scholarly journals Estimation of Risk Ratios in Cohort Studies with a Common Outcome: A Simple and Efficient Two-stage Approach

2013 ◽  
Vol 9 (2) ◽  
Author(s):  
Eric Tchetgen Tchetgen
Keyword(s):  
Epidemiologic Studies ◽  
Baseline Risk ◽  
Binary Outcome ◽  
Primary Interest ◽  
Two Stage ◽  
Effect Measure ◽  
Multiplicative Effect ◽  
Doubly Robust ◽  
Risk Ratios ◽  
Predicted Risk

AbstractThe risk ratio effect measure is often the main parameter of interest in epidemiologic studies with a binary outcome. In this paper, the author presents a simple and efficient two-stage approach to estimate the risk ratios directly, which does not directly rely on consistency for an estimate of the baseline risk. This latter property is a key advantage of the approach over existing methods, because, unlike these other methods, the proposed approach obviates the need to restrict the predicted risk probabilities to fall below one, in order to recover efficient inferences about risk ratios. An additional appeal of the approach is that it is easy to implement. Finally, when the primary interest is in the effect of a specific binary exposure, a simple doubly robust closed-form estimator is derived, for the multiplicative effect of the exposure. Specifically, we show how one can adjust for confounding by incorporating a working regression model for the propensity score so that the correct inferences about the multiplicative effect of the exposure are recovered if either this model is correct or a working model for the association between confounders and outcome risk is correct, but both do not necessarily hold.

scholarly journals Excess mortality during the COVID-19 outbreak in Italy: a two-stage interrupted time series analysis

2020 ◽  
Author(s):  
Matteo Scortichini ◽  
Rochelle Schneider dos Santos ◽  
Francesca De' Donato ◽  
Manuela De Sario ◽  
Paola Michelozzi ◽  
...  
Keyword(s):  
Time Series ◽  
Excess Mortality ◽  
Temporal Distribution ◽  
Interrupted Time Series ◽  
Baseline Risk ◽  
Mortality Data ◽  
Interrupted Time Series Analysis ◽  
Two Stage ◽  
Smooth Functions ◽  
The Impact

Background: Italy was the first country outside China to experience the impact of the COVID-19 pandemic, which resulted in a significant health burden. This study presents an analysis of the excess mortality across the 107 Italian provinces, stratified by sex, age group, and period of the outbreak. Methods: The analysis was performed using a two-stage interrupted time series design using daily mortality data for the period January 2015 - May 2020. In the first stage, we performed province-level quasi-Poisson regression models, with smooth functions to define a baseline risk while accounting for trends and weather conditions and to flexibly estimate the variation in excess risk during the outbreak. Estimates were pooled in the second stage using a mixed-effects multivariate meta-analysis. Results: In the period 15 February - 15 May 2020, we estimated an excess of 47,490 (95% empirical confidence intervals: 43,984 to 50,362) deaths in Italy, corresponding to an increase of 29.5% (95%eCI: 26.8 to 31.9%) from the expected mortality. The analysis indicates a strong geographical pattern, with the majority of excess deaths occurring in northern regions, where few provinces experienced up to 800% increase during the peak in late March. There were differences by sex, age, and area both in the overall impact and in its temporal distribution. Conclusions: This study offers a detailed picture of excess mortality during the first months of the COVID-19 pandemic in Italy. The strong geographical and temporal patterns can be related to implementation of lockdown policies and multiple direct and indirect pathways in mortality risk.

scholarly journals 1455Leveraging multi-omic negative controls for effect estimation in molecular epidemiologic studies: A simulation study

2021 ◽  
Vol 50 (Supplement_1) ◽  
Author(s):  
Jonathan Huang
Keyword(s):  
Machine Learning ◽  
Molecular Epidemiology ◽  
Maternal Blood ◽  
Epidemiologic Studies ◽  
Negative Control ◽  
Biological Knowledge ◽  
Methylation Array ◽  
Unmeasured Confounding ◽  
Dna Methylation Array ◽  
Doubly Robust

