Study design and participants
From July 1 to July 9, 2019, we conducted face-to-face surveys with 539 individuals (female = 447, male = 92), who were aged ≥ 40 years and resided in Settsu city, Osaka prefecture. The participants were recruited either when they participated in a specific health examination conducted at the Settsu Health Center or via Settsu’s public relations magazine and word of mouth. Details of study design and participants are described in our previous publication . Of these participants, we excluded participants who lived outside Settsu ity (n = 10), missed skeletal muscle mass data (n = 2), refused to participate (n = 1), and were aged less than 40 (n = 1). Ultimately, this cross-sectional study included 525 adults (436 females and 89 males) aged 40 to 91. Before the study began, participants provided written consent to participate after receiving information about the procedures and purpose of the study. The study protocol was approved by the Ethics Review Board of the National Institute of Health and Nutrition (Ikikenhatsu 178-1). This study was carried out in accordance with the principles outlined in the Declaration of Helsinki and was registered with the Japan Clinical Trials Registry (UMIN000036880).
Each participant was asked to visit a public building in Settsu city once between 10 AM and 4 PM and complete questionnaires on daily life (e.g., about dwelling and physical activity) before collecting each one’s objective weight and height measurements. To obtain self-reported body weight and height measurements, participants were asked, “What is your current body weight and height?” in self-administered questionnaires. They filled out self-reported height and weight using 0.1 cm and 0.1 kg units, respectively. Other health-related variables, which include smoking habit, drinking habit, exercise habit, self-rated health, self-rated physical fitness, and self-rated economic condition, were assessed using a self-administered questionnaire. Details of these questionnaires are described in our previous study . The heights of the participants were then objectively measured to the closest 0.1 cm using an analog height meter. Body weights were measured, and appendicular lean mass (ALM) was estimated using the multifrequency bioelectrical impedance analysis (MF-BIA) (MC-780A-N, TANITA, Tokyo, Japan), which were then validated using dual-energy X-ray absorptiometry (DXA) . Participants were evaluated in their underwear and were asked to stand barefoot on toe-and-heel electrodes while holding handgrips, with their arms hanging down a few centimeters from the hips. The details of the MF-BIA device used were previously described. All the procedures were conducted in July 2019.
Skeletal muscle mass measurements and definition of low muscle mass
We calculated the skeletal muscle index (SMI) as follows: SMI = ALM/height2. In this study, participants with lower SMI values for each sex were categorized as having “sarcopenia with low skeletal muscle mass.” According to the Asian Working Group for Sarcopenia, the cutoff SMI values using the bioelectrical impedance analysis (BIA) method were 7.0 and 5.7 kg/m2 in male and female, respectively .
Weight misperception was defined as the difference between self-reported and objectively measured values of weight and calculated as follows in our study. First, we calculated the difference of weight (kg) by subtracting objectively measured values from self-reported values of weight based on Ikeda’s study . Next, to calculate the difference (%) against the objectively measured value, the difference obtained was then divided by the objectively measured value and multiplied by 100 (represented as %). For example, weight misperception of an individual with self-reported weight = 65.0 kg and objectively measured weight = 66.0 kg can be calculated as follows: −1.0 kg ÷ 66.0 kg × 100 % = −1.51%.
For descriptive statistics, values of continuous variables were expressed as mean and standard deviation, or as median and interquartile range, while values of categorical variables were expressed as percentages (%). Continuous and ordinal participant characteristics were classified as follows: exercise habit (“Do you go walking or engage in other exercise at least once per week?” “yes” or “no”), smoking habit (“almost daily” or “sometimes” = yes, “used to, but quit,” or “never” = no), drinking habit (“almost daily,” or “sometimes,” = yes, and “almost never” or “never” = no); self-reported health (“very healthy” or “somewhat healthy” = good self-reported health, and “not very healthy” or “unhealthy” = poor self-reported health); self-reported physical fitness (“extremely confident” or “somewhat confident” = good self-reported physical fitness, and “slightly anxious” or “very anxious” = poor self-reported physical fitness); and self-reported socioeconomic status (“easy” or “somewhat easy” = high socioeconomic status, and “somewhat hard” or “hard” = low socioeconomic status). These variables were classified with reference to covariates used in our previous study , and these models were decided with reference to covariates used in previous studies, which examined the association between weight misperception on HRQoL .
Second, participants were first classified into sex-specific tertiles based on extent of weight misperception (under-reporters, acceptable reporters, and over-reporters) . In order to evaluate the relationship between weight misperception and prevalence of low muscle mass, we used a logistic regression model to calculate the sex, age, and multivariable adjusted ORs, as well as 95% CI. In the logistic regression model, sex and age were inputted as covariates to calculate the adjusted ORs. To calculate the multivariable-adjusted ORs, continuous variables (i.e., sex, age, and BMI) were inputted into the logistic regression model as covariates (model 1). Then, exercise habit, smoking, drinking habit, self-reported health, self-reported physical fitness, and self-reported socioeconomic status were added to model 1 to calculate the multivariable-adjusted OR (model 2). Furthermore, linear regression analyses were conducted to obtain the correlation between the means of self-reported and objectively measured height, weight, and BMI. We used Microsoft Office Excel 2017 and PASW Statistics version 20.0 (SPSS, IBM Corp., Armonk, N.Y., USA) for data processing and statistical analyses, respectively. A two-tailed p-value lower than 5% was considered statistically significant.