Selection bias and confounding in case-crossover analyses of environmental time-series data.

The case-crossover study design is a popular analytic tool for estimating the effects of triggers of acute outcomes by environmental exposures. Although this approach controls for time-invariant confounders by design, it may allow for selection bias and confounding by time-varying factors. We conducted a simulation study of the sensitivity of the symmetric bidirectional case-crossover design to time-varying patterns in exposure and outcome. We identified the effects of selection bias and confounding on symmetric bidirectional case-crossover results and offer strategies to eliminate or reduce these biases. Selection bias results when exposure in the reference periods is not identically representative of exposure in the hazard periods, even when the distribution of exposure is stationary. This bias can be estimated and removed. Selection bias also occurs when the distribution of exposure is nonstationary, but the adjusted symmetric bidirectional case-crossover methodology substantially controls for this. Confounding results from a common temporal pattern in the exposure and the outcome time series, but can also be the result of patterns in exposure and outcome that, although asymptotically uncorrelated, are correlated at finite series lengths. All three biases are reduced by choosing shorter referent-spacing lengths. This effect is illustrated using data on air pollution and daily deaths in Chicago.     

A regression shrinkage method tailored to qualitative regressors and clustered data

We propose a shrinkage method for estimation in linear regression models with qualitative regressors. Due to the nature of the shrinkage constraint, this method tends to give estimates that are exactly zero for some groups of coefficients belonging to the same regressor. The method hence results in concise models, since some of the regressors are entirely eliminated. In conjunction with this estimation method, a model with a fixed cluster effect turns out to be closely related to frailty models. We apply the method for modelling hospital readmissions.     

Bayesian inference from incomplete longitudinal data: A simple method to quantify sensitivity to nonignorable dropout

Bayesian approach has been increasingly used for analyzing longitudinal data. When dropout occurs in the study, analysis often relies on the assumption of ignorable dropout. Because ignorability is a critical and untestable assumption without obtaining additional data or making other unverifiable assumptions, it is important to assess the impact of departures from the ignorability assumption on the key Bayesian inferences. In this paper, we extend the Bayesian index of local sensitivity to non‐ignorability (ISNI) method proposed by Zhang and Heitjan to longitudinal data with dropout. We derive formulas for the Bayesian ISNI when the complete longitudinal data follow a linear mixed‐effect model. The calculation of the index only requires the posterior draws or summary statistics of these draws from the standard analysis of the ignorable model. Thus, our approach avoids fitting any complicated nonignorable model. One can use the method to evaluate which Bayesian parameter estimates or functions of these estimates in a linear mixed‐effect model are susceptible to nonignorable dropout and which ones are not. We illustrate the method using a simulation study and two real examples: rats data set and rheumatoid arthritis clinical trial data set. Copyright © 2009 John Wiley & Sons, Ltd.     

Non-parametric covariance methods for incidence density analyses of time-to-event data from a randomized clinical trial and their complementary roles to proportional hazards regression

The principal response criteria for many clinical trials involve time-to-event variables. Usual methods of analysis for this type of response criterion include product-limit estimators of cumulative survival for the treatment groups, (stratified) logrank tests to compare treatments, and proportional hazards regression models with treatment and relevant covariates. When adjustment for covariates is of some importance, the relative roles of these methods may be of some concern, particularly for confirmatory clinical trials which must provide convincing findings to regulatory agencies. Unadjusted methods may have lower power, but there are issues regarding adjustment for covariates that may be controversial. These issues include applicability of proportional hazards assumptions, whether the correct model has been specified, and whether there is parallelism between treatments for relationships with covariates. One way to address these issues is to use non-parametric analysis of covariance strategies with extensions to log incidence density estimation. The principal basis for this method is no association between covariates and treatment groups as provided by randomized assignment of patients to groups. The background theory and strategies for computation are described for this method. Aspects of its application are illustrated for a clinical trial with two treatment groups and 722 patients. The objective of analysis for this clinical trial is evaluation of treatment effects with and without adjustment for 22 a priori covariates and a stratification for three geographical regions.     

