Study sample
The present study had a cross-sectional design and uses data from the Japan Gerontological Evaluation Study (JAGES). JAGES was designed to describe the health status and social determinants of people aged 65 years and older, without disabilities, and not eligible for the long-term care insurance (LTCI) in Japan. We used the data “JAGES 2016,” which was obtained from self-reported questionnaires mailed to and filled-in by community-dwelling individuals in 39 municipalities in 2016. In the JAGES 2016 wave, self-administered questionnaires were mailed to functionally independent adults aged 65 years or older who did not receive benefits from the LTCI insurance in Japan. The survey was conducted in the municipalities between September 2016 and January 2017. The data included 180,021 individuals who answered the questionnaires with the basic items (response rate was 70.2%). Participants who did not answer questions regarding age or sex were excluded (n=2,030). The data consisted of individuals who were not eligible for LTCI on 1st April 2016. The data consisted of a three-stage hierarchal structure. The individual data were nested into 720 communities based on elementary or junior high school districts, and these communities were further nested into 39 municipalities.
Outcome variables
The occurrence of pneumonia and hospitalization due to it in the past year from September 2016 to January 2017 were the outcome variables. The occurrence of pneumonia was assessed by asking “Did you fall sick in the past year?” and instructing the participants to select an appropriate answer from the following items: “Influenza,” “Pneumonia”, and “none of them.” Hospitalization due to pneumonia was assessed for participants who answered “Influenza” or “pneumonia” in the previous question by asking “If the sickness was due to influenza or pneumonia, were you hospitalized with relation to it?” and instructing the participants to select an answer from the following options: “Not hospitalized,” “Hospitalized due to influenza,” “Hospitalized due to pneumonia,” “Contracted influenza while I was hospitalized for other diseases,” and “Developed pneumonia while I was hospitalized for other diseases.” Only participants who answered “Hospitalized due to pneumonia” were considered to be hospitalized. The participants who answered “Influenza” in the first question and “Hospitalized due to pneumonia” in the second were included because influenza can cause primary viral or secondary bacterial pneumonia 12.
Frailty
Frailty was assessed using the Kihon Check List (KCL) 13. KCL was developed by the Japanese Ministry of Health, Labor and Welfare to identify older adults requiring LTCI. KCL was included in the self-administered questionnaires of the JAGES 2016 wave. KCL consists of 25 questions classified into the following seven categories: instrumental activity of daily living (IADL), physical strength, nutritional status, oral function, homeboundness, cognitive function, and depressive mood (Supplemental Table 1). The scores from the KCL was well correlated with the validated assessments of physical strength, nutritional state, cognitive function, depressive mood, and the number of frailty phenotypes defined by the Cardiovascular Health Study criteria (CHS) 13. Frailty was categorized into three groups based on KCL scores: robust, 0–3; pre-frail, 4–7; and frail, ≥8; scores were calculated from the KCL questions, which were validated with the pre-frail and frail categories established by the CHS criteria13. KCL variables were generated for each of the seven categories. Each KCL variable was categorized into three groups, based on answers to questions: 0, not applicable; 1, applicable; and 2 or ≥2 applicable.
Covariates
Age was classified into two groups (65–74 years and ≥ 75 years). Educational attainment was categorized into five groups: < 6 years, 6–9 years, 10–12 years, ≥ 13 years, and others. Equalized income was calculated by dividing the normalized household gross income in 2015 by the square root of the number of household members, and was categorized into five groups: < 0.5, 0.50–0.99, 1.00–1.99, 2.00–3.99, and ≥ 4.00 million yen. The household structure was assessed by asking the respondents the question, “Who do you live with?”. They were asked to choose from the following options: “no one,” “spouse,” “son,” “daughter,” “spouse of child,” “grandchild,” “brother or sister, “father,” “mother,” “father-in-law,” “mother-in-law,” and “other.” The responses were classified into six groups as follows: living alone, living with a spouse, living with children, living with a spouse and children, living in three-generation households (living with/without a spouse, but with one of the sons/daughters/son’s or daughter’s spouse and grandchildren), and the in a household structure other than the above five categories. Marital status was assessed by asking, “What is your marital status?” and participants were instructed to select one from the following five options: “Married,” “Widowed,” “Divorced,” “Never married” and “Other.” Smoking status was assessed by asking, “Do you smoke cigarettes?” and the participants were instructed to select appropriate answers from the following items: “Never smoked,” “Quit smoking ≥5 years ago”, “Quit smoking < 5 years ago,” “Smoke sometimes” and “Smoke almost every day.” The population density of the municipality was categorized as follows: metropolitan (density over 4,000 people per km2), urban (density between 1,500 and 4,000 people per km2), semi-urban (density between 1,000 and 1,499 people per km2), and rural (density below 1,000 people per km2) 14. A municipality dummy variable was generated to adjust for differences in municipalities’ policies in preventing frailty 15. The diabetes, respiratory, heart, kidney/prostate gland, hematological, or immune disease status was assessed by asking participants whether they were receiving any treatment or experiencing after-effects of any of the above diseases. Pneumococcal vaccination status was assessed by asking the participants, “Did you get a pneumococcal vaccination in the last five years? They chose from the following options: “No,” “Yes, I got vaccinated using my municipality's subsidy”, and “Yes, but I did not get vaccinated using my municipality's subsidy”. The participants who chose the two latter options were categorized as vaccinated.
Statistical analysis
Multilevel Poisson regression analyses with random intercepts were performed to assess associations between frailty and pneumonia/hospitalization after adjusting for all the covariates. The data were structured in three levels: the individuals were nested within their elementary or junior high school districts and the districts were further nested within the municipalities. The covariates consist of all the covariates at individual-level and the municipality dummy variable used to adjust for differences in municipalities’ policies in older adult’s health including preventing frailty and pneumonia at municipality-level15,16. The individual-level covariates were: age group, sex, educational attainment, equivalized income, household structure, marital status, smoking status, municipality population density, diabetes, respiratory disease, heart disease, kidney/prostate gland disease, hematological/immune disease, and pneumococcal vaccination. The prevalence ratios (PRs) and 95% confidence intervals (95% CIs) were calculated after adjusting for all covariates. We used Stata version 14.2 (StataCorp., College Station, TX, USA) for all analyses, with a 2-tailed significance level set at 5%.
Ethical consideration
The process of obtaining informed consent in the present study was as follows: the questionnaire was sent by mail with the explanation of the study; the participants read the written explanation about the purpose of study and replied. Hence, we considered that informed consent was provided by those who replied and sent back the questionnaire. The JAGES protocol in 2016 was approved by the ethics committee of National Center for Geriatrics and Gerontology (No. 992) and the ethics committee of Chiba University (No. 2493). We followed the STROBE Statement to report our observational study. This study was performed in accordance with the principles of the Declaration of Helsinki. Informed consent was obtained from all participants.