Using linkage information to weight a genome-wide association of bipolar disorder.

Issues of multiple-testing and statistical significance in genomewide association studies (GWAS) have prompted statistical methods utilizing prior data to increase the power of association results. Using prior findings from genome-wide linkage studies on bipolar disorder (BPD), we employed a weighted false discovery approach (wFDR; [Roeder et al. 2006. Am J Hum Genet 78(2): 243­252]) to previously reported GWAS data drawn from the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD). Using this method, association signals are up or down-weighted given the linkage score in that genomic region. Although no SNPs in our sample reached genome-wide significance through the wFDR approach, the strongest single SNP result from the original GWAS results (rs4939921 in myosin VB) is strongly up-weighted as it occurs on a linkage peak of chromosome 18. We also identify regions on chromosome 9, 17, and 18 where modestly associated SNP clusters coincide with strong linkage scores, implicating them as possible candidate regions for further analysis. Moving forward, we believe the application of prior linkage information will be increasingly useful to future GWAS studies that incorporate rarer variants into their analysis.

Fitting ACE structural equation models to case-control family data.

Investigators interested in whether a disease aggregates in families often collect case-control family data, which consist of disease status and covariate information for members of families selected via case or control probands. Here, we focus on the use of case-control family data to investigate the relative contributions to the disease of additive genetic effects (A), shared family environment (C), and unique environment (E). We describe an ACE model for binary family data; this structural equation model, which has been described previously, combines a general-family extension of the classic ACE twin model with a (possibly covariate-specific) liability-threshold model for binary outcomes. We then introduce our contribution, a likelihood-based approach to fitting the model to singly ascertained case-control family data. The approach, which involves conditioning on the proband's disease status and also setting prevalence equal to a prespecified value that can be estimated from the data, makes it possible to obtain valid estimates of the A, C, and E variance components from case-control (rather than only from population-based) family data. In fact, simulation experiments suggest that our approach to fitting yields approximately unbiased estimates of the A, C, and E variance components, provided that certain commonly made assumptions hold. Further, when our approach is used to fit the ACE model to Austrian case-control family data on depression, the resulting estimate of heritability is very similar to those from previous analyses of twin data.

Obesity and weight gain in relation to depression: findings from the Stirling County Study.

OBJECTIVE: This study concerns the question of whether obese subjects in a community sample experience depression in a different way from the nonobese, especially whether they overeat to the point of gaining weight during periods of depression. DESIGN: A representative sample of adults was interviewed regarding depression and obesity. SUBJECTS: The sample consisted of 1396 subjects whose interviews were studied regarding relationships between obesity and depression and among whom 114 had experienced a major depressive episode at some point in their lives and provided information about the symptoms experienced during the worst or only episode of major depression. MEASUREMENTS: The Diagnostic Interview Schedule (DIS) was used to identify major depressive episodes. Information was also derived from the section on Depression and Anxiety (DPAX) of the Stirling Study Schedule. Obesity was calculated as a body mass index >30. Logistic regressions were employed to assess relationships, controlling for age and gender, by means of odds ratios and 95% confidence intervals. RESULTS: In the sample as a whole, obesity was not related to depression although it was associated with the symptom of hopelessness. Among those who had ever experienced a major depressive episode, obese persons were 5 times more likely than the nonobese to overeat leading to weight gain during a period of depression (P<0.002). These obese subjects, compared to the nonobese, also experienced longer episodes of depression, a larger number of episodes, and were more preoccupied with death during such episodes. CONCLUSIONS: Depression among obese subjects in a community sample tends to be more severe than among the nonobese. Gaining weight while depressed is an important marker of that severity. Further research is needed to understand and possibly prevent the associations, sequences and outcomes among depression, obesity, weight gain and other adversities.

A new statistical screening approach for finding pharmacokinetics-related genes in genome-wide studies.

Biomedical researchers usually test the null hypothesis that there is no difference of the population mean of pharmacokinetics (PK) parameters between genotypes by the Kruskal-Wallis test. Although a monotone increasing pattern with a number of alleles is expected for PK-related genes, the Kruskal-Wallis test does not consider a monotonic response pattern. For detecting such patterns in clinical and toxicological trials, a maximum contrast method has been proposed. We show how that method can be used with pharmacogenomics data to a develop test of association. Further, using simulation studies, we compare the power of the modified maximum contrast method to those of the maximum contrast method and the Kruskal-Wallis test. On the basis of the results of those studies, we suggest rules of thumb for which statistics to use in a given situation. An application of all three methods to an actual genome-wide pharmacogenomics study illustrates the practical relevance of our discussion.

A study of how socioeconomic status moderates the relationship between SNPs encompassing BDNF and ADHD symptom counts in ADHD families.

Recent animal research suggests that brain-derived neurotrophic factor (BDNF), may mediate response to different environmental stimuli. In this paper, we evaluated the possible role of BDNF as a moderator of attention deficit hyperactivity disorder (ADHD) in the context of different socioeconomic classes. We genotyped ten single nucleotide polymorphisms (SNPs) in and around BDNF in 229 families and evaluate whether there are SNP-by-socioeconomic status (SES) interactions for attention deficit hyperactivity. We developed three quantitative phenotypes for ADHD from nine inattentive and nine hyperactive-impulsive symptoms that were used in SNP-by-SES interaction analyses using a new methodology implemented in the computer program PBAT. Findings were adjusted for multiple comparisons using the false discovery rate. We found multiple significant SNP-by-SES interactions using the inattentive symptom count. This study suggests that different SES classes may modify the effect of the functional variant(s) in and around BDNF to have an impact on the number of ADHD symptom counts that are observed. The two exons within BDNF represent potential functional variants that may be causing the observed associations.

Approaches to handling incomplete data in family-based association testing.

The high throughput of data arising from the complete sequence of the human genome has left statistical geneticists with a rich and extensive information source. The wide availability of software and the increase in computing power has improved the possibilities to access and process such data. One problem is incompleteness of the data: unobserved or partially observed data points due to technical reasons or reasons associated with the patient's status or erroneous measurements of phenotype or genotype, to name a few. When not properly accounted for, these sources of incompleteness may seriously jeopardize the credibility of results from analyses. In this paper we provide some perspectives on the occurrence and analysis of different forms of incomplete data in family-based genetic association testing.

Incidence of adverse events and negligence in hospitalized patients: results of the Harvard Medical Practice Study I. 1991.

BACKGROUND: As part of an interdisciplinary study of medical injury and malpractice litigation, we estimated the incidence of adverse events, defined as injuries caused by medical management, and of the subgroup of such injuries that resulted from negligent or substandard care. METHODS: We reviewed 30121 randomly selected records from 51 randomly selected acute care, non-psychiatric hospitals in New York State in 1984. We then developed population estimates of injuries and computed rates according to the age and sex of the patients as well as the specialties of the physicians. RESULTS: Adverse events occurred in 3.7% of the hospitalizations (95% confidence interval 3.2 to 4.2), and 27.6% of the adverse events were due to negligence (95% confidence interval 22.5 to 32.6). Although 70.5% of the adverse events gave rise to disability lasting less than 6 months, 2.6% caused permanently disabling injuries and 13.6% led to death. The percentage of adverse events attributable to negligence increased in the categories of more severe injuries (Wald test chi(2) = 21.04, p<0.0001). Using weighted totals we estimated that among the 2671863 patients discharged from New York hospitals in 1984 there were 98609 adverse events and 27179 adverse events involving negligence. Rates of adverse events rose with age (p<0.0001). The percentage of adverse events due to negligence was markedly higher among the elderly (p<0.01). There were significant differences in rates of adverse events among categories of clinical specialties (p<0.0001), but no differences in the percentage due to negligence. CONCLUSIONS: There is a substantial amount of injury to patients from medical management, and many injuries are the result of substandard care.

Anxiety and depression: a 40-year perspective on relationships regarding prevalence, distribution, and comorbidity.

OBJECTIVE: Building on a report about the prevalence of depression over time, this paper examines historical trends regarding anxiety in terms of its prevalence, its distribution by age and gender, and its comorbidity with depression. Methods for conducting such time trend analysis are reviewed. METHOD: Representative samples of adults were selected and interviewed in 1952, 1970, and 1992. Logistic regressions were used for statistical analysis. RESULTS: Although twice as common as depression, the prevalence of anxiety was equally stable. Anxiety was consistently and significantly more characteristic of women than men. A re-distribution of rates in 1992 indicated that depression but not anxiety had significantly increased among younger women (P = 0.03). Throughout the study, approximately half of the cases of anxiety also suffered depression. CONCLUSION: The relationships between anxiety and depression remained similar over time with the exception that depression came to resemble anxiety as a disorder to which women were significantly more vulnerable than men. Social and historical factors should be investigated to assess their relevance to this change.

Tests of gene-environment interaction for case-parent triads with general environmental exposures.

As knowledge of the human genome continues to grow, more progress is being made towards not only identifying the genes involved in disease susceptibility but also in defining the synergistic role genes play with environmental exposures. The detection of gene-environment interactions is important as it can offer clinicians a potential means of intervention. The discovery of interactions relies heavily on powerful statistical methods. We present a test, FBAT-I, that can be used to investigate gene-environment interaction. The test uses the case-parent triad design and protects the statistical inference from potential spurious results due to population admixture.

Family study of the aggregation of eating disorders and mood disorders.

