BACKGROUND: High-dimensional propensity scoring (HDPS) is a method for empirically identifying potential confounders within large healthcare databases such as administrative claims data. However, this method has not yet been applied to large national health surveys such as the National Health and Aging Trends Study (NHATS), an ongoing nationally representative survey of older adults in the United States and important resource in gerontology research.
METHODS: In this Research Practice article, we present an overview of HDPS and describe the specific data transformation steps and analytic considerations needed to apply it to national health surveys. We applied HDPS within NHATS to investigate the association between self-reported visual difficulty and incident dementia, comparing HDPS to conventional confounder selection methods.
RESULTS: Among 7 207 dementia-free NHATS Wave 1 respondents, 528 (7.3%) had self-reported visual difficulty. In an unadjusted discrete time proportional hazards model accounting for the complex survey design of NHATS, self-reported visual difficulty was strongly associated with incident dementia (odds ratio [OR] 2.34, 95% confidence interval [CI]: 1.95-2.81). After adjustment for standard investigator-selected covariates via inverse probability weighting, the magnitude of this association decreased, but evidence of an association remained (OR 1.44, 95% CI: 1.11-1.85). Adding 75 HDPS-prioritized variables to the investigator-selected propensity score model resulted in further attenuation of the association between visual impairment and dementia (OR 0.94, 95% CI: 0.70-1.23).
CONCLUSIONS: HDPS can be successfully applied to national health surveys such as NHATS and may improve confounder adjustment. We hope developing this framework will encourage future consideration of HDPS in this setting.

BACKGROUND: Mediation analysis is a powerful tool to identify factors mediating the causal pathway of exposure to health outcomes. Mediation analysis has been extended to study a large number of potential mediators in high-dimensional data settings. The presence of confounding in observational studies is inevitable. Hence, it\'s an essential part of high-dimensional mediation analysis (HDMA) to adjust for the potential confounders. Although the propensity score (PS) related method such as propensity score regression adjustment (PSR) and inverse probability weighting (IPW) has been proposed to tackle this problem, the characteristics with extreme propensity score distribution of the PS-based method would result in the biased estimation.
METHODS: In this article, we integrated the overlapping weighting (OW) technique into HDMA workflow and proposed a concise and powerful high-dimensional mediation analysis procedure consisting of OW confounding adjustment, sure independence screening (SIS), de-biased Lasso penalization, and joint-significance testing underlying the mixture null distribution. We compared the proposed method with the existing method consisting of PS-based confounding adjustment, SIS, minimax concave penalty (MCP) variable selection, and classical joint-significance testing.
RESULTS: Simulation studies demonstrate the proposed procedure has the best performance in mediator selection and estimation. The proposed procedure yielded the highest true positive rate, acceptable false discovery proportion level, and lower mean square error. In the empirical study based on the GSE117859 dataset in the Gene Expression Omnibus database using the proposed method, we found that smoking history may lead to the estimated natural killer (NK) cell level reduction through the mediation effect of some methylation markers, mainly including methylation sites cg13917614 in CNP gene and cg16893868 in LILRA2 gene.
CONCLUSIONS: The proposed method has higher power, sufficient false discovery rate control, and precise mediation effect estimation. Meanwhile, it is feasible to be implemented with the presence of confounders. Hence, our method is worth considering in HDMA studies.

While frameworks to systematically assess bias in systematic reviews and meta-analyses (SRMAs) and frameworks on causal inference are well established, they are less frequently integrated beyond the data analysis stages. This paper proposes the use of Directed Acyclic Graphs (DAGs) in the design stage of SRMAs. We hypothesize that DAGs created and registered a priori can offer a useful approach to more effective and efficient evidence synthesis. DAGs provide a visual representation of the complex assumed relationships between variables within and beyond individual studies prior to data analysis, facilitating discussion among researchers, guiding data analysis, and may lead to more targeted inclusion criteria or set of data extraction items. We illustrate this argument through both experimental and observational case examples.

