Study cohort
Data were collected as part of the National Institute for Longevity Sciences–Longitudinal Study of Aging (NILS-LSA). In this project, the normal aging process has been assessed over time using detailed questionnaires, medical check-ups, anthropometric measurements, physical fitness tests, and nutritional examinations. Participants in the NILS-LSA included randomly selected age- and sex-stratified individuals from the non-institutionalized residents in the institute neighbourhood areas of Obu City and Higashiura Town in Aichi Prefecture in Japan. The first wave of the NILS-LSA was conducted from 1997 to 2000 and included 2267 participants (1139 men, 1128 women; age range, 40–79 years). Details of the NILS-LSA study have been reported elsewhere [17]. Subjects were followed-up every 2 years from the second to seventh wave (2000–2012).
Follow-up survey and vital statistics records
In July 2013, a self-administered questionnaire was mailed to participants to assess health status, including a “requirement for long-term care” under the new long-term care insurance system that started in Japan in 2000 [18, 19]. In addition, we obtained death records for all participants and obtained information from local government regarding which participants had moved to other areas. To clarify causes of death, we used National Vital Statistics records that were available until the end of December 2014. Mortality or disability was defined according to the National Vital Statistics records, or self-reported long-term care insurance certification, respectively. The main outcome in this study was “composite outcome for mortality or disability” as we combined these outcomes to increase the statistical power (to increase the number of cases).
Study subjects
Among the 2267 participants who participated in the first wave, we excluded 289 patients with a history of Parkinson’s disease (n = 5) or for whom data were missing (n = 284) (Fig. 1). Of the 1978 participants still being followed as of 2014, 389 men and women were categorized as having died (n = 299) or as needing long-term care insurance certification (n = 90). Of the 1589 men and women categorized as “censored,” 1481 were confirmed as alive according to information from the local government, and 108 had moved away from the local area or dropped out.
The study protocol was reviewed and approved by the Committee of Ethics of Human Research at the National Center for Geriatrics and Gerontology. Written informed consent was obtained from all subjects during the first to the seventh wave of the NILS-LSA. In the survey mailed in 2013, we explained that returning the self-administered questionnaire represented informed consent. Death data were obtained by means of secondary usage of demographic statistics by predetermined procedures. We used anonymous death data that had already been collected by the Ministry of Health, Labour and Welfare in Japan and do not require consent of individuals.
Assessment of muscle mass
ALM (kg), which represents the appendicular fat-free mass minus the bone mineral content [10], was assessed using a QDR-4500 dual-energy X-ray absorptiometry (DXA) system (Hologic, Bedford, MA).
We calculated 5 indices using ALM (kg): ALM divided by leg length (kg/m); ALM divided by height (kg/m); ALM divided by height squared (kg/m2); ALM divided by weight and multiplied by 100 (kg/kg*100); and ALM divided by BMI and multiplied by 10 (kg/kg/m2*10). BMI was calculated as weight divided by height squared.
Other measurements
History of stroke, hypertension, hyperlipidemia, heart disease, and diabetes (past and current), current smoking status (yes/no), education (≤9 or ≥10 years of school), and annual family income (<5,500,000 or ≥5,500,000 yen per year) were collected using self-reported questionnaires, and medical doctors or trained staff confirmed the information [20]. Sarcopenia was defined according to AWGS [6] criteria using grip strength, gait speed, and muscle mass. All measurements were assessed in the first-wave survey of the NILS-LSA.
Statistical analysis
Sex differences in baseline characteristics of participants were analysed using the t test or χ2 test. The difference in the prevalence of mortality or disability by sex-stratified quintiles of each ALM index (ALM and adjusted ALMs) or sarcopenia-related indices was analysed using the χ2 test. Crude and adjusted Cox proportional hazard models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for mortality or disability by sex-stratified quintiles of each ALM index (ALM and adjusted ALMs) or sarcopenia-related indices. For Cox models, follow-up time (years) was calculated by the duration (days) that had elapsed since the day on which each participant entered the first wave of the NILS-LSA. The last day of follow-up for each participant was used for analysis, as either the date of death, the earliest day of needing long-term care (event group), the latest day of last participation in the NILS-LSA, or December 2014, whichever came first (censored group). When the participants were missing, that is, they moved away from the local area or dropped out (n = 108), we considered the last participation day of the NILS-LSA as the last day of follow-up.
Variables considered for adjustment were age, smoking status, education, family income, history of stroke, hypertension, heart disease, hyperlipidemia, and diabetes mellitus (multivariate-adjusted). In sub-analyses, we calculated receiver-operating-characteristic (ROC) curves on disability or mortality according to the sex-stratified quintiles of each ALM index (ALM and adjusted ALMs). The area under the ROC curve (AUC) was calculated with the multivariate-adjusted logistic regression model.
Probability levels of <.05 and <.10 were considered significant and marginally significant, respectively. All statistical analyses were conducted using Statistical Analysis System software version 9.3 (SAS Institute, Cary, NC).
Additional analyses
To clarify age-related changes in each ALM index (ALM alone and ALM with each of the 5 adjustments), we attempted to describe trends in changes to these indices over 12 years according to age in the first wave of the study of the NILS-LSA.
Among the 2267 participants (age range, 40–79 years) who participated in the first wave, we excluded patients with a history of Parkinson’s disease (n = 5) or for whom data were missing (n = 96) (Fig. 1). We selected subjects who also participated in more than one study wave from the second to the seventh wave, because variables could be followed-up at least once from the first wave. A total of 1838 subjects (951 men, 887 women) who were in the first wave were available for analysis. Each wave was conducted every 2 years. Mean (SD) interval and participation times between the first and last waves of participation for each participant were 5.5 (4.1) years and 3.7 (2.0) times, respectively.
For repeated-measures analyses of each ALM index, a mixed-effects model was used. To estimate fixed effects on each ALM index by follow-up time, both age at baseline and the interaction of follow-up time × age were substituted into the model. To clarify the impact of habitual lifestyles on each ALM index, we additionally adjusted the mixed effect model using lifestyle-related factors including smoking status, alcohol intake, total physical activity, and energy intake.