1. Study subjects
The study subjects consisted of the participants in the Japan Multi-Institutional Collaborative Cohort Study (J-MICC) Study, which is an ongoing prospective cohort study undertaken in collaboration with universities/research institutions throughout Japan16,17. The J-MICC Study examines how lifestyle habits and genetic factors (e.g., genetic variants) mutually affect the occurrence of lifestyle-related diseases, mainly cancer, after tracking approximately 100,000 subjects for 20 years. It was launched in specific areas of Japan (Chiba, Shizuoka-Sakuragaoka, Shizuoka, Okazaki, Aichi Cancer Center, Daiko, Iga, Takashima, Kyoto, Tokushima, Fukuoka, Saga, Kagoshima, and the Kyusyu and Okinawa Population Study area) in 200517. By the end of 2014, 92,530 men and women of 35–69 years of age were recruited. Most participants were community residents and health check-up examinees.
The study protocol was approved by ethical committees of Nagoya University Graduate School of Medicine and all collaborating universities/institutions (Aichi Cancer Center, Chiba Cancer Center, Nagoya City University Graduate School of Medical Sciences, Shiga University of Medical Science, Kyoto Prefectural University of Medicine, Kyushu University Graduate School of Medical Sciences, Saga University Faculty of Medicine, Kagoshima University Graduate School of Medical and Dental Sciences, Tokushima University Graduate School of Biomedical Sciences, and University of Shizuoka). Written informed consent was obtained from all participants. The present study was conducted according to the principles expressed in the World Medical Association Declaration of Helsinki.
2. Baseline survey
This cross-sectional study used data from the baseline survey of the J-MICC Study. The baseline assessment included data collection using a self-administered questionnaire on demographic characteristics (age and gender), psychosocial factors (perceived stress and coping strategies), lifestyle factors (drinking and smoking habits, physical activity, sleeping hours, and dietary habits), and disease history, as well as physical measurements (height and weight) and blood collection.
2.1. Psychosocial factors
Perceived stress was assessed using the question, “How much stress did you feel during the last year?” The subjects were requested to select one of the following answers: 1) “I felt no stress at all,” 2) “I felt little stress,” 3) “I felt moderate stress,” and 4) “I felt much stress.” The level of perceived stress was classified as low for answers 1) and 2), medium for answer 3), and high for answer 4). We used these three levels of perceived stress as categorical or ordinal variables in the subsequent analyses. Although the above measurement of perceived stress is simple, it showed fair one-year reproducibility (weighted k=0.55)18 and was significantly associated with stress-related behaviors (e.g., coping strategies, smoking, physical activity, and sleeping hours18) and urinary cortisol levels19.
For coping strategies, we used five items selected from a dispositional version of the General Coping Questionnaire20 or the Brief Coping Orientation to Problems Experienced21. After the query, “How do you cope with various problems and unfavorable events you experience in daily life?” subjects were requested to answer the frequency (four response categories: “seldom,” “sometimes,” “often,” and “very often”) of each of the following coping strategies: 1) “I express my negative feelings and thoughts” (termed ‘emotional expression’); 2) “I consult with someone close and ask him/her for encouragement” (termed ‘emotional support seeking’ [ESS]); 3) “I try to interpret the problem in a favorable way” (termed ‘positive reappraisal’); 4) “I try hard to solve the problem” (termed ‘problem solving’); and 5) “I let the problem take its own course” (termed ‘disengagement’). The level of each coping strategy was classified as low for the frequency of “seldom”, medium for "sometimes", and high for “often” or “very often”. These three levels were used as categorical or ordinal variables for each coping strategy. The one-year reproducibility, as estimated by the weighted κ statistic, was reported to be 0.41 for emotional expression, 0.49 for ESS, 0.30 for positive reappraisal, 0.48 for problem solving, and 0.31 for disengagement18.
2.2. Covariates
For drinking habit, subjects were classified into never, former, and current drinkers and for current drinkers, total ethanol consumption (g/day) was estimated from the reported consumption frequency and the amounts of alcoholic beverages, as well as beverage-specific ethanol concentrations. Smoking status was categorized as never, former, or current smoker, with further classification by the number of cigarettes per day. Physical activity was estimated as the metabolic equivalent (MET)-hours per week, based on the frequency and duration of daily and leisure time activities22. Sleeping hours per day was asked in an open-ended manner. Energy intake was estimated using a validated short food frequency questionnaire23. Spearman’s correlation coefficient between estimated energy intake by the questionnaire and that by 3-day weighed dietary records was 0.36 in men and 0.37 in women24. Subjects were considered to have a history of hypertension, diabetes, or hyperlipidemia if they currently had these conditions or if they had been diagnosed with or treated for these conditions by physicians. Height and weight were measured on the day of the survey and the body mass index (BMI) was calculated as the weight in kilograms divided by the square of height in meters (kg/m2).
2.3. Estimation of the eGFR
Venous blood was sampled for the determination of creatinine and other biochemical measurements on the day of the survey. Serum creatinine was measured at external laboratories using an enzymatic method25. The eGFR (mL/min/1.73 m2) was estimated using the following formula, taking into account serum creatinine (mg/dl), age (years), and gender: 194 ´ creatinine -1.094 ´ age -0.287 for men and this estimate was multiplied by 0.739 for women26.
3.Statistical analysis
From 92,530 participants in the baseline survey, we excluded subjects with the following conditions: missing data on perceived stress (n=1,810) or serum creatinine (n=19,936), creatinine levels of <0.2 or >2.0 mg/dl (n=108), or a history of renal disease (n=34). Consequently, the 70,642 remaining subjects were included in the analysis. The dataset used in the present study was fixed on March 12, 2020. Statistical analyses were performed using SAS (Ver. 9.4 for Windows; SAS Institute, Cary, NC, USA).
All analyses were conducted by gender with adjustment for age because age was strongly associated with both the exposures (e.g., perceived stress) and the outcome (i.e., eGFR), and thereby exerted a large confounding effect. We examined possible associations of perceived stress, coping strategies, and eGFR with covariates using the age-adjusted Spearman's rank correlation coefficient (r). In our main analyses, multiple regression models were run with the eGFR as a dependent variable and each of perceived stress and coping strategies as the main independent variables. The P value for trend was based on the statistical significance of each psychosocial variable as an ordinal variable. The following four models were constructed: 1) Model 1 was adjusted for age and study area, 2) Model 2 was additionally adjusted for lifestyle factors (drinking, smoking, physical activity, sleeping hours, energy intake, and BMI), 3) Model 3 was additionally adjusted for perceived stress and coping strategies, and 4) Model 4 was additionally adjusted for the history of hypertension, diabetes, and hyperlipidemia. We regarded the results in Model 3 as the main effects of psychosocial variables because the history of hypertension, diabetes, and hyperlipidemia included in Model 4 may represent the main mediators linking these variables to the eGFR.
When a significant association was found between a psychosocial variable and the eGFR in the above analyses, we estimated the adjusted mean (and 95% confidence interval [CI]) of the eGFR according to that variable in both gender, with the LSMEANS statement of the GLM procedure of SAS. We also examined whether an interaction existed between perceived stress and each coping strategy and the eGFR by including a corresponding interaction term in the above multiple regression models. When a significant interaction was detected, a stratified analysis was conducted to estimate the adjusted mean eGFR according to perceived stress and an identified coping strategy. All P values reported were two-tailed, and P values of <0.05 were considered statistically significant.