Exposure-outcome Relationship Research Articles

Exposure range matters: considering non-linear associations in the meta-analysis of environmental pollutant exposure using examples of per- and polyfluoroalkyl substances and birth outcomes.

Meta-analysis is a powerful analytic method for summarizing effect estimates across studies. However, conventional meta-analysis often assumes a linear exposure-outcome relationship and does not account for variability over the exposure ranges. In this work, we first used simulation techniques to illustrate that the linear-based meta-analytical approach may result in oversimplistic effect estimation based on three plausible non-linear exposure-outcome curves (S-shape, inverted U-shape, and M-shape). We showed that subgroup meta-analysis that stratifies on exposure levels can investigate non-linearity and identify the consistency of effect magnitudes in these simulated examples. Next, we examined the heterogeneity of effect estimates across exposure ranges in two published linear-based meta-analyses of prenatal exposure to per- and polyfluoroalkyl substances (PFAS) on changes in mean birth weight or risk of preterm birth. The re-analysis found some varying effect sizes and potential heterogeneity when restricting to different PFAS exposure ranges, but findings were sensitive to the cut-off choices used to rank the exposure levels. Finally, we discussed methodological challenges and recommendations for detecting and interpreting potential non-linear associations in meta-analysis. Using meta-analysis without accounting for exposure range could contribute to literature inconsistency for exposure-induced health effects and impede evidence-based policymaking. Therefore, investigating result heterogeneity by exposure range is recommended.

Is precarious employment an occupational hazard? Evidence from Ontario, Canada

ObjectivesTo examine the association between precarious employment and risk of occupational injury or illness in Ontario, Canada.MethodsWe combined accepted lost-time compensation claims from the Workplace Safety and Insurance Board with...

Methods for extending inferences from observational studies: considering causal structures, identification assumptions, and estimators.

Most prior work in quantitative approaches to generalizability and transportability emphasizes extending causal effect estimates from randomized trials to target populations. Extending findings from observational studies is also of scientific interest, and identifiability assumptions and estimation methods differ from randomized settings when there is selection on both the exposure and exposure-outcome mediators in combination with exposure-outcome confounders (and both confounders and mediators can modify exposure-outcome effects). We argue that this causal structure is common in observational studies, particularly in the field of lifecourse epidemiology, e.g., when extending estimates of the effect of an early-life exposure on a later-life outcome from a cohort enrolled in mid- to late-life. We describe identifiability assumptions and identification using observed data in such settings, highlighting differences from work extending findings from randomized trials. We describe statistical methods, including weighting, outcome modeling, and doubly robust approaches to estimate potential outcome means and verage treatment effects in the target population and illustrate performance of the methods in a simulation study. We show that in the presence of selection into the study sample on both exposure and confounders, estimators must be able to address confounding in the target population. When there is also selection on mediators of the exposure-outcome relationship, estimators need to be able to use different sets of variables to account for selection (including the mediator), and confounding. We discuss conceptual implications of our results, as well as highlight unresolved practical questions for applied work to extend findings from observational studies to target populations.

Causal inference and machine learning in endocrine epidemiology.

With the rapid development of computer science, there is an increasing demand for the use of causal inference methods and machine learning in the research of endocrine disorders and their long-term health outcomes. However, studies on the effective and appropriate applications of these approaches in real-world data and clinical settings are still limited. This review will illustrate the use of causal inference and machine learning in epidemiological research within the field of endocrinology and metabolism. It will examine each concept of causal inference and machine learning through application examples of endocrine disorders. Subsequently, the paper will discuss the integration of machine learning within the causal inference framework, including (i) the estimation of treatment effects or the causal relationship between exposure and outcomes, and (ii) the evaluation of heterogeneity in such treatment effects (or exposure-outcome causal relationship) based on individuals' characteristics. Accurately assessing causal relationships and their heterogeneity across different individuals is crucial not only for determining effective interventions, but also for the appropriate allocation of medical resources and reducing healthcare disparities. By illustrating some application examples in endocrinology, this review aims to enhance readers' understanding and application of causal inference and machine learning in future epidemiological studies focusing on endocrine disorders.