People with mental disorders exhibit increased mortality due to colorectal cancer, despite having a similar incidence to the general population. We aimed to evaluate the extent to which people with mental disorders participate in organised colorectal cancer screening.
We conducted a population-based cohort study of all Danish residents aged 50-74 years who were invited to undergo biennial faecal immunochemical testing between March 1, 2014, and Sept 30, 2018. We used national registry data from all first-time invitees. The primary endpoint was participation within 90 days of invitation. We calculated the proportion who participated and assessed their screening results and adherence to and completeness of follow-up colonoscopy according to their history of mental disorders, classified as none, mild or moderate, or severe. We computed crude and adjusted participation differences in percentage points and participation ratios using the pseudo-observations method.
Of 2 036 704 people who were invited, we included 2 036 352 in the final cohort, of whom 1 008 045 (49·5%) were men and 1 028 307 (50·5%) were women, with a mean age of 60·7 years (SD 8·3, range 49-78). Data on ethnicity were not collected. Compared with people with no mental disorders, the adjusted analysis showed lower participation among people with mild or moderate mental disorders (men: participation difference -4·4 percentage points [95% CI -4·7 to -4·1]; women: -3·8 percentage points [-4·1 to -3·6]) and severe mental disorders (men: participation difference -13·8 percentage points [-14·3 to -13·3]; women: -15·4 percentage points [-15·8 to -14·9]). People with mental disorders had a higher proportion of positive faecal immunochemical test results, lower adherence to colonoscopy, and more incomplete colonoscopies than people without mental disorders.
People with mental disorders were less likely to participate in colorectal cancer screening than those without these disorders. Patients with mental disorders could benefit from support or encouragement from their general practitioner or mental health-care facility to participate in cancer screening. Potential interventions should consider type of mental disorder, as needs might differ.
Danish Cancer Society, Danish Health Foundation.

A vast amount of data regarding the effects of radiation stressors on transcriptional changes has been produced over the past few decades. These data have shown remarkable consistency across platforms and experimental design, enabling increased understanding of early molecular effects of radiation exposure. However, the value of transcriptomic data in the context of risk assessment is not clear and represents a gap that is worthy of further consideration. Recently, benchmark dose (BMD) modeling has shown promise in correlating a transcriptional point of departure (POD) to that derived using phenotypic outcomes relevant to human health risk assessment. Although frequently applied in chemical toxicity evaluation, our group has recently demonstrated application within the field of radiation research. This approach allows the possibility to quantitatively compare radiation-induced gene and pathway alterations across various datasets using BMD values and derive meaningful biological effects. However, before BMD modeling can confidently be used, an understanding of the impact of confounding variables on BMD outputs is needed.
To this end, BMD modeling was applied to a publicly available microarray dataset (Gene Expression Omnibus #GSE23515) that used peripheral blood ex-vivo gamma-irradiated at 0.82 Gy/min, at doses of 0, 0.1, 0.5 or 2 Gy, and assessed 6 hours post-exposure. The dataset comprised six female smokers (F-S), six female nonsmokers (F-NS), six male smokers (M-S), and six male nonsmokers (M-NS).
A combined total of 412 genes were fit to models and the BMD distribution was noted to be bi-modal across the four groups. A total of 74, 41, 62 and 62 genes were unique to the F-NS, M-NS, F-S and M-S groups. Sixty-two BMD modeled genes and nine pathways were common across all four groups. There were no differential sensitivity of BMD responses in the robust common genes and pathways.
For radiation-responsive genes and pathways common across the study groups, the BMD distribution of transcriptional activity was unaltered by sex and smoking status. Although further validation of the data is needed, these initial findings suggest BMD values for radiation relevant genes and pathways are robust and could be explored further in future studies.

