Graphical models are useful tools in causal inference, and causal directed acyclic graphs (DAGs) are used extensively to determine the variables for which it is sufficient to control for confounding to estimate causal effects. We discuss the following ten pitfalls and tips that are easily overlooked when using DAGs: 1) Each node on DAGs corresponds to a random variable and not its realized values; 2) The presence or absence of arrows in DAGs corresponds to the presence or absence of individual causal effect in the population; 3) “Non-manipulable” variables and their arrows should be drawn with care; 4) It is preferable to draw DAGs for the total population, rather than for the exposed or unexposed groups; 5) DAGs are primarily useful to examine the presence of confounding in distribution in the notion of confounding in expectation; 6) Although DAGs provide qualitative differences of causal structures, they cannot describe details of how to adjust for confounding; 7) DAGs can be used to illustrate the consequences of matching and the appropriate handling of matched variables in cohort and case-control studies; 8) When explicitly accounting for temporal order in DAGs, it is necessary to use separate nodes for each timing; 9) In certain cases, DAGs with signed edges can be used in drawing conclusions about the direction of bias; and 10) DAGs can be (and should be) used to describe not only confounding bias but also other forms of bias. We also discuss recent developments of graphical models and their future directions.
Background: Serum gamma-glutamyltransferase has been recognized as the risk factor of cardiovascular and metabolic diseases. However, the association between serum gamma-glutamyltransferase and the risk of chronic kidney disease is not well known, and no prospective studies have examined separately the relationship of serum gamma-glutamyltransferase with the risk of proteinuria versus that of low estimated glomerular filtration rate (eGFR).
Methods: We prospectively followed 9,341 Japanese men who did not have low eGFR, proteinuria, or diabetes, and did not take antihypertensive medications at entry for the analysis of proteinuria, and we followed 9,299 men for the analysis of low eGFR. We defined “persistent proteinuria” as proteinuria detected two or more times consecutively and persistently as ≥1+ on urine dipstick at the annual check-up until the end of follow-up. Low eGFR was defined as eGFR <60 mL/min/1.73 m2.
Results: During the 11-year observation period, 151 men developed persistent proteinuria and 1,276 men developed low eGFR. In multivariate models, the highest quartile (≥71 IU/L) of serum gamma-glutamyltransferase was independently related to the development of persistent proteinuria (hazard ratio 3.39; 95% confidence interval, 1.92–5.97) compared with the lowest quartile (≤25 IU/L). In joint analysis of alcohol consumption and serum gamma-glutamyltransferase, non-drinkers in the highest tertile (≥58 IU/L) of serum gamma-glutamyltransferase had the highest risk of persistent proteinuria. However, there was no association between serum gamma-glutamyltransferase and low eGFR.
Conclusion: In middle-aged Japanese men, elevated serum gamma-glutamyltransferase was independently associated with future persistent proteinuria, but not with low eGFR.
Background: Evidence for primary prevention of chronic kidney disease (CKD) is insufficient. The population-based prospective Uonuma CKD cohort study aims to explore associations of lifestyle and other risk factors with CKD. We report here the study design and baseline profiles.
Methods: All 67,322 residents aged ≥40 years in Minamiuonuma City, Uonuma City, and Yuzawa Town, Niigata Prefecture, Japan and 11,406 participants who attended local health-check examinations were targeted for baseline questionnaire and biochemical sampling, respectively. Information was gathered from 43,217 (64.2%) questionnaires and 8,052 (70.6%) biochemical samples; 6,945 participants consented to both questionnaire and biochemical sampling at baseline, conducted between fiscal years 2012 and 2015. Participants provided information regarding sociodemographic, lifestyle, and self-reported outcomes. Urine albumin-to-creatinine ratio (ACR) and estimated glomerular filtration rate (eGFR) were measured. The primary outcome is CKD based on self-report and biochemical/clinical diagnosis.
Results: Mean age of questionnaire respondents was 63.3 (standard deviation [SD], 12.5) years for men and 64.3 (SD, 13.3) years for women. Among participants who submitted urine samples, median ACR was 10.0 (interquartile range [IQR], 5.0–24.0) mg/g for men and 13.0 (IQR, 7.7–27.0) mg/g for women, and median eGFR was 73.6 mL/min/1.73 m2 (IQR, 63.5–84.5) for men and 73.5 mL/min/1.73 m2 (IQR, 64.4–83.5) for women. ACR 30 mg/g or more was found in 1,741 participants (21.7%) and eGFR <60 mL/min/1.73 m2 in 1,361 participants (16.9%).
Conclusion: The Uonuma CKD cohort study was established to investigate the impact of lifestyle on CKD development and to provide data for preventing the onset and progression of CKD.
Background: MicroRNAs (miRNAs) play crucial roles in the development of various diseases, including chronic kidney disease (CKD). Although previous studies in clinically severe patients have investigated associations between CKD and miRNAs, with particular attention on renal fibrosis, relationships in a general population have yet to be established. The aim of this study was to examine the relationship between expression level of circulating miRNAs and CKD in a middle-aged Japanese population.
Methods: A final total of 513 individuals (216 men and 297 women) who participated in the health check-up program in 2012 were included in our analysis. Quantitative real-time polymerase chain reaction was used to determine expression levels of 22 miRNAs. Estimated glomerular filtration rate (eGFR) was calculated based on serum creatinine level, sex, and age. Participants with eGFR <60 mL/min/1.73 m2 were defined as having CKD.