Pharmacovigilance in Pregnancy Studies, Exposures and Outcomes Ascertainment, and Findings from Low- and Middle-Income Countries: A Scoping Review.

Pharmacovigilance (PV), or the ongoing safety monitoring after a medication has been licensed, plays a crucial role in pregnancy, as clinical trials often exclude pregnant people. It is important to understand how pregnancy PV projects operate in low- and middle-income countries (LMICs), where there is a disproportionate lack of PV data yet ahigh burden of adverse pregnancy outcomes. We conducted a scoping review to assess how exposures and outcomes were measured in recently published pregnancy PV projects in LMICs. We utilized a search string, secondary review, and team knowledge to review publications focusing on therapeutic or vaccine exposures among pregnant people in LMICs. We screened abstracts for relevance before conducting a full text review, and documented measurements of exposures and outcomes (categorized as maternal, birth, or neonatal/infant) among other factors, including study topic, setting, and design, comparator groups, and funding sources. We identified 31 PV publications spanning at least 24 LMICs, all focusing on therapeutics or vaccines for infectious diseases, including HIV (n = 17), tuberculosis (TB; n = 9), malaria (n = 7), pertussis, tetanus, and diphtheria (n = 1), and influenza (n = 3). As for outcomes, n = 15, n = 31, and n = 20 of the publications covered maternal, birth, and neonatal/infant outcomes, respectively. Among HIV-specific publications, the primary exposure-outcome relationship of focus was exposure to maternal antiretroviral therapy and adverse outcomes. For TB-specific publications, the main exposures of interest were second-line drug-resistant TB and isoniazid-based prevention therapeutics for pregnant people living with HIV. For malaria-specific publications, the primary exposure-outcome relationship of interest was antimalarial medication exposure during pregnancy and adverse outcomes. Among vaccine-focused publications, the exposure was assessed during a specific time during pregnancy, with an overall interest in vaccine safety and/or efficacy. The study settings were frequently from Africa, designs varied from cohort or cross-sectional studies to clinical trials, and funding sources were largely from high-income countries. The published pregnancy PV projects were largely centered in Africa and concerned with infectious diseases. This may reflect the disease burden in LMICs but also funding priorities from high-income countries. As the prevalence of non-communicable diseases increases in LMICs, PV projects will have to broaden their scope. Birth and neonatal/infant outcomes were most reported, with fewer reporting on maternal outcomes and none on longer-term child outcomes; additionally, heterogeneity existed in definitions and ascertainment of specific measures. Notably, almost all projects covered a single therapeutic exposure, missing an opportunity to leverage their projects to cover additional exposures, add scientific rigor, create uniformity across health services, and bolster existing health systems. For many publications, the timing of exposure, specifically by trimester, was crucial to maternal and neonatal safety. While currently published pregnancy PV literature offer insights into the PV landscape in LMICs, further work is needed to standardize definitions and measurements, integrate PV projects across health services, and establish longer-term monitoring.

Causal relationship between thyroid dysfunction and gastric cancer: a two-sample Mendelian randomization study.