A logistic regression model applied to Chagas' disease

The focus of this paper is the application of statistical models to the study of socioeconomic conditioning factors in perinatal Chagas' disease conducted in Rosario, Argentina. A case (154) and control (158) design was applied to investigate socioeconomic and cultural differences in pregnant women in Hospital Roque Sáenz Pe?a as to their infection status. Logistic regression models were used to evaluate the importance of antecedents linked to the infection and socioeconomic and cultural factors for infection status. For pregnant women, the importance of antecedents linked to the infection was confirmed and the women's level of schooling stood out as the predominant socioeconomic condition associated with infection. Log-linear models were used to explore the associations between certain explanatory variables. This approach pointed up the most relevant associations between such factors and Chagas' disease and provided a better understanding of the framework of relationships among them.     

Inflation of the type I error rate when a continuous confounding variable is categorized in logistic regression analyses

This paper demonstrates an inflation of the type I error rate that occurs when testing the statistical significance of a continuous risk factor after adjusting for a correlated continuous confounding variable that has been divided into a categorical variable. We used Monte Carlo simulation methods to assess the inflation of the type I error rate when testing the statistical significance of a risk factor after adjusting for a continuous confounding variable that has been divided into categories. We found that the inflation of the type I error rate increases with increasing sample size, as the correlation between the risk factor and the confounding variable increases, and with a decrease in the number of categories into which the confounder is divided. Even when the confounder is divided in a five-level categorical variable, the inflation of the type I error rate remained high when both the sample size and the correlation between the risk factor and the confounder were high.     

A regression method for analysing ordinal data from intervention trials

We consider the case of a community intervention trial evaluated with a series of cross-sectional surveys and having outcomes measured on an ordinal scale. We propose a modelling procedure that combines ridit analysis and linear regression methods. We use the multinomial distribution as the basis for variance estimation of the mean ridits and then use simple regression models to estimate differences (for example, between intervention and comparison areas) among the ridits. We illustrate this procedure with data from a community intervention trial promoting condom use, with the adoption of consistent condom use measured on a 5-point ordinal scale.     

基于正则化回归的变量选择方法在高维数据中的应用

变量筛选和模型估计一直是高维数据的研究热点,而高维数据的维度灾难问题日渐突出,传统的统计分析方法因模型不稳定不再适用,本文对高维数据中基于正则化回归的变量选择方法的原理、适用的数据类型及优缺点、调整参数的选择进行综述。     

Measures of clustering and heterogeneity in multilevel Poisson regression analyses of rates/count data

Multilevel data occur frequently in many research areas like health services research and epidemiology. A suitable way to analyze such data is through the use of multilevel regression models. These models incorporate cluster‐specific random effects that allow one to partition the total variation in the outcome into between‐cluster variation and between‐individual variation. The magnitude of the effect of clustering provides a measure of the general contextual effect. When outcomes are binary or time‐to‐event in nature, the general contextual effect can be quantified by measures of heterogeneity like the median odds ratio or the median hazard ratio, respectively, which can be calculated from a multilevel regression model. Outcomes that are integer counts denoting the number of times that an event occurred are common in epidemiological and medical research. The median (incidence) rate ratio in multilevel Poisson regression for counts that corresponds to the median odds ratio or median hazard ratio for binary or time‐to‐event outcomes respectively is relatively unknown and is rarely used. The median rate ratio is the median relative change in the rate of the occurrence of the event when comparing identical subjects from 2 randomly selected different clusters that are ordered by rate. We also describe how the variance partition coefficient, which denotes the proportion of the variation in the outcome that is attributable to between‐cluster differences, can be computed with count outcomes. We illustrate the application and interpretation of these measures in a case study analyzing the rate of hospital readmission in patients discharged from hospital with a diagnosis of heart failure.     

The concept of residual confounding in regression models and some applications.

In this paper the concept of residual confounding is generalized to various types of regression models such as logistic regression or Cox regression. Residual confounding and a newly suggested parameter, the relative residual confounding, are defined on the regression parameters of the models. The estimator gives the proportion of confounding which has been removed by incomplete adjustment. The concept quantifies the effects of categorizing continuous covariables and of model misspecification. These are investigated by a simulation study and with data from an epidemiological investigation. A case-control study of laryngeal cancer is used to illustrate the residual confounding effect of arbitrary transformation of a continuous confounder, smoking, on the effect of alcohol consumption on laryngeal cancer risk. The data also showed that categorization into two levels can yield high residual confounding. The parameters described in this paper are of some use in quantifying the effect of inadequate adjustment for confounding variables.     

