Using a monotone single-index model to stabilize the propensity score in missing data problems and causal inference

The augmented inverse weighting method is one of the most popular methods for estimating the mean of the response in causal inference and missing data problems. An important component of this method is the propensity score. Popular parametric models for the propensity score include the logistic, probit, and complementary log-log models. A common feature of these models is that the propensity score is a monotonic function of a linear combination of the explanatory variables. To avoid the need to choose a model, we model the propensity score via a semiparametric single-index model, in which the score is an unknown monotonic nondecreasing function of the given single index. Under this new model, the augmented inverse weighting estimator (AIWE) of the mean of the response is asymptotically linear, semiparametrically efficient, and more robust than existing estimators. Moreover, we have made a surprising observation. The inverse probability weighting and AIWEs based on a correctly specified parametric model may have worse performance than their counterparts based on a nonparametric model. A heuristic explanation of this phenomenon is provided. A real-data example is used to illustrate the proposed methods.     

Estimating returns to hospital volume: Evidence from advanced cancer surgery

High-volume hospitals typically perform better than low-volume hospitals. In this paper, we study whether such patterns reflect a causal effect of case volume on patient outcomes. To this end, we exploit closures and openings of entire cancer clinics in Swedish hospitals which provides sharp and arguably exogenous variation in case volumes. Using detailed register data on more than 100,000 treatment episodes of advanced cancer surgery, our results suggest substantial positive effects of operation volume on survival. Complementary analyses point to learning-by-doing as an important explanation.     

Microhaplotypes in forensic genetics

Microhaplotype loci (microhaps, MHs) are a novel type of molecular marker of less than 300 nucleotides, defined by two or more closely linked SNPs associated in multiple allelic combinations. The value of these markers is enhanced by massively parallel sequencing (MPS), which allows the sequencing of both parental haplotypes at each of the many multiplexed loci. This review describes the features of these multi-SNP markers and documents their value in forensic genetics, focusing on individualization, biogeographic ancestry inference, and mixture deconvolution. Foreseeable applications also include missing person identification, relationship testing, and medical diagnostic applications. The technique is not restricted to humans.     

Association of heart rate and diabetes among 0.5 million adults in the China Kadoorie biobank: Results from observational and Mendelian randomization analyses

Background and aimsObservational studies have associated resting heart rate with incident diabetes. Whether the associations are causal remains unclear. We aimed to examine the shape and strength of the associations and assessed the causal relevance of such associations in Chinese adults.Methods and resultsThe China Kadoorie Biobank enrolled 512,891 adults in China. Cox proportional hazard regression models was conducted to estimate hazard ratios (HRs) for the associations of resting heart rate with type 2 diabetes and total diabetes. Among 92,724 participants, 36 single-nucleotide polymorphisms (SNPs) related to resting heart rate were used to construct genetic risk score. We used Mendelian randomization analyses to make the causal inferences. During a median follow-up of 9 years, 7872 incident type 2 diabetes and 13,349 incident total diabetes were documented. After regression dilution bias adjustment, each 10 bpm higher heart rate was associated with about a 26% higher risk of type 2 diabetes (HR, 1.26 [95% CI, 1.23, 1.29]) and 23% higher risk of total diabetes (HR, 1.23 [95% CI, 1.20, 1.26]). Instrumental variable analyses showed participants at top quintile compared with those at bottom quintile had 30% higher risk for type 2 diabetes (HR, 1.30 [95% CI, 1.17, 1.43]), and 10% higher risk for total diabetes (HR, 1.10 [95% CI, 1.02, 1.20]).ConclusionsThis study provides evidence that resting heart rate is an important risk factor for diabetes risk. The results suggest that novel treatment approaches targeting reduction of high heart rate for incidence of diabetes may be worth further investigation.     

Activity spaces in studies of the environment and physical activity: A review and synthesis of implications for causality

Activity spaces are increasingly used to understand how people interact with their environment and engage in activity but their use may raise challenges regarding causal inference. We conducted a systematic review of findings and the methodological, analytical and conceptual issues relevant to causal inference. Studies were included if they comprised a spatial summary of locations visited, assessed any part of the causal pathway between the environment, physical activity and health, and used quantitative or qualitative methods. We searched seven electronic databases in January 2018 and screened 11910 articles for eligibility. Forty-seven studies were included for review. Studies answered research questions about features of or environmental features within activity spaces using a range of spatial and temporal summary techniques. The conceptual challenge of using activity spaces to strengthen causal inference was rarely considered, although some studies discussed circularity, temporality, and plausibility. Future studies should use longitudinal and experimental designs and consider the potential and actual use of spaces for physical activity, and their relationship with total levels of activity.     