Abstract Background Exploratory null-hypothesis significance testing (e.g. GWAS, EWAS) form the backbone of molecular epidemiology, however methods to identify true causal signals are underdeveloped. Via plasmode simulation, I evaluate two approaches to quantitatively control for shared unmeasured confounding and recover unbiased effects using complementary epigenomes and biologically-informed structural assumptions. Methods I adapt proposed negative control-based estimators, the control outcome calibration approach (COCA) and proximal g-computation (PG) to case studies in perinatal molecular epidemiology. COCA may be employed when maternal epigenome has no direct effects on phenotype and proxy shared unmeasured confounders and PG further with suitable genetic instruments (e.g. mQTLs). Baseline covariates were extracted from 777 mother-child pairs in a birth cohort with maternal blood and fetal cord DNA methylation array data. Treatment and outcome values were simulated in 2000 bootstraps. Bootstrapped, ordinary (COCA) and 2-stage (PG) least squares were fitted to estimate treatment effects and standard errors under various common settings of missing confounders (e.g. paternal data). Doubly-robust, machine learning estimators were explored. Results COCA and PG performed well in simplistic data generating processes. However, in real-world cohort simulations, COCA performed acceptably only in settings with strong proxy confounders, but otherwise poorly (median bias 610%; coverage 29%). PG performed slightly better. Alternatively, simple covariate adjustment for maternal methylation outperformed (median bias 22%; 71% coverage) COCA, PG, and machine learning estimators. Discussion Molecular epidemiology provides key opportunity to leverage biological knowledge against unmeasured confounding. Negative control calibration or adjustments may help under limited scenarios where assumptions are fulfilled, but should be tested with suitable simulations. Key messages Quantitative approaches for unmeasured confounding in molecular epidemiology are a critical gap. Negative control calibration or adjustment may help under limiting scenarios. Proposed estimators should be tested in simulation settings that closely mimic target data.

scholarly journals Individual Participant Data Meta-Analysis for a Binary Outcome: One-Stage or Two-Stage?

PLoS ONE ◽  
2013 ◽  
Vol 8 (4) ◽  
pp. e60650 ◽  
Author(s):  
Thomas P. A. Debray ◽  
Karel G. M. Moons ◽  
Ghada Mohammed Abdallah Abo-Zaid ◽  
Hendrik Koffijberg ◽  
Richard David Riley
Keyword(s):  
Meta Analysis ◽  
Individual Participant Data ◽  
Binary Outcome ◽  
Two Stage ◽  
One Stage ◽  
Individual Participant

Doubly robust testing and estimation of model-adjusted effect-measure modification with complex survey data

2012 ◽  
Vol 32 (4) ◽  
pp. 673-684 ◽  
Author(s):  
Hao W. Zheng ◽  
Babette A. Brumback ◽  
Xiaomin Lu ◽  
Erin D. Bouldin ◽  
Michael B. Cannell ◽  
...  
Keyword(s):  
Survey Data ◽  
Complex Survey ◽  
Complex Survey Data ◽  
Robust Testing ◽  
Effect Measure ◽  
Effect Measure Modification ◽  
Doubly Robust

scholarly journals Adjusting for time of infection or positive test when estimating the risk of a post-infection outcome in an epidemic

2021 ◽  
Author(s):  
Shaun R Seaman ◽  
Tommy Nyberg ◽  
Christopher E Overton ◽  
David Pascall ◽  
Anne M Presanis ◽  
...  
Keyword(s):  
Post Infection ◽  
Positive Test ◽  
Test Time ◽  
Binary Outcome ◽  
Calendar Time ◽  
Infection Time ◽  
Time Of Infection ◽  
Infectious Pathogen ◽  
Mean Time ◽  
Risk Ratios

When comparing the risk of a post-infection binary outcome, e.g. hospitalisation, for two variants of an infectious pathogen, it is important to adjust for calendar time of infection to avoid the confounding that would occur if the relative incidence of the two variants and the variant-specific risks of the outcome both change over time. Infection time is typically unknown and time of positive test used instead. Likewise, time of positive test may be used instead of infection time when assessing how the risk of the binary outcome changes over calendar time. Here we show that if mean time from infection to positive test is correlated with the outcome, the risk conditional on positive test time depends on whether incidence of infection is increasing or decreasing over calendar time. This complicates interpretation of risk ratios adjusted for positive test time. We also propose a simple sensitivity analysis that indicates how these risk ratios may differ from the risk ratios adjusted for infection time.

scholarly journals 2034 Effect of balanced crystalloids on renal outcomes among critically ill adults does not differ from 0.9% saline across baseline risk of renal outcomes

2018 ◽  
Vol 2 (S1) ◽  
pp. 42-42
Author(s):  
Andrew C. McKown ◽  
Todd W. Rice ◽  
Matthew W. Semler
Keyword(s):  
Logistic Regression Model ◽  
Pragmatic Trial ◽  
Baseline Risk ◽  
Renal Outcomes ◽  
Group Assignment ◽  
Heterogeneity Of Treatment Effect ◽  
Critically Ill Adults ◽  
Ill Adults ◽  
Predicted Risk ◽  
Balanced Crystalloids

