著者

Etsuji Suzuki Tomohiro Shinozaki Eiji Yamamoto

出版者

Japan Epidemiological Association

雑誌

Journal of Epidemiology (ISSN:09175040)

巻号頁・発行日

pp.JE20190192, (Released:2020-02-01)

参考文献数

70

被引用文献数

9 36

---

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.

著者

Tomohiro Shinozaki Etsuji Suzuki

出版者

Japan Epidemiological Association

雑誌

Journal of Epidemiology (ISSN:09175040)

巻号頁・発行日

vol.30, no.9, pp.377-389, 2020-09-05 (Released:2020-09-05)

参考文献数

84

被引用文献数

13 20

---

Epidemiologists are increasingly encountering complex longitudinal data, in which exposures and their confounders vary during follow-up. When a prior exposure affects the confounders of the subsequent exposures, estimating the effects of the time-varying exposures requires special statistical techniques, possibly with structural (ie, counterfactual) models for targeted effects, even if all confounders are accurately measured. Among the methods used to estimate such effects, which can be cast as a marginal structural model in a straightforward way, one popular approach is inverse probability weighting. Despite the seemingly intuitive theory and easy-to-implement software, misunderstandings (or “pitfalls”) remain. For example, one may mistakenly equate marginal structural models with inverse probability weighting, failing to distinguish a marginal structural model encoding the causal parameters of interest from a nuisance model for exposure probability, and thereby failing to separate the problems of variable selection and model specification for these distinct models. Assuming the causal parameters of interest are identified given the study design and measurements, we provide a step-by-step illustration of generalized computation of standardization (called the g-formula) and inverse probability weighting, as well as the specification of marginal structural models, particularly for time-varying exposures. We use a novel hypothetical example, which allows us access to typically hidden potential outcomes. This illustration provides steppingstones (or “tips”) to understand more concretely the estimation of the effects of complex time-varying exposures.

著者

Tomohiro Shinozaki Etsuji Suzuki

出版者

Japan Epidemiological Association

雑誌

Journal of Epidemiology (ISSN:09175040)

巻号頁・発行日

pp.JE20200226, (Released:2020-07-18)

参考文献数

84

被引用文献数

20

---

Epidemiologists are increasingly encountering complex longitudinal data, in which exposures and their confounders vary during follow-up. When a prior exposure affects the confounders of the subsequent exposures, estimating the effects of the time-varying exposures requires special statistical techniques, possibly with structural (i.e., counterfactual) models for targeted effects, even if all confounders are accurately measured. Among the methods used to estimate such effects, which can be cast as a marginal structural model in a straightforward way, one popular approach is inverse probability weighting. Despite the seemingly intuitive theory and easy-to-implement software, misunderstandings (or “pitfalls”) remain. For example, one may mistakenly equate marginal structural models with inverse probability weighting, failing to distinguish a marginal structural model encoding the causal parameters of interest from a nuisance model for exposure probability, and thereby failing to separate the problems of variable selection and model specification for these distinct models. Assuming the causal parameters of interest are identified given the study design and measurements, we provide a step-by-step illustration of generalized computation of standardization (called the g-formula) and inverse probability weighting, as well as the specification of marginal structural models, particularly for time-varying exposures. We use a novel hypothetical example, which allows us access to typically hidden potential outcomes. This illustration provides steppingstones (or “tips”) to understand more concretely the estimation of the effects of complex time-varying exposures.

著者

Masao Iwagami Tomohiro Shinozaki

出版者

Society for Clinical Epidemiology

雑誌

Annals of Clinical Epidemiology (ISSN:24344338)

巻号頁・発行日

vol.4, no.2, pp.33-40, 2022 (Released:2022-04-04)

参考文献数

27

被引用文献数

19

---

Matching is a technique through which patients with and without an outcome of interest (in case-control studies) or patients with and without an exposure of interest (in cohort studies) are sampled from an underlying cohort to have the same or similar distributions of some characteristics. This technique is used to increase the statistical efficiency and cost efficiency of studies. In case-control studies, besides time in risk set sampling, controls are often matched for each case with respect to important confounding factors, such as age and sex, and covariates with a large number of values or levels, such as area of residence (e.g., post code) and clinics/hospitals. In the statistical analysis of matched case-control studies, fixed-effect models such as the Mantel-Haenszel odds ratio estimator and conditional logistic regression model are needed to stratify matched case-control sets and remove selection bias artificially introduced by sampling controls. In cohort studies, exact matching is used to increase study efficiency and remove or reduce confounding effects of matching factors. Propensity score matching is another matching method whereby patients with and without exposure are matched based on estimated propensity scores to receive exposure. If appropriately used, matching can improve study efficiency without introducing bias and could also present results that are more intuitive for clinicians.