Bayesian bent‐cable growth mixture tobit models for longitudinal data with skewness and detection limit: application to AIDS studies

This paper presents a new Bayesian methodology for identifying a transition period for the development of drug resistance to antiretroviral drug or therapy in HIV/AIDS studies or other related fields. Estimation of such a transition period requires an availability of longitudinal data where growth trajectories of a response variable tend to exhibit a gradual change from a declining trend to an increasing trend rather than an abrupt change. We assess this clinically important feature of the longitudinal HIV/AIDS data using the bent‐cable framework within a growth mixture Tobit model. To account for heterogeneity of drug resistance among subjects, the parameters of the bent‐cable growth mixture Tobit model are also allowed to differ by subgroups (subpopulations) of patients classified into latent classes on the basis of trajectories of observed viral load data with skewness and left‐censoring. The proposed methods are illustrated using real data from an AIDS clinical study. Copyright © 2016 John Wiley & Sons, Ltd.     

横断面研究能否进行因果推断

基于变量调查(或测量)的共时性、统计学关联及幸存者偏倚等原因,横断面研究被认为不能进行因果推断,这是当前的"共识"。本文基于因果思维,借助因果图,首先明确定义真实截面和测量截面,并识别截面概念仅存在于理论的特性。实际横断面研究中,测量变量的共时性并不存在,而是无一例外地表现为非共时性时序,其实质上相当于测量变量间互为独立性假设,或不存在有差别错分偏倚。类似于累积病例对照研究和历史性队列研究,横断面研究均为暴露和结局已存在或发生后进行的测量,这种测量相当于对变量值的历史重建或"考古"。这类研究进行因果推断的共性前提条件之一是,测量变量与其历史变量间必须存在着因果律。测量变量均为真实变量的替代者,测量变量间的时序在因果推断上并不重要。应加强对横断面研究分析性角色的认识。     

Causal Diagram Techniques for Urologic Oncology Research

Large quantities of data are now available to medical researchers; however, observational studies are plagued by bias and confounding. Additionally, much of this research only speculates on variable associations, leaving prospective randomized clinical trials as the sole purveyors of claims about causal relations between variables. There has been a growing movement of causal inference that uses new techniques to investigate causality using observational data. These techniques include the implementation of directed acyclic graphs, which allow researchers to explicitly and reproducibly define the causal relationships between study variables, thus making statistical analysis more robust. Directed acyclic graphs further allow researchers to identify confounding and other sources of bias and to discover causal effects among complex networks of variables. This review aims to introduce these techniques to the general urology and urologic oncology research communities in order to provide a basic understanding of causal inference and analysis and call for integration of these practices more generally in research methodology.     

基于大自然时间轴的测量时序分类法

因果推断中，时序（或方向性）的概念尚未完全明确。本文从因果思维出发，以真实因和真实果的发生时间将大自然时间轴划分为3个时域和2个时点，从而锚定了因果推断只能实现于第3时域。测量时序可分为5种类型：跨第1和3时域纵向时序（实验性时序）、跨第2和3时域纵向时序、同时域纵向时序、同时域逆纵向时序和同时域横向时序（观察性时序）。这种分类法适用于首次或多次测量、及时和延后测量等所有测量策略。除了实验中真实因的测量（或干预措施）在其发生之前（第1时域）或观察和实验中真实因的测量在真实果发生之前（第2时域）的情形外，所有其他测量策略类似于历史重建或"考古"，测量时序的重要性次于测量的准确性。从研究设计应整合偏倚设计的观点来看，本文提出基于大自然时间轴的测量时序五分类法，概念清楚并将有助于判断研究过程中可能出现的偏倚，为正确进行因果推断研究奠定基础。     

Combining individual and aggregated data to investigate the role of socioeconomic disparities on cancer burden in Italy

Quantifying socioeconomic disparities and understanding the roots of inequalities are growing topics in cancer research. However, socioeconomic differences are challenging to investigate mainly due to the lack of accurate data at individual-level, while aggregate indicators are only partially informative. We implemented a multiple imputation algorithm within a statistical matching framework that combines diverse sources of data to estimate individual-level associations between income and risk of breast and lung cancer, adjusting for potential confounding factors in Italy. The framework is computationally flexible and can be adapted to similar contexts.