OBJECTIVES/SPECIFIC AIMS: Traditional clinical trials typically enroll a homogenous population to test the efficacy of an intervention. Pragmatic trials deliberately enroll a more diverse population to enhance generalizability, but doing so may increase heterogeneity of treatment effect among subpopulations. For example, the effect of a treatment on an outcome may vary based on patients’ sex, comorbidities, or baseline risk of experiencing the outcome. We hypothesized that heterogeneity of treatment effect by baseline risk for the outcome could be demonstrated in a large pragmatic clinical trial. METHODS/STUDY POPULATION: We performed a prespecified secondary analysis of a recent pragmatic trial comparing balanced crystalloids Versus 0.9% saline among critically ill adults. The primary endpoint of the trial was major adverse kidney events within 30 days of ICU admission, censored at hospital discharge (MAKE30). MAKE30 is a composite outcome of all-cause mortality, new renal replacement therapy, or persistent renal dysfunction. Using a previously published model with high predictive accuracy for MAKE30 (area under the curve=0.903), we calculated the baseline risk of MAKE30 for all trial participants. We then developed a logistic regression model for MAKE30 with independent covariates of fluid group assignment, baseline risk of MAKE30 as a nonlinear continuous variable, and the interaction between group assignment and MAKE30 baseline risk. RESULTS/ANTICIPATED RESULTS: Among 15,802 patients from 5 intensive care units enrolled in the original trial, 126 had missing variables for predicted risk of MAKE30. Mean predicted risk of MAKE30 among all patients was 15.4%; median was 4.4% (interquartile range 2.2%–17.1%). Predicted risk of MAKE30 did not significantly differ between groups (p=0.61 by Mann-Whitney U-test). The incidence of MAKE30 in the trial was 14.9%, and the prediction model was well-calibrated overall (AUC=0.891). In a logistic regression model examining the interaction between group assignment and predicted risk of MAKE30, group assignment significantly affected MAKE30 (odds ratio saline:balanced 1.13, 95% CI: 1.02–1.27, p=0.02), but we observed no interaction between the effect of group assignment on MAKE30 and patients’ predicted risk of MAKE30 at baseline (p=0.66 for interaction term). DISCUSSION/SIGNIFICANCE OF IMPACT: In a large pragmatic trial demonstrating a significant difference in the primary outcome of MAKE30 between balanced crystalloids and saline, a previously published model accurately predicted MAKE30 using baseline factors. However, contrary to our hypothesis, the baseline risk of MAKE30 did not modify the effect of fluid group on the observed incidence of MAKE30. Our analysis could not account for unmeasured confounders and may be underpowered to detect a significant interaction. Our findings suggest that the impact of balanced crystalloids versus normal saline on renal outcomes in critically patients is consistent across all levels of risk.

scholarly journals Coffee Intake and Obesity: A Meta-Analysis

Nutrients ◽  
2019 ◽  
Vol 11 (6) ◽  
pp. 1274 ◽  
Author(s):  
Ariel Lee ◽  
Woobin Lim ◽  
Seoyeon Kim ◽  
Hayeong Khil ◽  
Eugene Cheon ◽  
...  
Keyword(s):  
Weighted Mean Difference ◽  
Meta Analysis ◽  
Epidemiologic Studies ◽  
Inverse Association ◽  
Central Adiposity ◽  
Binary Outcome ◽  
Coffee Intake ◽  
Continuous Outcome ◽  
Summary Relative Risk ◽  
The Relationship

Many studies have explored the relationship between coffee—one of the most commonly consumed beverages today—and obesity. Despite inconsistent results, the relationship has not been systematically summarized. Thus, we conducted a meta-analysis by compiling data from 12 epidemiologic studies identified from PubMed and Embase through February 2019. The included studies assessed obesity by body mass index (BMI, a measure of overall adiposity) or waist circumference (WC, a measure of central adiposity); analyzed the measure as a continuous outcome or binary outcome. Using random effects model, weighted mean difference (WMD) and 95% confidence interval (CI) were obtained for continuous outcomes; summary relative risk (RR) and 95% CI for the highest vs. lowest categories of coffee intake were estimated for binary outcome. For BMI, WMD was −0.08 (95% CI −0.14, −0.02); RR was 1.49 (95% CI 0.97, 2.29). For WC, WMD was −0.27 (95% CI −0.51, −0.02) and RR was 1.07 (95% CI 0.84, 1.36). In subgroup analysis by sex, evidence for an inverse association was more evident in men, specifically for continuous outcome, with WMD −0.05 (95% CI −0.09, −0.02) for BMI and −0.21 (95% CI −0.35, −0.08) for WC. Our meta-analysis suggests that higher coffee intake might be modestly associated with reduced adiposity, particularly in men.

