Confounding is a major concern in epidemiology. Despite its significance, the different notions of confounding have not been fully appreciated in the literature, leading to confusion of causal concepts in epidemiology. In this article, we aim to highlight the importance of differentiating between the subtly different notions of confounding from the perspective of counterfactual reasoning. By using a simple example, we illustrate the significance of considering the distribution of response types to distinguish causation from association, highlighting that confounding depends not only on the population chosen as the target of inference, but also on the notions of confounding in distribution and confounding in measure. This point has been relatively underappreciated, partly because some literature on the concept of confounding has only used the exposed and unexposed groups as the target populations, while it would be helpful to use the total population as the target population. Moreover, to clarify a further distinction between confounding “in expectation” and “realized” confounding, we illustrate the usefulness of examining the distribution of exposure status in the target population. To grasp the explicit distinction between confounding in expectation and realized confounding, we need to understand the mechanism that generates exposure events, not the product of that mechanism. Finally, we graphically illustrate this point, highlighting the usefulness of directed acyclic graphs in examining the presence of confounding in distribution, in the notion of confounding in expectation.
Background: Childhood attention deficit hyperactivity disorder (ADHD) is reported to be more prevalent among socioeconomically disadvantaged groups in various countries. The effect of poverty on child development appears to depend on how long poverty lasts. The timing of poverty also seems to be important for childhood outcomes. Lifetime socioeconomic status may shape current health. Thus, we investigated the effects of household income changes from birth to 4 years on the occurrence of ADHD.
Methods: Data were obtained from 18,029 participants in the Korean National Health Insurance cohort who were born in 2002 and 2003. All individuals were followed until December 2013 or the occurrence of ADHD, whichever came first. Household income trajectories were estimated using the national health insurance premium and the group-based model. Cox proportional hazard models were used to compare incidence rates between different income trajectory groups after adjustment for possible confounding risk factors.
Results: Of 18,029 participants, 554 subjects (3.1%) were identified as having ADHD by age 10 or 11. Seven household income trajectories within three categories were found. Children living in decreasing, consistently low, and consistently mid-low income households had an increased risk of ADHD compared to children who consistently lived in the mid-high household income group.
Conclusions: Children who live in decreasing-income or consistently low-income households have a higher risk for ADHD. Promotion of targeted policies and priority support may help reduce ADHD in this vulnerable group.
Background: Studies on the associations between persistent organic pollutants (POPs) and smoking according to gender and smoking amount (cigarettes/day) are limited, and the results regarding the relationship between POPs and smoking are not completely consistent across studies.
Objectives: The smoking rate in Korea is one of the highest among the Organization for Economic Cooperation and Development (OECD) countries. We investigated the association between serum concentrations of POPs and cigarette smoking in Koreans by smoking status (never-smoker/ever-smoker) and smoking amount (cigarettes/day) according to gender.
Methods: Serum concentrations of 32 polychlorinated biphenyls (PCBs) and 19 organochlorine pesticides (OCPs) were measured in 401 participants (232 men and 169 women) who received health examinations during the Korean Cancer Prevention Study-II. We compared POP levels in ever-smokers and never-smokers and conducted multivariate logistic regression analyses to identify associations between POPs and smoking.
Results: Among women, the concentrations of PCB 156, PCB 167, and PCB 180 were significantly higher in ever-smokers than in never-smokers. After adjustments for age, body mass index, gamma-glutamyl transpeptidase, and alcohol intake, serum PCB 157 concentration was positively associated with male ever-smokers (OR 2.26; 95% CI, 1.01–5.04). In addition, trans-nonachlordane in OCPs as well as PCBs was significantly positively related with female ever-smokers (OR 3.21; 95% CI, 1.04–9.86). We found that subjects who smoked fewer than 15 cigarettes/day had a higher risk of having high POP concentrations than never-smokers.
Conclusions: These results indicate that smoking may be associated with human serum POPs levels.
Background: Development of periodontal disease (PD) may be affected by socioeconomic status. This study examined the relationship between occupational status and PD in a 5-year prospective cohort of Japanese workers.