Inequities in CHD incidence and case fatality by neighborhood deprivation

BACKGROUND: Research has not firmly established whether living in a deprived neighborhood predicts the incidence and case fatality of coronary heart disease (CHD), and whether effects vary across sociodemographic groups. METHODS: Prospective follow-up study of all Swedish women and men, aged 35 to 74 (1.9 million women, 1.8 million men). Women and men, without a history of CHD, were assessed on December 31, 1995, and followed from January 1, 1996 through December 31, 2000, for first fatal or nonfatal CHD event (130,024 cases); data were analyzed in 2006. Neighborhood-level deprivation (index of education, income, unemployment, welfare assistance) was categorized as low, moderate, and high deprivation. RESULTS: Age-standardized CHD incidence was 1.9 times higher for women and 1.5 times higher for men in high- versus low-deprivation neighborhoods; 1-year case fatality from CHD was 1.6 times higher for women and 1.7 times higher for men in high versus low deprivation neighborhoods. The higher incidence in more deprived neighborhoods was observed across all individual-level sociodemographic groups (age, marital status, family income, education, immigration status, mobility, and urban/rural status). In multilevel logistic regression models, neighborhood deprivation remained significantly associated with both CHD incidence and case fatality for women and men after adjusting for the seven sociodemographic factors (p values <0.01). Effects were slightly stronger for women than men in an ancillary analysis that tested for gender differences. CONCLUSIONS: The clustering of CHD and subsequent mortality among adults in deprived neighborhoods raises important clinical and public health concerns, and calls for a reframing of health problems to include neighborhood social environments, as they may affect health.     

Bayesian analyses of longitudinal binary data using Markov regression models of unknown order

We present non-homogeneous Markov regression models of unknown order as a means to assess the duration of autoregressive dependence in longitudinal binary data. We describe a subject's transition probability evolving over time using logistic regression models for his or her past outcomes and covariates. When the initial values of the binary process are unknown, they are treated as latent variables. The unknown initial values, model parameters, and the order of transitions are then estimated using a Bayesian variable selection approach, via Gibbs sampling. As a comparison with our approach, we also implement the deviance information criterion (DIC) for the determination of the order of transitions. An example addresses the progression of substance use in a community sample of n = 242 American Indian children who were interviewed annually four times. An extension of the Markov model to account for subject-to-subject heterogeneity is also discussed.     

Semiparametric regression on cumulative incidence function with interval‐censored competing risks data

Many biomedical and clinical studies with time‐to‐event outcomes involve competing risks data. These data are frequently subject to interval censoring. This means that the failure time is not precisely observed but is only known to lie between two observation times such as clinical visits in a cohort study. Not taking into account the interval censoring may result in biased estimation of the cause‐specific cumulative incidence function, an important quantity in the competing risks framework, used for evaluating interventions in populations, for studying the prognosis of various diseases, and for prediction and implementation science purposes. In this work, we consider the class of semiparametric generalized odds rate transformation models in the context of sieve maximum likelihood estimation based on B‐splines. This large class of models includes both the proportional odds and the proportional subdistribution hazard models (i.e., the Fine–Gray model) as special cases. The estimator for the regression parameter is shown to be consistent, asymptotically normal and semiparametrically efficient. Simulation studies suggest that the method performs well even with small sample sizes. As an illustration, we use the proposed method to analyze data from HIV‐infected individuals obtained from a large cohort study in sub‐Saharan Africa. We also provide the R function ciregic that implements the proposed method and present an illustrative example. Copyright © 2017 John Wiley & Sons, Ltd.     

Colours of fruit and vegetables and 10-year incidence of CHD

The colours of the edible part of fruit and vegetables indicate the presence of specific micronutrients and phytochemicals. The extent to which fruit and vegetable colour groups contribute to CHD protection is unknown. We therefore examined the associations between fruit and vegetables of different colours and their subgroups and 10-year CHD incidence. We used data from a prospective population-based cohort including 20?069 men and women aged 20-65 years who were enrolled between 1993 and 1997. Participants were free of CVD at baseline and completed a validated 178-item FFQ. Hazard ratios (HR) for the association between green, orange/yellow, red/purple, white fruit and vegetables and their subgroups with CHD were calculated using multivariable Cox proportional hazards models. During 10 years of follow-up, 245 incident cases of CHD were documented. For each 25?g/d increase in the intake of the sum of all four colours of fruit and vegetables, a borderline significant association with incident CHD was found (HR 0·98; 95?% CI 0·97, 1·01). No clear associations were found for the colour groups separately. However, each 25?g/d increase in the intake of deep orange fruit and vegetables was inversely associated with CHD (HR 0·74; 95?% CI 0·55, 1·00). Carrots, their largest contributor (60?%), were associated with a 32?% lower risk of CHD (HR 0·68; 95?% CI 0·48, 0·98). In conclusion, though no clear associations were found for the four colour groups with CHD, a higher intake of deep orange fruit and vegetables and especially carrots may protect against CHD.     