Gastric cancer is one of the most common cancers worldwide, and its development is associated with a variety of factors. Previous observational studies have reported that thyroid dysfunction is associated with the development of gastric cancer. However, the exact relationship between the two is currently unclear. We used a two-sample Mendelian randomization (MR) study to reveal the causal relationship between thyroid dysfunction and gastric cancer for future clinical work. This study is based on a two-sample Mendelian randomization design, and all data are from public GWAS databases. We selected hyperthyroidism, hypothyroidism, free thyroxine (FT4), and thyroid-stimulating hormone (TSH) as exposures, with gastric cancer as the outcome. We used three statistical methods, namely Inverse-variance weighted (IVW), MR-Egger, and weighted median, to assess the causal relationship between thyroid dysfunction and gastric cancer. The Cochran's Q test was used to assess the heterogeneity among SNPs in the IVW analysis results, and MR-PRESSO was employed to identify and remove IVs with heterogeneity from the analysis results. MR-Egger is a weighted linear regression model, and the magnitude of its intercept can be used to assess the horizontal pleiotropy among IVs. Finally, the data were visualized through the leave-one-out sensitivity test to evaluate the influence of individual SNPs on the overall causal effect. Funnel plots were used to assess the symmetry of the selected SNPs, forest plots were used to evaluate the confidence and heterogeneity of the incidental estimates, and scatter plots were used to assess the exposure-outcome relationship. All results were expressed as odds ratios (OR) and 95% confidence intervals (95% CI). P<0.05 represents statistical significance. According to IVW analysis, there was a causal relationship between hypothyroidism and gastric cancer, and hypothyroidism could reduce the risk of gastric cancer (OR=0.936 (95% CI:0.893-0.980), P=0.006).This means that having hypothyroidism is a protective factor against stomach cancer. This finding suggests that hypothyroidism may be associated with a reduced risk of gastric cancer.Meanwhile, there was no causal relationship between hyperthyroidism, FT4, and TSH and gastric cancer. In this study, we found a causal relationship between hypothyroidism and gastric cancer with the help of a two-sample Mendelian randomisation study, and hypothyroidism may be associated with a reduced risk of gastric cancer, however, the exact mechanism is still unclear. This finding provides a new idea for the study of the etiology and pathogenesis of gastric cancer, and our results need to be further confirmed by more basic experiments in the future.

Examining the impact of a 9-component bundle and the INICC multidimensional approach on catheter-associated urinary tract infection rates in 32 countries across Asia, Eastern Europe, Latin America, and the Middle East

Catheter-Associated Urinary Tract Infections (CAUTIs) frequently occur in the intensive care unit (ICU) and are correlated with a significant burden. We implemented a strategy involving a 9-element bundle, education, surveillance of CAUTI rates and clinical outcomes, monitoring compliance with bundle components, feedback of CAUTI rates and performance feedback. This was executed in 299 ICUs across 32 low- and middle-income countries. The dependent variable was CAUTI per 1,000 UC days, assessed at baseline and throughout the intervention, in the second month, third month, 4 to 15 months, 16 to 27 months, and 28 to 39 months. Comparisons were made using a 2-sample t test, and the exposure-outcome relationship was explored using a generalized linear mixed model with a Poisson distribution. Over the course of 978,364 patient days, 150,258 patients utilized 652,053 UC-days. The rates of CAUTI per 1,000 UC dayswere measured. The rates decreased from 14.89 during the baseline period to 5.51 in the second month (risk ratio [RR]=0.37; 95% confidence interval [CI]=0.34-0.39; P<.001), 3.79 in the third month (RR=0.25; 95% CI=0.23-0.28; P<.001), 2.98 in the 4 to 15 months (RR=0.21; 95% CI=0.18-0.22; P<.001), 1.86 in the 16 to 27 months (RR=0.12; 95% CI=0.11-0.14; P<.001), and 1.71 in the 28 to 39 months (RR=0.11; 95% CI=0.09-0.13; P<.001). Our intervention, without substantial costs or additional staffing, achieved an 89% reduction in CAUTI incidence in ICUs across 32 countries, demonstrating feasibility in ICUs of low- and middle-income countries.

A Mendelian randomization study of the effect of serum 25-hydroxyvitamin D levels on autoimmune thyroid disease.

The influence of vitamin D on autoimmune thyroid disease (AITD) remains a subject of ongoing debate. This study employs Mendelian randomization (MR) to investigate the causal correlations of serum 25-hydroxyvitamin D (25[OH]D) levels with autoimmune thyroiditis (AIT), autoimmune hyperthyroidism (AIH), and Graves disease (GD). Data on single nucleotide polymorphisms related to serum 25(OH)D levels, AIT, AIH, and GD were sourced from UK Biobank and FinnGen. Inverse variance weighted, MR-Egger, and weighted median were employed to test the exposure-outcome causal relationship. Assessments of horizontal pleiotropy, heterogeneity, and stability were performed using the MR-Egger intercept, Cochran's Q test, and leave-one-out sensitivity analysis, respectively. The results of MR analysis showed increased serum 25(OH)D levels was associated with a reduced risk of AIT (OR 0.499, 95% CI 0.289 to 0.860, p = 0.012) but not causal associated with AIH (OR 0.935, 95% CI 0.695 to 1.256, p = 0.654) and GD (OR 0.813, 95% CI 0.635 to 1.040, p = 0.100). Intercept analysis showed no horizontal pleiotropy (p > 0.05), and Cochran's Q test showed no heterogeneity (p > 0.05). Sensitivity analysis suggested that these results were robust. An increased serum 25(OH)D level is associated with AIT risk reduction but unrelated to AIH and GD. This finding suggests that vitamin D supplementation can be valuable for preventing and treating AIT.

