for type B, and 62% (95% CI, 36-78%) for the H1N1pdm09 strain among adults aged >60 years [4] . Darvishian et al conducted an individual participant data meta-analysis of TND studies and demonstrated that influenza vaccination is moderately effective against laboratory-confirmed influenza in this age group during epidemic seasons but not during non-epidemic seasons [8] . On the other hand, evidence is still limited for the beneficial effect of influenza vaccination against influenza-related severe outcomes such as pneumonia. Previous studies have estimated the IVE against all-cause pneumonia or influenza-related hospitalization in older adults [9, [26] [27] [28] ; however, these studies used less specific outcomes and may have underestimated the true IVE [17] . The TND study by Grijalva et al demonstrated that the overall estimate of IVE against hospitalization with laboratory-confirmed influenza pneumonia was 56.7% (95% CI, 31.9-72.5%) [10] . However, their study included all age groups, and only 16% of their patients were aged !65 years. In their analysis restricted to patients aged !65 years, IVE showed a positive effect but did not reach a statistically significant level (48.4%; 95% CI, À33.3% to 80%). Therefore, the authors concluded that additional studies were needed to establish the IVE against pneumonia in older adults. Our study targeted this age group and demonstrated that the vaccine effectively reduces the risk of laboratory-confirmed influenza pneumonia by 58.3% (95% CI, 28.8-75.6%). Our IVE estimates against influenza pneumonia in older adults may be higher than generally expected values. IVE is commonly lower for severe outcomes than for medically attended influenza and is lower in older adults than in children [3, 9] . In addition, IVE is usually lower for the H3N2 stain than for the H1N1pdm09 strain [4] . However, our estimates are not dissimilar to those of previous reports: in the study by Grijalva et al, the IVE against influenza pneumonia related to the H3N2 strain in all age groups was 45.1% (95% CI, À9.3% to 72.4%) [10] , and in another study conducted during the 2011-12 influenza season when H3N2 was the dominant circulating strain, the IVE against influenza hospitaliza- [29] . The use of sputum samples in our study may also explain our high IVE estimate. Identification of influenza from sputum samples may be more sensitive and specific than that from upper respiratory tract samples in diagnosing influenza pneumonia and may provide less biased IVE estimates [17, 30, 31] . Consistent findings in our sensitivity analyses also support the robustness of our IVE estimates. Although a higher IVE estimate was observed for primary influenza pneumonia, IIV was also effective at preventing influenzaassociated bacterial pneumonia (49.1%; 95% CI, 17.1-68.7%). This finding is important because influenza-bacterial co-infection increases the risk of severe outcomes [32] . Our finding also suggests that IIV may be effective at preventing influenza pneumonia death; however, the association did not reach a statistically significant level because of the limited sample size. It was unexpected that the IVE was significantly higher among people with immunosuppressive conditions (85.9%; 95% CI, 67.4-93.9%) than among those without (48.7%; 95% CI, 2.7-73%). The opposite finding was observed in the study by Grijalva et al, which included children and adults (À21.9% vs. 73.4%) [10] . This difference might be, at least partially, explained by a lower HIV prevalence in our patients. Although seasonal influenza vaccinations have been recommended for adults with immunosuppressive conditions [33] , only a few studies have evaluated the IVE against clinical outcomes among this population [34] . Our finding provides supporting evidence for the current recommendations but needs to be confirmed in future studies. Recent studies have suggested that statins may reduce the IVE against medically attended influenza among older adults by their immunomodulatory effects [35] [36] [37] [38] . However, such an effect has not been observed in our study. Although the degree of its effect remains controversial, statins are also known to modify the risk of pneumonia and pneumonia-related outcomes [39] [40] [41] . The impact of statin use on the IVE may be different according to influenza outcomes. Influenza infection is a threat to older adults because of its potential to cause pneumonia and secondary bacterial infections [13] . The burden of pneumonia is rapidly increasing in highincome countries such as Japan because of the aging population [2] . Therefore, the prevention of influenza pneumonia is an important public health measure in controlling pneumonia. The moderate effectiveness observed in our study supports the current seasonal influenza vaccination policy. In Japan, the proportion of people vaccinated against influenza among adults aged !65 years has been increasing but still remains approximately 60% [42] . In addition to improving vaccination coverage, an introduction of newer vaccines such as the more immunogenic high-dose influenza vaccine must be considered [43, 44] . On the other hand, it must be noted that only 5% of pneumonia cases have influenza pneumonia, and thus, the impact of influenza vaccination on allcause pneumonia is limited [45] . Newer multidimensional approaches are needed to reduce the pneumonia burden in the aging population. Our study has limitations. Influenza vaccination history was not documented for 22% of our patients. However, our sensitivity analysis using multiple imputations showed very robust estimates. We believe that the exclusion of this patient group did not affect our IVE estimates. Although all potential confounders were considered, unmeasured confounders may have remained. Recently, Andrew et al argued that frailty must be considered in estimating IVE for older adults [29] . We have not measured the frailty of our patients but measured their performance status and BMI. We confirmed that the inclusion of performance status