A sequential Cox approach for estimating the causal effect of treatment in the presence of time‐dependent confounding applied to data from the Swiss HIV Cohort Study

When estimating the effect of treatment on HIV using data from observational studies, standard methods may produce biased estimates due to the presence of time‐dependent confounders. Such confounding can be present when a covariate, affected by past exposure, is both a predictor of the future exposure and the outcome. One example is the CD4 cell count, being a marker for disease progression for HIV patients, but also a marker for treatment initiation and influenced by treatment. Fitting a marginal structural model (MSM) using inverse probability weights is one way to give appropriate adjustment for this type of confounding. In this paper we study a simple and intuitive approach to estimate similar treatment effects, using observational data to mimic several randomized controlled trials. Each ‘trial’ is constructed based on individuals starting treatment in a certain time interval. An overall effect estimate for all such trials is found using composite likelihood inference. The method offers an alternative to the use of inverse probability of treatment weights, which is unstable in certain situations. The estimated parameter is not identical to the one of an MSM, it is conditioned on covariate values at the start of each mimicked trial. This allows the study of questions that are not that easily addressed fitting an MSM. The analysis can be performed as a stratified weighted Cox analysis on the joint data set of all the constructed trials, where each trial is one stratum. The model is applied to data from the Swiss HIV cohort study. Copyright © 2010 John Wiley & Sons, Ltd.     

A review of the use of routine data on CHD in general practice research in Scotland

BACKGROUND: The primary aim of the Platform Project is to maximise the use of routine data for primary care research in Scotland. AIMS: To assess the extent to which routine data is available and has been used in studies on CHD in General Practice research in Scotland. To assess the advantages and limitations of using routine data in this setting. METHODS: Literature review using a variety of databases catalogues and websites, bibliographies of articles retrieved and searching through journals by hand not available electronically. RESULTS: This review has found that the use of routine data in CHD studies in General Practice research in Scotland remains small. There has been little work undertaken which has combined the use of routine data with other research methods. Limitations with routine data exist particularly with regard to risk factors and ethnicity. However, despite such limitations there exists an increasingly extensive range of data, which exists to help explain tends in CHD, which so far has been largely underused.     

A grouped data regression approach to estimating economic and social influences on individual drinking behaviour

General Household Survey (GHS) data sets, covering the period 1978-1990, are pooled to investigate the relationship between the riskiness of individuals' self-reported drinking behaviour and a wide range of personal characteristics and economic factors. A grouped data regression approach is used to reduce problems with the inaccuracy of self-reports of alcohol consumption and clustering of observations in the consumption data. Results for males aged 18 to 24 years are presented, and possible methods for interpreting the results of grouped data regression are illustrated. Controlling for other factors, current smokers are estimated to be at a 75% higher risk of drinking over recommended levels than non-smokers. Particular attention is paid to the interactions between the price of alcohol, income and heavy drinking. At average levels of income, a 5% increase in the real price of alcohol is predicted to reduce the probability of 'at-risk' drinking by 1.5%. At lower initial levels of income, drinking patterns are found to be more responsive to both price and income changes. Grouped data regression is proposed as a way of focusing policy analysis on individual risks of alcohol-related health and social problems.     

Relationship of diet to the incidence of esophageal and stomach cancer in Japan

The food intake of households in 1,040 census tracts sampled in 1974, 1975, and 1976 for the nationwide nutrition survey of the Japanese Ministry of Health and Welfare, and the standardized mortality ratios (SMRs) of stomach and esophageal cancers from 1969 to 1974 for the cities, towns, and villages of the nutrition survey areas, were linked and their relationship was observed statistically by correlation analysis and multiple regression analysis. The results obtained are summarized as follows: