Power Of Case-control Studies Research Articles

A retrospective likelihood approach for efficient integration of multiple omics factors in case-control association studies.

Integrative omics, the joint analysis of outcome and multiple types of omics data, such as genomics, epigenomics, and transcriptomics data, constitute a promising approach for powerful and biologically relevant association studies. These studies often employ a case-control design, and often include nonomics covariates, such as age and gender, that may modify the underlying omics risk factors. An open question is how to best integrate multiple omics and nonomics information to maximize statistical power in case-control studies that ascertain individuals based on the phenotype. Recent work on integrative omics have used prospective approaches, modeling case-control status conditional on omics, and nonomics risk factors. Compared to univariate approaches, jointly analyzing multiple risk factors with a prospective approach increases power in nonascertained cohorts. However, these prospective approaches often lose power in case-control studies. In this article, we propose a novel statistical method for integrating multiple omics and nonomics factors in case-control association studies. Our method is based on a retrospective likelihood function that models the joint distribution of omics and nonomics factors conditional on case-control status. The new method provides accurate control of Type I error rate and has increased efficiency over prospective approaches in both simulated and real data.

Informed Conditioning on Clinical Covariates Increases Power in Case-Control Association Studies

Genetic case-control association studies often include data on clinical covariates, such as body mass index (BMI), smoking status, or age, that may modify the underlying genetic risk of case or control samples. For example, in type 2 diabetes, odds ratios for established variants estimated from low–BMI cases are larger than those estimated from high–BMI cases. An unanswered question is how to use this information to maximize statistical power in case-control studies that ascertain individuals on the basis of phenotype (case-control ascertainment) or phenotype and clinical covariates (case-control-covariate ascertainment). While current approaches improve power in studies with random ascertainment, they often lose power under case-control ascertainment and fail to capture available power increases under case-control-covariate ascertainment. We show that an informed conditioning approach, based on the liability threshold model with parameters informed by external epidemiological information, fully accounts for disease prevalence and non-random ascertainment of phenotype as well as covariates and provides a substantial increase in power while maintaining a properly controlled false-positive rate. Our method outperforms standard case-control association tests with or without covariates, tests of gene x covariate interaction, and previously proposed tests for dealing with covariates in ascertained data, with especially large improvements in the case of case-control-covariate ascertainment. We investigate empirical case-control studies of type 2 diabetes, prostate cancer, lung cancer, breast cancer, rheumatoid arthritis, age-related macular degeneration, and end-stage kidney disease over a total of 89,726 samples. In these datasets, informed conditioning outperforms logistic regression for 115 of the 157 known associated variants investigated (P-value = 1×10−9). The improvement varied across diseases with a 16% median increase in χ2 test statistics and a commensurate increase in power. This suggests that applying our method to existing and future association studies of these diseases may identify novel disease loci.

Flexible matching in case-control studies of gene-environment interactions.

Because of the lack of power of case-control study designs to detect gene-environment interactions, flexible matching has recently been proposed as a method of improving efficiency. In this paper, the authors consider a large-sample approximation method that allows estimation of the most efficient matching strategy when genotype and exposure are either independent or associated. The authors provide tables of the sample sizes required to detect gene-environment interactions if this flexible matching strategy is followed, and they make brief comparisons with other study designs.

Degree of matching and gain in power and efficiency in case-control studies.

Frequency matching can be used to increase the precision and power of case-control studies. Unmatched and frequency-matched designs are only two distinct possibilities of control selection in a continuum regarding the frequency of the matching factor in controls. We assessed the power and efficiency of case-control studies under a variety of assumptions regarding the prevalence and the effects of the matching factor and the exposure of interest as well as their association in the population. For each set of parameters, we simulated 10,000 case-control studies varying the degree of matching, that is, the proportion of the matching factor in selected controls over a wide range including the proportion in cases (matched design) and the population (unmatched design) as two special options. Traditional frequency matching increased the precision and power in most scenarios, but most of the gain was often achieved by incomplete (less than perfect) matching. Even greater gains were sometimes observed by increasing the prevalence of the matching factor in controls above the one in cases. In the scenarios assessed, perfect matching was neither necessary nor the optimum degree of matching in many circumstances. It might be worthwhile to evaluate the optimum degree of matching for specific settings in the design of case-control studies.