Two-stage least square [2SLS] and two-stage residual inclusion [2SRI] are popularly used instrumental variable (IV) methods to address medication nonadherence in pragmatic trials with point treatment settings. These methods require assumptions, e.g., exclusion restriction, although they are known to handle unmeasured confounding. The newer IV-method, nonparametric causal bound [NPCB], showed promise in reducing uncertainty compared to usual IV-methods. The inverse probability-weighted per-protocol [IP-weighted PP] method is useful in the same setting but requires different assumptions, e.g., no unmeasured confounding. Although all of these methods are aimed to address the same nonadherence problem, comprehensive simulations to compare performances of them are absent in the literature.
We performed extensive simulations to compare the performances of the above methods in addressing nonadherence when: (1) exclusion restriction satisfied and no unmeasured confounding, (2) exclusion restriction is met but unmeasured confounding present, and (3) exclusion restriction is violated. Our simulations varied parameters such as, levels of adherence rates, unmeasured confounding, and exclusion restriction violations. Risk differences were estimated, and we compared performances in terms of bias, standard error (SE), mean squared error (MSE), and 95% confidence interval coverage probability.
For setting (1), 2SLS and 2SRI have small bias and nominal coverage. IP-weighted PP outperforms these IV-methods in terms of smaller MSE but produces high MSE when nonadherence is very high. For setting (2), IP-weighted-PP generally performs poorly compared to 2SLS and 2SRI in term of bias, and both-stages adjusted IV-methods improve precision than naive IV-methods. For setting (3), IV-methods perform worst in all scenarios, and IP-weighted-PP produces unbiased estimates and small MSE when confounders are adjusted. NPCB produces larger uncertainty bound width in almost all scenarios. We also analyze a two-arm trial to estimate vitamin-A supplementation effect on childhood mortality after addressing nonadherence.
Understanding finite sample characteristics of these methods will guide future researchers in determining suitable analysis strategies. Since assumptions are different and often untestable for IP-weighted PP and IV methods, we suggest analyzing data using both IP-weighted PP and IV approaches in search of a robust conclusion.

There are usually unknown or unmeasured confounders in the observational study, which is a significant challenge in epidemiological causal association research. This paper presents a tool for identification and effect assessment of unknown/unmeasured confounders in observational studies: probe variables. It can be divided into three forms: exposure probe variable, outcome probe variable, and mediation probe variable. The first two types can identify unknown/unmeasured confounding factors and estimate their size of effect to reveal the real correlation between exposure and outcome. The mediation probe variable controls for \"mediating factors\" to identify unmeasured confounders between exposure and results. The most significant difficulty in this theory\'s practice is selecting and determining \"probe variables.\" Improper probe variables may introduce unknown confounders, which may lead to false identification of unmeasured confounders. Probe variables can be recommended as a sensitivity analysis in observational studies to help readers truly understand the association between exposure and outcomes and to increase the strength of evidence in observational epidemiological studies.
观察性研究中往往存在未知或未测量的混杂因素，是流行病学因果关联研究中的重大挑战。本文介绍一种可以应用在观察性研究中的一种对未知/未测量混杂因素进行识别和效应评估的工具——“探针变量”。其主要可以分为暴露探针变量、结局探针变量以及中介探针3种形式，前2种不仅可以对未知/未测量混杂因素进行识别，也可以对其效应量进行估计，从而揭示真实的暴露与结局之间的关联。而中介探针则是针对“中介因子”进行控制，从而识别暴露和结局之间是否存在未测量混杂因素。该理论实践过程中最大的困难在于“探针变量”的选择和确定，不恰当的“探针变量”可能引入新的混杂，导致未测量混杂因素识别不准确。“探针变量”可以推荐作为观察性研究报告中的一项敏感性分析内容，有助于读者真实理解暴露与结局之间的关联，增加观察性流行病学研究中的证据力度。.

The objective of this study was to examine the association between periodontitis and risk of incident Parkinson\'s disease using large-scale cohort data on the entire population of South Korea. Health checkup data from 6,856,180 participants aged 40 and older were provided by the National Health Insurance Service of South Korea between January 1, 2009, and December 31, 2009, and the data were followed until December 31, 2017. The hazard ratio (HR) of Parkinson\'s disease and 95% confidence interval (CI) were estimated using a Cox proportional hazards model adjusted for potential confounders. The incidence probability of Parkinson\'s disease was positively correlated with the presence of periodontitis. The HR of Parkinson\'s disease for the participants without the need of further dentist visits was 0.96 (95% CI 0.921-1.002); the HR of Parkinson\'s disease increased to 1.142 (95% CI 1.094-1.193) for the individuals who needed further dentist visits. Compared to individuals without periodontitis and without metabolic syndrome, the HR of incident Parkinson\'s disease gradually increased for individuals with periodontitis, with metabolic syndrome, and with both periodontitis and metabolic syndrome. People with periodontitis and metabolic syndrome had the highest HR of incident Parkinson\'s disease, at 1.167 (95% CI 1.118-1.219). In conclusion, a weak association between periodontitis and Parkinson\'s disease was suggested after adjusting for confounding factors from the population-based large-scale cohort of the entire South Korean population.