著者

Takahiro Tabuchi Sho Fujihara Tomohiro Shinozaki Hiroyuki Fukuhara

出版者

Japan Epidemiological Association

雑誌

Journal of Epidemiology (ISSN:09175040)

巻号頁・発行日

pp.JE20170163, (Released:2018-05-19)

参考文献数

38

被引用文献数

11

---

Background: Our objective in this study was to find determinants of high-school dropout in a deprived area of Japan using longitudinal data, including socio-demographic and junior high school-period information.Methods: We followed 695 students who graduated the junior high school located in a deprived area of Japan between 2002 and 2010 for 3 years after graduation (614 students: follow-up rate, 88.3%). Multivariable log-binomial regression models were used to calculate the prevalence ratios (PRs) for high-school dropout, using multiple imputation (MI) to account for non-response at follow-up.Results: The MI model estimated that 18.7% of students dropped out of high school in approximately 3 years. In the covariates-adjusted model, three factors were significantly associated with high-school dropout: ≥10 days of tardy arrival in junior high school (PR 6.44; 95% confidence interval [CI], 1.69–24.6 for “10–29 days of tardy arrival” and PR 8.01; 95% CI, 2.05–31.3 for “≥30 days of tardy arrival” compared with “0 day of tardy arrival”), daily smoking (PR 2.01; 95% CI, 1.41–2.86) and severe problems, such as abuse and neglect (PR 1.66; 95% CI, 1.16–2.39). Among students with ≥30 days of tardy arrival in addition to daily smoking or experience of severe problems, ≥50% high-school dropout rates were observed.Conclusions: Three determinants of high-school dropout were found: smoking, tardy arrival, and experience of severe problems. These factors were correlated and should be treated as warning signs of complex behavioral and academic problems. Parents, educators, and policy makers should work together to implement effective strategies to prevent school dropout.

著者

Satomi Odani Tomohiro Shinozaki Kenji Shibuya Takahiro Tabuchi

出版者

Japan Epidemiological Association

雑誌

Journal of Epidemiology (ISSN:09175040)

巻号頁・発行日

vol.32, no.4, pp.195-203, 2022-04-05 (Released:2022-04-05)

参考文献数

31

被引用文献数

7

---

Background: Coronavirus disease 2019 (COVID-19) has disproportionately affected the most vulnerable populations. We assessed the prevalence and disparities of economic hardships and their impact on health deterioration in Japan.Methods: Data were obtained from a nation-wide, cross-sectional, internet-based, self-reported survey conducted during August–September, 2020 with individuals aged 15–79 years in Japan (n = 25,482). Economic hardships and changes in various physical and mental health status were measured using sample-weighted data. Adjusted prevalence ratios (APRs) were estimated to investigate the associations between economic hardships and health outcomes.Results: During April–September, 2020 in Japan, 25.0%, 9.6%, 7.9%, and 3.1% of the respondents experienced income loss, money shortage, financial anxiety and financial exploitation, respectively, with higher prevalence among workers (vs non-workers). Stratifying by sex and working status, income loss was associated with physical health deterioration (APRs ranged from 1.45–1.95), mental health deterioration (APRs ranged from 1.47–1.68), and having serious psychological distress (APRs ranged from 1.41–2.01) across all strata. Shortage of money and financial anxiety were also associated with increased likelihood of all adverse health outcomes assessed, regardless of whether the hardships were pre-existing or experienced first time. Among non-working individuals, financial exploitation was associated with physical health deterioration among males (APR 1.88) and mental health deterioration among both males (APR 1.80) and females (APR 2.23), while such associations were not observed among working individuals.Conclusions: During the early phase of the COVID-19 epidemic, COVID-19-related economic hardships were associated with physical and mental health deterioration in Japan, particularly among the vulnerable populations. Timely and prompt responses are warranted to mitigate both economic and health burdens.

著者

Miwa Yamaguchi Yosuke Inoue Tomohiro Shinozaki Masashige Saito Daisuke Takagi Katsunori Kondo Naoki Kondo

出版者

Japan Epidemiological Association

雑誌

Journal of Epidemiology (ISSN:09175040)

巻号頁・発行日

vol.29, no.10, pp.363-369, 2019-10-05 (Released:2019-10-05)

参考文献数

46

被引用文献数

39

---

Background: This study aimed to examine the contextual effects of community-level social capital on the onset of depressive symptoms using a longitudinal study design.Methods: We used questionnaire data from the 2010 and 2013 waves of the Japan Gerontological Evaluation Study that included 14,465 men and 14,600 women aged over 65 years from 295 communities. We also used data of a three-wave panel (2006–2010–2013) to test the robustness of the findings (n = 7,424). Using sex-stratified multilevel logistic regression, we investigated the lagged associations between three scales of baseline community social capital and the development of depressive symptoms.Results: Community civic participation was inversely associated with the onset of depressive symptoms (men: adjusted odds ratio [AOR] 0.93; 95% confidence interval [CI], 0.88–0.99 and women: AOR 0.94; 95% CI, 0.88–0.997 per 1 standard deviation unit change in the score), while no such association was found in relation to the other two scales on social cohesion and reciprocity. This association was attenuated by the adjustment of individual responses to the civic participation component. Individual-level scores corresponding to all three community social capital components were significantly associated with lower risks for depressive symptoms. The results using the three-wave data set showed statistically less clear but similar associations.Conclusions: Promoting environment and services enhancing to community group participation might help mitigate the impact of late-life depression in an aging society.