Methods: In total, 19,633 participants had initial examinations at the Aichi Health Promotion Foundation, of whom 8210 participants aged 20 years or older did not have PD. Follow-up examinations were conducted for 3757 participants, accounting for 45.8% of baseline participants. Ultimately, 3390 participants were analyzed according to the criterion of job classification at baseline, which was based on the International Standard Classification of Occupations, 1987. Oral examinations were performed using the Community Periodontal Index (CPI). The CPI scores were coded as follows: healthy (score of 0); bleeding after probing (1); dental calculus (2); shallow pockets (3); and deep pockets (4). Participants with one or more sextants with a score >2 were diagnosed with PD. Poisson regression analysis was performed to adjust for age and other potential confounders.
Results: Overall, 31.6% of men and 23.8% of women had developed PD (CPI scores of 3 or 4). The adjusted relative risk (RR) for PD (CPI scores of 3 or 4) in men was not significant. On the other hand, the adjusted RRs for PD (CPI score of 4) in men were 2.52-, 2.39-, and 2.74-fold higher for skilled workers, sales persons, and drivers, respectively, than for professionals. In contrast, we found no gradient in women.
Conclusions: We found a gradient related to the risk of developing PD according to occupational status among men in a Japanese worker population.
Background: Healthy life expectancy (HLE) is used as one of the primary objectives of fundamental health promotion plans and social development plans. Activity limitation is used to calculate HLE, but little study has been done to identify determinants of activity limitation in order to extend HLE. The purpose of this study is to identify diseases and injuries that commonly lead to activity limitation to prioritize countermeasures against activity limitation.
Methods: We used anonymous data from the 2007 “Comprehensive Survey of Living Conditions,” collected by the Ministry of Health, Labour and Welfare of Japan according to the Statistics Act, Article 36. We used logistic regression analyses and calculated odds ratios (ORs) after adjusting for age and sex. Limitation in daily activities was applied as the dependent variable, and each disease/injury was applied as an independent variable in this analysis. Furthermore, population attributable fractions (PAFs) were calculated.
Results: The provided data included 98,789 subjects. We used data for 75,986 valid subjects aged 12 years or older. The following diseases showed high PAF: backache (PAF 13.27%, OR 3.88), arthropathia (PAF 7.61%, OR 4.82), eye and optical diseases (PAF 6.39%, OR 2.01), and depression and other mental diseases (PAF 5.70%, OR 11.55). PAFs of cerebrovascular diseases, hypertension, and diabetes were higher for males than for females; on the other hand, PAFs of orthopedic diseases were higher among females.
Conclusions: Our results indicate that orthopedic diseases, ophthalmic diseases, and psychiatric diseases particularly affect activity limitation.
Background: Health literacy has been increasingly recognized as one of the most important social determinants for health. However, an appropriate and comprehensive assessment tool is not available in many Asian countries. This study validates a comprehensive health literacy survey tool European health literacy questionnaire (HLS-EU-Q47) for the general public in several Asian countries.
Methods: A cross-sectional survey based on multistage random sampling in the target countries. A total of 10,024 participants aged ≥15 years were recruited during 2013–2014 in Indonesia, Kazakhstan, Malaysia, Myanmar, Taiwan, and Vietnam. The questionnaire was translated into local languages to measure general health literacy and its three domains. To evaluate the validity of the tool in these countries, data were analyzed by confirmatory factor analysis, internal consistency analysis, and regression analysis.
Results: The questionnaire was shown to have good construct validity, satisfactory goodness-of-fit of the data to the hypothetical model in three health literacy domains, high internal consistency (Cronbach's alpha >0.90), satisfactory item-scale convergent validity (item-scale correlation ≥0.40), and no floor/ceiling effects in these countries. General health literacy index score was significantly associated with level of education (P from <0.001 to 0.011) and perceived social status (P from <0.001 to 0.016), with evidence of known-group validity.
Conclusions: The HLS-EU-Q47 was a satisfactory and comprehensive health literacy survey tool for use in Asia.