scholarly journals Graphing and reporting heterogeneous treatment effects through reference classes

Trials ◽  
2020 ◽  
Vol 21 (1) ◽  
Author(s):  
James A. Watson ◽  
Chris C. Holmes
Keyword(s):  
Predictive Model ◽  
Treatment Effect ◽  
Multiple Testing ◽  
Treatment Effects ◽  
Population Heterogeneity ◽  
Baseline Risk ◽  
Two Stage ◽  
Heterogeneous Treatment Effects ◽  
Reference Classes ◽  
Baseline Covariates

Abstract Background Exploration and modelling of heterogeneous treatment effects as a function of baseline covariates is an important aspect of precision medicine in randomised controlled trials (RCTs). Randomisation generally guarantees the internal validity of an RCT, but heterogeneity in treatment effect can reduce external validity. Estimation of heterogeneous treatment effects is usually done via a predictive model for individual outcomes, where one searches for interactions between treatment allocation and important patient baseline covariates. However, such models are prone to overfitting and multiple testing and typically demand a transformation of the outcome measurement, for example, from the absolute risk in the original RCT to log-odds of risk in the predictive model. Methods We show how reference classes derived from baseline covariates can be used to explore heterogeneous treatment effects via a two-stage approach. We first estimate a risk score which captures on a single dimension some of the heterogeneity in outcomes of the trial population. Heterogeneity in the treatment effect can then be explored via reweighting schemes along this axis of variation. This two-stage approach bypasses the search for interactions with multiple covariates, thus protecting against multiple testing. It also allows for exploration of heterogeneous treatment effects on the original outcome scale of the RCT. This approach would typically be applied to multivariable models of baseline risk to assess the stability of average treatment effects with respect to the distribution of risk in the population studied. Case study We illustrate this approach using the single largest randomised treatment trial in severe falciparum malaria and demonstrate how the estimated treatment effect in terms of absolute mortality risk reduction increases considerably in higher risk strata. Conclusions ‘Local’ and ‘tilting’ reweighting schemes based on ranking patients by baseline risk can be used as a general approach for exploring, graphing and reporting heterogeneity of treatment effect in RCTs. Trial registration ISRCTN clinical trials registry: ISRCTN50258054. Prospectively registered on 22 July 2005.

scholarly journals A Note on the Control Function Approach with an Instrumental Variable and a Binary Outcome

2014 ◽  
Vol 3 (1) ◽  
Author(s):  
Eric Tchetgen Tchetgen
Keyword(s):  
Logistic Regression ◽  
Instrumental Variable ◽  
Logistic Model ◽  
Control Function ◽  
Function Approach ◽  
Continuous Exposure ◽  
Binary Outcome ◽  
Two Stage ◽  
Second Stage ◽  
Control Function Approach

AbstractUnobserved confounding is a well-known threat to causal inference in non-experimental studies. The instrumental variable design can under certain conditions be used to recover an unbiased estimator of a treatment effect even if unobserved confounding cannot be ruled out with certainty. For continuous outcomes, two stage least squares is the most common instrumental variable estimator used in epidemiologic applications. For a rare binary outcome, an analogous linear-logistic two stage procedure can be used. Alternatively, a control function approach is sometimes used which entails entering the residual from the first stage linear model for exposure as a covariate in a second stage logistic regression of the outcome on the treatment. Both strategies for binary response have previously formally been justified only for continuous exposure, which has impeded widespread use of the approach outside of this setting. In this note, we consider the important setting of binary exposure in the context of a binary outcome. We provide an alternative motivation for the control function approach which is appropriate for binary exposure, thus establishing simple conditions under which the approach may be used for instrumental variable estimation when the outcome is rare. In the proposed approach, the first stage regression involves a logistic model of the exposure conditional on the instrumental variable, and the second stage regression is a logistic regression of the outcome on the exposure adjusting for the first stage residual. In the event of a non-rare outcome, we recommend replacing the second stage logistic model with a risk ratio regression.

Sign in / Sign up

Export Citation Format

Share Document