or BMI category in the final model did not change the IVE estimates. Our observation is based on the analyses of older patients aged !65 years and there- fore may not be generalizable to younger adults. Finally, our sample size was too small to estimate subtype-specific IVE. Seasonal influenza vaccination is moderately effective against laboratory-confirmed influenza pneumonia in adults aged !65 years. Considering the increasing burden of pneumonia in an aging population, we must improve influenza vaccination coverage and establish newer approaches. Konosuke Morimoto reports speaker fees from Taisho Toyama Pharmaceutical, Pfizer, and Asahi Kasei Pharma. All other authors declare no competing interests. Populations at risk for severe or complicated influenza illness: systematic review and meta-analysis The burden and etiology of community-onset pneumonia in the aging Japanese population: a multicenter prospective study Efficacy and effectiveness of influenza vaccines: a systematic review and meta-analysis Variable influenza vaccine effectiveness by subtype: a systematic review and meta-analysis of test-negative design studies WHO. WHO vaccine-preventable diseases: monitoring system Innate immunity in aging: impact on macrophage function The effect of ageing of the immune system on vaccination responses Effectiveness of seasonal influenza vaccination in community-dwelling elderly people: an individual participant data meta-analysis of test-negative design case-control studies Effectiveness of influenza vaccines in preventing severe influenza illness among adults: a systematic review and meta-analysis of test-negative design case-control studies Association between hospitalization with community-acquired laboratory-confirmed influenza pneumonia and prior receipt of influenza vaccination Recurrent pneumonia among Japanese adults: disease burden and risk factors Serotype-specific effectiveness of 23-valent pneumococcal polysaccharide vaccine against pneumococcal pneumonia in adults aged 65 years or older: a multicentre, prospective, test-negative design study The impact of virus infections on pneumonia mortality is complex in adults: a prospective multicentre observational study Viral pathogens associated with acute respiratory infections in central vietnamese children Vaccine effectiveness against medically attended laboratory-confirmed influenza in Japan Impact of 2009 pandemic influenza among Vietnamese children based on a populationbased prospective surveillance from Methodologic issues regarding the use of three observational study designs to assess influenza vaccine effectiveness The test-negative design: validity, accuracy and precision of vaccine efficacy estimates compared to the gold standard of randomised placebo-controlled clinical trials. Euro surveillance: bulletin Europeen sur les maladies transmissibles = Theoretical basis of the testnegative study design for assessment of influenza vaccine effectiveness Negative impact of prior influenza vaccination on current influenza vaccination among people infected and not infected in prior season: a testnegative case-control study in Japan Toxicity and response criteria of the Eastern Cooperative Oncology Group Potential effect of virus interference on influenza vaccine effectiveness estimates in test-negative designs Vaccines for preventing influenza in the elderly Influenza vaccine response in community-dwelling German prefrail and frail individuals Antibody response to influenza vaccination in the elderly: a quantitative review Influenza vaccine effectiveness in preventing hospitalization for pneumonia in the elderly Influenza vaccine effectiveness among elderly nursing home residents: a cohort study Effectiveness of influenza vaccine in the community-dwelling elderly The importance of frailty in the assessment of influenza vaccine effectiveness against influenza-related hospitalization in elderly people Yield of sputum for viral detection by reverse transcriptase PCR in adults hospitalized with respiratory illness Comparison of sputum and nasopharyngeal swabs for detection of respiratory viruses Mechanisms of severe mortalityassociated bacterial co-infections following influenza virus infection Vaccination recommendations for the adult immunosuppressed patient: a systematic review and comprehensive field synopsis Influenza vaccines in immunosuppressed adults with cancer. Cochrane Database Syst Rev Influence of statins on influenza vaccine response in elderly individuals Impact of statins on influenza vaccine effectiveness against medically attended acute respiratory illness Effect of statin use on influenza vaccine effectiveness How regulatory T-cell induction by statins may impair influenza vaccine immunogenicity and effectiveness The effect of statins on mortality from severe infections and sepsis: a systematic review and metaanalysis Is statin use associated with reduced mortality after pneumonia? A systematic review and meta-analysis Statin use and hospital length of stay among adults hospitalized with community-acquired pneumonia Estimated influenza vaccination rates in Japan Comparative effectiveness of high-dose versus standard-dose influenza vaccines among US medicare beneficiaries in preventing postinfluenza deaths during Comparative effectiveness of high-dose versus standard-dose influenza vaccination on numbers of US nursing home residents admitted to hospital: a cluster-randomised trial Inactivated influenza vaccines for prevention of community-acquired pneumonia: the limits of using nonspecific outcomes in vaccine effectiveness studies We are grateful to all the Adult Pneumonia Study Group-Japan contributors. We would like to thank Professor Koya Ariyoshi, Dr. Eiichiro Sando, and Dr. Tomoko Ishifuji for their contribution to the study. We also thank Rina Shiramizu, Kyoko Uchibori for performing the PCR and Yumi Araki for administrative work. This study was supported by Nagasaki University and Pfizer Japan, Inc. Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.vaccine.2018.04. 037.