UNASSIGNED: Out of hospital cardiac arrest (OHCA) remains a leading cause of mortality among adults in the United States. Environmental impact on incidence and outcomes of OHCA has not been fully investigated in recent years. Previous studies showed a possible increase in incidence and mortality in winter season and during seasons with temperature extremes. This study examines seasonal variation in incidence and outcomes of OHCA in the United States.Retrospective study of adult OHCA using the Nationwide Emergency Department Sample was carried out. Monthly incidence rate per 100,000 ED presentations was calculated. Survival rates for each month of admission were examined by hospital region. Multivariate analyses were conducted to determine the effect of the season and month of admission on survival.A total of 122,870 adult OHCA cases presented to emergency departments (EDs) in 2014 and were included. Average incidence of OHCA cases was 147 per 100,000 ED presentations. Overall survival rate in the study population was 5.6% (95% confidence intervals [CI] = 5.4%-5.9%). Patients had an average age of 65.5 (95% CI: 65.3-65.7) years and were mainly men (61.8%). Rates of OHCA presentations were highest during December and January (9.9% and 10.0%) while survival rates were lowest during December (4.6%) and highest in June (6.9%). Regional variation in OHCA outcomes was also noted with highest average survival rate in West (7.8%) and lowest in South (4.3%). After adjusting for confounders including region of hospital, Summer season (Ref: all other seasons), and more specifically month of June (Ref: all other months) were found to be positively associated with survival (OR 1.27, 95% CI [1.07-1.52], P-value = .008) and (OR 1.43, 95% CI [1.08-1.89], P-value = .012 respectively).Incidence and outcomes of out of hospital cardiac arrest presentations to the emergency departments in the United States have seasonal variation. Both incidence and mortality of OHCA increase during colder months, and survival is significantly higher in summer season or in June. Exploring how to use this variation to improve outcomes through refresher training of medical providers or through other mitigation plans is needed.

Studies suggest that high occupational physical activity increases mortality risk. However, it is unclear whether this association is causal or can be explained by a complex network of socioeconomic and behavioural factors. We aimed to examine the association between occupational physical activity and longevity, taking a complex network of confounding variables into account.
In this prospective cohort study, we linked data from Norwegian population-based health examination surveys, covering all parts of Norway with data from the National Population and Housing Censuses and the Norwegian Cause of Death Registry. 437 378 participants (aged 18-65 years; 48·7% men) self-reported occupational physical activity (mutually exclusive groups: sedentary, walking, walking and lifting, and heavy labour) and were followed up from study entry (between February, 1974, and November, 2002) to death or end of follow-up on Dec 31, 2018, whichever came first. We estimated differences in survival time (death from all causes, cardiovascular disease, and cancer) between occupational physical activity categories using flexible parametric survival models adjusted for confounding factors.
During a median of 28 years (IQR 25-31) from study entry to the end of follow-up, 74 203 (17·0%) of the participants died (all-cause mortality), of which 20 111 (27·1%) of the deaths were due to cardiovascular disease and 29 886 (40·3%) were due to cancer. Crude modelling indicated shorter mean survival times among men in physically active occupations than in those with sedentary occupations. However, this finding was reversed following adjustment for confounding factors (birth cohort, education, income, ethnicity, prevalent cardiovascular disease, smoking, leisure-time physical activity, body-mass index), with estimates suggesting that men in occupations characterised by walking, walking and lifting, and heavy labour had life expectancies equivalent to 0·4 (95% CI -0·1 to 1·0), 0·8 (0·3 to 1·3), and 1·7 (1·2 to 2·3) years longer, respectively, than men in the sedentary referent category. Results for mortality from cardiovascular disease and cancer showed a similar pattern. No clear differences in survival times were observed between occupational physical activity groups in women.
Our results suggest that moderate to high occupational physical activity contributes to longevity in men. However, occupational physical activity does not seem to affect longevity in women. These results might inform future physical activity guidelines for public health.
The Norwegian Research Council (grant number 249932/F20).