Results: Three different miRNAs (miR-17, miR-21, and miR-150) showed significant correlations with eGFR after Bonferroni correction and were selected for further analyses. Expression levels of miR-17, miR-21, and miR-150 miRNAs were positively associated with eGFR after adjusting for potential confounders (P = 0.004, 0.002, and 0.004, respectively). Logistic regression analyses showed significantly lower odds ratios for CKD (eGFR <60 mL/min/1.73 m2) in the highest tertile of all three miRNAs (miR-17, miR-21, and miR-150) compared with the lowest tertile (P = 0.003, 0.01, and 0.02, respectively).
Conclusions: We found that three circulating miRNAs were significantly associated with CKD in a general Japanese population, which suggested that these miRNAs may be biomarkers for CKD among general adults.
Background: Left atrial dimension (LAD) and other parameters of echocardiography have been reported to be associated with the risk of atrial fibrillation (AF). However, few studies have investigated the associations between echocardiographic parameters and the risk of AF in the Asian general population, which has a low AF incidence.
Methods: A prospective cohort study was performed in 1,424 individuals in the Suita study with echocardiographic parameters, including LAD, and no history of AF. After echocardiography, the participants were followed using 12-lead electrocardiography and questionnaires to detect AF incidence. The multivariable-adjusted hazard ratios (HRs) of echocardiographic parameters for AF incidence were estimated after adjustment for the risk factors of the AF risk score.
Results: During the median 6.0 years of follow-up, 31 AF cases occurred. The multivariable-adjusted HR of a 1-mm increase in LAD for AF was 1.18 (95% confidence interval [CI], 1.08–1.28). The multivariable-adjusted HR for AF of a 1-standard-deviation increase in LAD was higher than that of left ventricular internal dimensions in diastole, left ventricular mass, ejection fraction, and percent fractional shortening, and it was the only significant factor. In 667 participants with both LAD and LA volume (LAV) measurements, LAD and LAV were independently associated with the risk of AF incidence.
Conclusions: LAD on echocardiography was an independent risk factor of incident AF in the Japanese population. LAD might be useful for identifying individuals with a high risk of AF in health check-ups of the general population.
Background: Positive and negative psychological factors are associated with mortality and cardiovascular disease. This study prospectively investigated associations of daily frequency of laughter with mortality and cardiovascular disease in a community-based population.
Methods: This study included 17,152 subjects ≥40 years old who participated in an annual health check in Yamagata Prefecture. Self-reported daily frequency of laughter was grouped into three categories (≥1/week; ≥1/month but <1/week; <1/month). Associations of daily frequency of laughter with increase in all-cause mortality and cardiovascular disease incidence were determined using Cox proportional hazards modeling.
Results: During follow-up (median, 5.4 years), 257 subjects died and 138 subjects experienced cardiovascular events. Kaplan-Meier analysis revealed that all-cause mortality and cardiovascular disease incidence were significantly higher among subjects with a low frequency of laughter (log-rank P < 0.01). Cox proportional hazard model analysis adjusted for age, gender, hypertension, smoking, and alcohol drinking status showed that risk of all-cause mortality was significantly higher in subjects who laughed <1/month than in subjects who laughed ≥1/week (hazard ratio [HR] 1.95; 95% confidence interval [CI], 1.16–3.09). Similarly, risk of cardiovascular events was higher in subjects who laughed ≥1/month but <1/week than in subjects who laughed ≥1/week (HR 1.62; 95% CI, 1.07–2.40).
Conclusion: Daily frequency of laughter represents an independent risk factor for all-cause mortality and cardiovascular disease in a Japanese general population.
Background: It is uncertain whether health counselling after community-based health checkups for high-risk individuals of lifestyle-related disease enhances their referral to physicians.
Methods: We performed a clustered randomized controlled trial of untreated high-risk individuals aged 40 to 74 years who were screened from the annual health checkup in 2014 and 2015 under the national health insurance in 43 municipalities around Japan, assigning 21 intervention and 22 usual care municipalities. The high-risk conditions were severe forms of hypertension, diabetes, dyslipidemia (for men), and proteinuria. For the intervention group, the theory-based health counselling was performed to enhance referrals to physicians, while each municipality performed its own standard counselling for the usual care group. Data on clinical visits and risk factors were collected systematically and anonymously from the databases of health insurance qualification, health insurance claims, and annual health checkups. Hypotheses are that the cumulative proportion of seeing physicians (clinical visits) is higher in the intervention than the usual care groups, and that those in the intervention group have lower cumulative incidence of composite outcomes associated with lifestyle-related diseases.
Results: The numbers of subjects for the analyses were 8,977 in the intervention group and 6,733 in the usual care group. Among them, 6,758 had hypertension, 2,147 had diabetes, 2,861 had dyslipidemia, and 1,221 had proteinuria in the intervention group, with corresponding numbers of 4,833, 1,517, 2,262, and 845, respectively, in the usual care group. There were no material differences in mean levels and proportions of major cardiovascular risk factors between the two groups.
Conclusions: We expect to provide scientific evidence on the effectiveness of health counselling.