Intraoperative hypotension associated with postoperative acute kidney injury in hypertension patients undergoing non-cardiac surgery: a retrospective cohort study.

Acute kidney injury (AKI) is a common surgical complication and is associated with intraoperative hypotension. However, the total duration and magnitude of intraoperative hypotension associated with AKI remains unknown. In this study, the causal relationship between the intraoperative arterial pressure and postoperative AKI was investigated among chronic hypertension patients undergoing non-cardiac surgery. A retrospective cohort study of 6552 hypertension patients undergoing non-cardiac surgery (2011 to 2019) was conducted. The primary outcome was AKI as diagnosed with the Kidney Disease-Improving Global Outcomes criteria and the primary exposure was intraoperative hypotension. Patients' baseline demographics, pre- and post-operative data were harvested and then analyzed with multivariable logistic regression to assess the exposure-outcome relationship. Among 6552 hypertension patients, 579 (8.84%) had postoperative AKI after non-cardiac surgery. The proportions of patients admitted to ICU (3.97 vs. 1.24%, p < 0.001) and experiencing all-cause death (2.76 vs. 0.80%, p < 0.001) were higher in the patients with postoperative AKI. Moreover, the patients with postoperative AKI had longer hospital stays (13.50 vs. 12.00days, p < 0.001). Intraoperative mean arterial pressure (MAP) < 60mmHg for >20min was an independent risk factor of postoperative AKI. Furthermore, MAP <60mmHg for >10min was also an independent risk factor of postoperative AKI in patients whose MAP was measured invasively in the subgroup analysis. Our work suggested that MAP < 60mmHg for >10min measured invasively or 20min measured non-invasively during non-cardiac surgery may be the threshold of postoperative AKI development in hypertension patients. This work may serve as a perioperative management guide for chronic hypertension patients. clinical trial number: ChiCTR2100050209 (8/22/2021). http://www.chictr.org.cn/showproj.aspx?proj=132277.

Decreasing central line-associated bloodstream infections rates in intensive care units in 30 low- and middle-income countries: An INICC approach

Central line (CL)-associated bloodstream infections (CLABSIs) occurring in the intensive care unit (ICU) are common and associated with a high burden. We implemented a multidimensional approach, incorporating an 11-element bundle, education, surveillance of CLABSI rates and clinical outcomes, monitoring compliance with bundle components, feedback of CLABSI rates and clinical outcomes, and performance feedback in 316 ICUs across 30 low- and middle-income countries. Our dependent variables were CLABSI per 1,000-CL-days and in-ICU all-cause mortality rates. These variables were measured at baseline and during the intervention, specifically during the second month, third month, 4 to 16 months, and 17 to 29 months. Comparisons were conducted using a two-sample t test. To explore the exposure-outcome relationship, we used a generalized linear mixed model with a Poisson distribution to model the number of CLABSIs. During 1,837,750 patient-days, 283,087 patients, used 1,218,882 CL-days. CLABSI per 1,000 CL-days rates decreased from 15.34 at the baseline period to 7.97 in the 2nd month (relative risk (RR)=0.52; 95% confidence interval [CI]=0.48-0.56; P<.001), 5.34 in the 3rd month (RR=0.35; 95% CI=0.32-0.38; P<.001), and 2.23 in the 17 to 29 months (RR=0.15; 95% CI=0.13-0.17; P<.001). In-ICU all-cause mortality rate decreased from 16.17% at baseline to 13.68% (RR=0.84; P=.0013) at 17 to 29 months. The implemented approach was effective, and a similar intervention could be applied in other ICUs of low- and middle-income countries to reduce CLABSI and in-ICU all-cause mortality rates.