Exposure measurement errors, risk estimate and statistical power in case-control studies using dichotomous analysis of a continuous exposure variable.

Non-differential errors in exposure measurements have been shown to lead to differential misclassification of exposure. As a consequence, the common tenet that, in absence of bias, imprecise exposure assessment can only bias the risk estimates conservatively does not necessarily hold. We investigate the effects of exposure measurement errors on the risk estimate and on statistical power. We used a computer model that simulates a case-control study. We used both hypothetical data and data modelled on empirical measurements of environmental magnetic fields exposure. Measurement errors are found to have a lesser impact on risk estimates and statistical power than would have been the case had misclassification been truly non-differential. However, for a given cutpoint, a bias away from the null cannot be excluded. The predominant direction of the errors is found to have important consequences on both the study power and the risk estimates. When sufficient empirical data are available, computer modelling may give a more accurate estimate of the effects of measurement errors than algebraic corrections.

The National Bladder Cancer Study: Employment in the Chemical Industry<xref ref-type="fn" rid="FN2">2</xref>

The relationship between bladder cancer employment in the chemical industry was assessed in a study of 2,982 incident cases and 5,782 population controls. There were 190 cases and 369 controls who had ever been employed in the chemical industry [odds ratio (OR) = 1.0; 95% confidence interval (CI) = 0.8, 1.2]. Employment in the production of organic chemicals was associated with a 1.3-fold increased risk among men (95% CI = 0.8, 2.1). Risk increased with duration of employment, reaching an OR of 2.4 for 20 or more years (chi for trend = 1.57; P = .06). Women who had worked in the plastics industry had a 3.3-fold increased bladder cancer risk. Within the plastics and rubber industry, increased risks for bladder cancer were found for men in mixing, filtering, grinding, and other dusty operations (OR = 4.6; 95% CI = 1.0, 20.4) and men in heat-associated operations (OR = 2.8; 95% CI = 0.5, 15.3). A 1.4-fold risk among men in agricultural chemicals was attributable to risks in the pesticides subdivision (OR = 2.3; 95% CI = 0.6, 8.2). Men performing dusty operations (i.e., mixing, filtering, sifting, grinding, and crushing) in any industry had an OR of 1.4 (95% CI = 0.8, 2.7). Despite the large number of study subjects, few statistically significant findings were observed and should be evaluated with consideration of the large number of comparisons made in the analysis. The statistical power of case-control studies to detect risks associated with particular occupational exposures is limited by the small proportion of the population employed in any specific occupation or industry.

The use of polytomous dual response data to increase power in case-control studies: An application to the association between dietary fat and breast cancer

Computer simulation has been used to compare various methods for analyzing data from case-control studies when two response variables are available to measure a risk factor. When such variables are continuous or polytomous, and a dose-response relationship exists, it is concluded that power may be increased by including individuals who are discordant on their responses. Failure to use the most appropriate method can lead to a substantial loss in power. The results are illustrated by reanalysis of data from an earlier case-control study of diet and breast cancer. This reanalysis has strengthened the evidence for a positive association between saturated fat intake and risk of breast cancer, with some suggestion that this association is limited to premenopausal women, in whom the increase in risk may be substantial.

Case definition and power in case-control studies.

A narrow case definition can lead to greater in clarity in case-control studies, but it can also lead to either a gain or a loss in the power of statistical tests. Under plausible assumptions, the loss of power can be in excess of 30 per cent, and the gain can be in excess of 50 per cent. If at least one third of the total study size consists of narrowly defined cases, the loss of power will often be negligible, whereas the gain can be substantial.