Preoperative hs-CRP/HDL ratio is associated with increased risk for postoperative SIRS in elderly patients: a retrospective cohort study.

Systemic inflammatory response syndrome (SIRS) greatly affects postoperative lives of afflicted aged patients. This study aimed to determine whether preoperative high hs-CRP/HDL ratio (CHR) was associated with an increased risk of postoperative SIRS in the elderly population. This retrospective cohort study included data on patients aged ≥ 65years who underwent general anesthesia surgery at two clinical centers between January 2015 and September 2020. The primary exposure was preoperative CHR which was divided into two groups (≤ 12.82 and > 12.82) based on its normal range in our hospital, and the primary outcome was the incidence of postoperative SIRS. Targeted maximum likelihood estimation analyses were used to model the exposure-outcome relationship. The analysis included 5595 elderly patients, of whom 1410 (25.20%) developed SIRS within three postoperative days. Targeted maximum likelihood estimation analysis revealed that elderly patients with CHR > 12.82 vs. CHR ≤ 12.82 was associated with increased risk of postoperative SIRS (aOR = 1.40, 95% CI [1.33, 1.48], P < 0.001). Those results were consistent both in subgroup analyses and sensitivity analyses. Compared with patients with CHR ≤ 12.82, patients with CHR > 12.82 had a higher prevalence of postoperative SIRS (49.06% vs. 22.70%), postoperative in-hospital mortality (3.40% vs. 0.65%), a longer hospital stay after surgery [10 (IQR, 6-16) vs. 8 (IQR, 5-11) days] and higher direct medical cost [10070 (IQR, 6878-15577) vs. 7117 (IQR, 4079-10314) euros, all P < 0.001]. In elderly patients, preoperative CHR > 12.82 was significantly associated with a higher risk of postoperative SIRS.

Abstract 744: Treatment inequities among patients diagnosed with Hodgkin's lymphoma with and without HIV (2018-2019): an analysis of the US National Cancer Database

Abstract Background: People living with HIV (PLWH) with Hodgkin’s lymphoma(HL) have similar response and survival rates to people without HIV when standard of care therapies are given. However, PLWH with HL are less likely to receive cancer treatment for unclear reasons. Our objective was to examine the role of social determinants of health(SDOH) with non-receipt of cancer treatment among PLWH with HL. Methods: Using the U.S. National Cancer Database, we identified 11,517 persons aged 18-90 years, diagnosed with HL (ICD-O-3 codes; C770-779) between 2018-2019. Most patients were diagnosed with classical HL(9650; n=4534, 39%) or nodular sclerosis classical Hodgkin’s lymphoma (9663; n=4725, 41%). As per NCCN guidelines, receipt of first-line treatment was defined as receiving any chemotherapy, radiotherapy or a combination of both. We compared treatment receipt by HIV status and associations of SDOH (e.g., zip-code median household income) with non-receipt of treatment among those with and without HIV. Using multivariable logistic regression, we evaluated the association of treatment receipt with HIV status after adjustment for age, sex, and stage at diagnosis. We stratified this model by zip-code level income, education and race/ethnicity to evaluate the role of SDOH in this exposure-outcome relationship. For each model, we calculated cluster-robust standard errors to account for non-independence within clusters at the facility level to adjust for correlated patient characteristics within hospitals. Results: Our sample (n=11,517) included 60% adults aged 50 years or below, 55% men, 71% NH-White and 14% NH-Black. About one-third were treated at an academic/research program and 57% were privately insured. PLWH included 6% of the population. Overall, 12% did not receive first-line recommended treatment and were less likely to receive treatment compared to those without HIV (13% vs. 9%, p=0.012). We identified the following characteristics amongst PLWH to be more likely not to receive treatment compared to those without HIV, persons who were: aged 40 years or younger(13% vs. 6%, p=0.001), male(15% vs. 9%, p&amp;lt;0.001), NH-White(13% vs. 8%) and NH-Black(14% vs. 11%) persons(p&amp;lt;0.001), persons residing in areas with lower levels of educational attainment (20% vs. 10%, p&amp;lt;0.001) and low income areas (14% vs 9%, p&amp;lt;0.001), and were Medicaid insured (17% vs. 8%, p&amp;lt;.001). Stratifying by area-level education, this association persisted among those residing in areas with lower levels of education and income. Conclusion: PLWH with HL were less likely to receive cancer treatment and continue to experience inequities in cancer treatment receipt, with SDOH playing a potential role in inequitable cancer care delivery. Citation Format: Jessica Yasmine Islam, Jennifer McGee, Yi Guo, Marlene Camacho-Rivera, Lubomir Sokol, Susan T. Vadaparampil, Gita Suneja. Treatment inequities among patients diagnosed with Hodgkin's lymphoma with and without HIV (2018-2019): an analysis of the US National Cancer Database [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 744.

Association between unmet social needs and healthy lifestyle parenting behaviors.

To examine cross-sectional associations of food and housing security risks and healthy lifestyle parenting behaviors related to nutrition and physical activity among families with children with overweight/obesity. We surveyed 407 parents of children ages 6-12 years with overweight/obesity. Exposures were measures of food and housing insecurity risk. Outcomes were healthy lifestyle parenting behaviors related to nutrition and physical activity. Logistic regression models for each exposure-outcome relationship were adjusted for parental educational attainment, parental cohabitation status, household size, and household income. In multivariable-adjusted models, food insecurity was associated with significantly lower odds of parent modeling exercise {aOR 0.60 [95% confidence interval (CI): 0.37, 0.96]} and parent modeling eating healthy foods [aOR 0.42 (95% CI: 0.24, 0.73)]. Housing insecurity was associated with significantly lower odds of parent modeling exercise [aOR 0.57 (95% CI: 0.35, 0.95)]. Food insecurity and housing insecurity may be barriers to parents adopting and modeling healthy lifestyle parenting behaviors related to physical activity and nutrition.

Simultaneous adjustment of uncontrolled confounding, selection bias and misclassification in multiple-bias modelling.

Adjusting for multiple biases usually involves adjusting for one bias at a time, with careful attention to the order in which these biases are adjusted. A novel, alternative approach to multiple-bias adjustment involves the simultaneous adjustment of all biases via imputation and/or regression weighting. The imputed value or weight corresponds to the probability of the missing data and serves to 'reconstruct' the unbiased data that would be observed based on the provided assumptions of the degree of bias. We motivate and describe the steps necessary to implement this method. We also demonstrate the validity of this method through a simulation study with an exposure-outcome relationship that is biased by uncontrolled confounding, exposure misclassification, and selection bias. The study revealed that a non-biased effect estimate can be obtained when correct bias parameters are applied. It also found that incorrect specification of every bias parameter by +/-25% still produced an effect estimate with less bias than the observed, biased effect. Simultaneous multi-bias analysis is a useful way of investigating and understanding how multiple sources of bias may affect naive effect estimates. This new method can be used to enhance the validity and transparency of real-world evidence obtained from observational, longitudinal studies.

Neighbourhood-related socioeconomic perinatal health inequalities: An illustration of the mediational g-formula and considerations for the big data context.

Advances in computing power have enabled the collection, linkage and processing of big data. Big data in conjunction with robust causal inference methods can be used to answer research questions regarding the mechanisms underlying an exposure-outcome relationship. The g-formula is a flexible approach to perform causal mediation analysis that is suited for the big data context. Although this approach has many advantages, it is underused in perinatal epidemiology and didactic explanation for its implementation is still limited. The aim of this was to provide a didactic application of the mediational g-formula by means of perinatal health inequalities research. The analytical procedure of the mediational g-formula is illustrated by investigating whether the relationship between neighbourhood socioeconomic status (SES) and small for gestational age (SGA) is mediated by neighbourhood social environment. Data on singleton births that occurred in the Netherlands between 2010 and 2017 (n=1,217,626) were obtained from the Netherlands Perinatal Registry and linked to sociodemographic national registry data and neighbourhood-level data. The g-formula settings corresponded to a hypothetical improvement in neighbourhood SES from disadvantaged to non-disadvantaged. At the population level, a hypothetical improvement in neighbourhood SES resulted in a 6.3% (95% confidence interval [CI] 5.2, 7.5) relative reduction in the proportion of SGA, that is the total effect. The total effect was decomposed into the natural direct effect (5.6%, 95% CI 5.1, 6.1) and the natural indirect effect (0.7%, 95% CI 0.6, 0.9). In terms of the magnitude of mediation, it was observed the natural indirect effect accounted for 11.4% (95% CI 9.2, 13.6) of the total effect of neighbourhood SES on SGA. The mediational g-formula is a flexible approach to perform causal mediation analysis that is suited for big data contexts in perinatal health research. Its application can contribute to providing valuable insights for the development of policy and public health interventions.

Medicaid payer status is associated with increased 90-day morbidity and resource utilization following primary shoulder arthroplasty: a propensity score-matched analysis.

Medicaid payer status has been shown to affect risk-adjusted patient outcomes and health care utilization across multiple medical specialties and orthopedic procedures. However, there is a paucity of data regarding the impact of Medicaid payer status on 90-day morbidity and resource utilization following primary shoulder arthroplasty (reverse total shoulder arthroplasty [rTSA], anatomic total shoulder arthroplasty [aTSA], and hemiarthroplasty [HA]). The purpose of this study was to examine 90-day readmissionand reoperationrates, hospital length of stay (LOS), and direct cost following primary shoulder arthroplasty in the Medicaid population. The National Readmission Database was queried for all patients undergoing primary aTSA, rTSA, and HA from 2011 to 2016. Medicaid or non-Medicaid payer status was determined. Patient demographic characteristics and comorbidities, along with 90-day readmission, 90-day reoperation, LOS, and inflation-adjusted cost, were queried. Propensity score matching was used to control for baseline differences in cohorts that could be acting as confounders in the exposure-outcome relationship. This was achieved with 1-to-1 propensity score matching between Medicaid and non-Medicaid patients. Odds ratios (ORs) and 95% confidence intervals (CIs) for 90-day readmission and reoperation rates were calculated, and a comparison of LOS and cost was performed between the propensity score-matched cohorts. A total of 4667 Medicaid and 161,147 non-Medicaid patients were identified from the 2011-2016 National Readmission Databases. Propensity score analysis was performed, and 4637 Medicaid patients were matched to 4637 non-Medicaid patients; each group comprised 1504 rTSAs (32.4%), 1934 aTSAs (41.7%), and 1199 HAs (25.9%). Patients with Medicaid payer status yielded significant increases in the 90-day all-cause readmission rate of 11.6% vs. 9.3% (P<.001; OR, 1.28 [95% CI, 1.12-1.46]), 90-day shoulder-related readmission rate of 3.3% vs. 2.3% (P=.004; OR, 1.44 [95% CI, 1.12-1.85]), and 90-day reoperation rate of 2.0% vs. 1.3% (P=.008; OR, 1.54 [95% CI, 1.12-1.94]). Furthermore, there was an increased risk of an extended LOS (ie, LOS > 2 days) (28.4% vs. 25.7%; P=.004; OR, 1.14 [95% CI, 1.04-1.25]) along with increased direct cost (median, $17,612 vs. $16,775; P<.001). This study demonstrates that Medicaid payer status is independently associated with increased 90-day readmission and reoperation rates, LOS, and direct cost following primary shoulder arthroplasty. Providers may have a disincentive to treat patient populations who require increased resource utilization following surgery. Risk adjustment models accounting for Medicaid payer status will be necessary to ensure good access to care for this patient population by avoiding penalties for physicians and hospital systems.

A machine learning based approach towards high-dimensional mediation analysis

Mediation analysis is used to investigate the role of intermediate variables (mediators) that lie in the path between an exposure and an outcome variable. While significant research has focused on developing methods for assessing the influence of mediators on the exposure-outcome relationship, current approaches do not easily extend to settings where the mediator is high-dimensional. These situations are becoming increasingly common with the rapid increase of new applications measuring massive numbers of variables, including brain imaging, genomics, and metabolomics. In this work, we introduce a novel machine learning based method for identifying high dimensional mediators. The proposed algorithm iterates between using a machine learning model to map the high-dimensional mediators onto a lower-dimensional space, and using the predicted values as input in a standard three-variable mediation model. Hence, the machine learning model is trained to maximize the likelihood of the mediation model. Importantly, the proposed algorithm is agnostic to the machine learning model that is used, providing significant flexibility in the types of situations where it can be used. We illustrate the proposed methodology using data from two functional Magnetic Resonance Imaging (fMRI) studies. First, using data from a task-based fMRI study of thermal pain, we combine the proposed algorithm with a deep learning model to detect distributed, network-level brain patterns mediating the relationship between stimulus intensity (temperature) and reported pain at the single trial level. Second, using resting-state fMRI data from the Human Connectome Project, we combine the proposed algorithm with a connectome-based predictive modeling approach to determine brain functional connectivity measures that mediate the relationship between fluid intelligence and working memory accuracy. In both cases, our multivariate mediation model links exposure variables (thermal pain or fluid intelligence), high dimensional brain measures (single-trial brain activation maps or resting-state brain connectivity) and behavioral outcomes (pain report or working memory accuracy) into a single unified model. Using the proposed approach, we are able to identify brain-based measures that simultaneously encode the exposure variable and correlate with the behavioral outcome.

Reporting and methodological quality of studies that use Mendelian randomisation in UK Biobank: a meta-epidemiological study

ObjectivesTo identify whether Mendelian randomisation (MR) studies are appropriately conducted and reported in enough detail for other researchers to accurately replicate and interpret them.DesignCross-sectional meta-epidemiological study.Data sourcesWeb of Science, EMBASE,...

Software Application Profile: SUMnlmr, an R package that facilitates flexible and reproducible non-linear Mendelian randomization analyses

Abstract Motivation Mendelian randomization methods that estimate non-linear exposure-outcome relationships typically require individual-level data. This package implements non-linear Mendelian randomization methods using stratified summarized data, facilitating analyses where individual-level data cannot easily be shared, and additionally increasing reproducibility as summarized data can be reported. Dependence on summarized data means the methods are independent of the form of the individual-level data, increasing flexibility to different outcome types (such as continuous, binary or time-to-event outcomes). Implementation SUMnlmr is available as an R package (version 3.1.0 or higher). General features The package implements the previously proposed fractional polynomial and piecewise linear methods on stratified summarized data that can either be estimated from individual-level data using the package or supplied by a collaborator. It constructs plots to visualize the estimated exposure-outcome relationship, and provides statistics to assess preference for a non-linear model over a linear model. Availability The package is freely available from GitHub [https://github.com/amymariemason/SUMnlmr].

E-values for effect heterogeneity and approximations for causal interaction.

BackgroundEstimates of effect heterogeneity (i.e. the extent to which the causal effect of one exposure varies across strata of a second exposure) can be biased if the exposure–outcome relationship is subject to uncontrolled confounding whose severity differs across strata of the second exposure.MethodsWe propose methods, analogous to the E-value for total effects, that help to assess the sensitivity of effect heterogeneity estimates to possible uncontrolled confounding. These E-value analogues characterize the severity of uncontrolled confounding strengths that would be required, hypothetically, to ‘explain away’ an estimate of multiplicative or additive effect heterogeneity in the sense that appropriately controlling for those confounder(s) would have shifted the effect heterogeneity estimate to the null, or alternatively would have shifted its confidence interval to include the null. One can also consider shifting the estimate or confidence interval to an arbitrary non-null value. All of these E-values can be obtained using the R package EValue.ResultsWe illustrate applying the proposed E-value analogues to studies on: (i) effect heterogeneity by sex of the effect of educational attainment on dementia incidence and (ii) effect heterogeneity by age on the effect of obesity on all-cause mortality.ConclusionReporting these proposed E-values could help characterize the robustness of effect heterogeneity estimates to potential uncontrolled confounding.