BACKGROUND: Family studies have suggested that eating disorders and mood disorders may coaggregate in families. To study further this question, data from a family interview study of probands with and without major depressive disorder was examined. METHOD: A bivariate proband predictive logistic regression model was applied to data from a family interview study, conducted in Innsbruck, Austria, of probands with (N = 64) and without (N = 58) major depressive disorder, together with 330 of their first-degree relatives. RESULTS: The estimated odds ratio (OR) for the familial aggregation of eating disorders (anorexia nervosa, bulimia nervosa and binge-eating disorder) was 7.0 (95 % CI 1.4, 28; P = 0.006); the OR for the familial aggregation of mood disorders (major depression and bipolar disorder) was 2.2 (0.92, 5.4; P = 0.076); and for the familial coaggregation of eating disorders with mood disorders the OR was 2.2 (1.1, 4.6; P = 0.035). CONCLUSIONS: The familial coaggregation of eating disorders with mood disorders was significant and of the same magnitude as the aggregation of mood disorders alone--suggesting that eating disorders and mood disorders have common familial causal factors.

Estimation and tests of haplotype-environment interaction when linkage phase is ambiguous.

In the study of complex traits, the utility of linkage analysis and single marker association tests can be limited for researchers attempting to elucidate the complex interplay between a gene and environmental covariates. For these purposes, tests of gene-environment interactions are needed. In addition, recent studies have indicated that haplotypes, which are specific combinations of nucleotides on the same chromosome, may be more suitable as the unit of analysis for statistical tests than single genetic markers. The difficulty with this approach is that, in standard laboratory genotyping, haplotypes are often not directly observable. Instead, unphased marker phenotypes are collected. In this article, we present a method for estimating and testing haplotype-environment interactions when linkage phase is potentially ambiguous. The method builds on the work of Schaid et al. [2002] and is applicable to any trait that can be placed in the generalized linear model framework. Simulations were run to illustrate the salient features of the method. In addition, the method was used to test for haplotype-smoking exposure interaction with data from the Childhood Asthma Management Program.

Multiple informants: mortality associated with psychiatric disorders in the Stirling County Study.

This paper applies new statistical procedures for analyzing multiple-source information about the relation of psychiatric diagnoses to mortality. The data come from the Stirling County Study, a longitudinal community investigation of adults, that collected multiple-source reports (self-report and physician-report) about psychiatric disorders. These reports are used as predictors of mortality risk over a 16-year follow-up period (1952-1968). Despite extensive efforts, one or both of these reports were sometimes missing. Missingness of self-report was related to demographic characteristics as well as to physician-reports of psychiatric diagnosis. The statistical procedures used here draw together into a single frame of reference both informant reports for the initial Stirling survey and relate these to mortality risk using weighted generalized estimating equation regression models for time to event data. This unified method has two advantages over traditional approaches: 1) the relative predictiveness of each informant can be assessed and 2) all subjects contribute to the analysis. The methods are applicable to other areas of epidemiology where multiple informant reports are used. The results for self-reports and physician-reports of disorders were comparable: Psychiatric diagnosis was associated with higher mortality, particularly among younger subjects.

The family based association test method: strategies for studying general genotype--phenotype associations.

With possibly incomplete nuclear families, the family based association test (FBAT) method allows one to evaluate any test statistic that can be expressed as the sum of products (covariance) between an arbitrary function of an offspring's genotype with an arbitrary function of the offspring's phenotype. We derive expressions needed to calculate the mean and variance of these test statistics under the null hypothesis of no linkage. To give some guidance on using the FBAT method, we present three simple data analysis strategies for different phenotypes: dichotomous (affection status), quantitative and censored (eg, the age of onset). We illustrate the approach by applying it to candidate gene data of the NIMH Alzheimer Disease Initiative. We show that the RC-TDT is equivalent to a special case of the FBAT method. This result allows us to generalise the RC-TDT to dominant, recessive and multi-allelic marker codings. Simulations compare the resulting FBAT tests to the RC-TDT and the S-TDT. The FBAT software is freely available.

Maximum likelihood analysis of logistic regression models with incomplete covariate data and auxiliary information.

This article presents a new method for maximum likelihood estimation of logistic regression models with incomplete covariate data where auxiliary information is available. This auxiliary information is extraneous to the regression model of interest but predictive of the covariate with missing data. Ibrahim (1990, Journal of the American Statistical Association 85, 765-769) provides a general method for estimating generalized linear regression models with missing covariates using the EM algorithm that is easily implemented when there is no auxiliary data. Vach (1997, Statistics in Medicine 16, 57-72) describes how the method can be extended when the outcome and auxiliary data are conditionally independent given the covariates in the model. The method allows the incorporation of auxiliary data without making the conditional independence assumption. We suggest tests of conditional independence and compare the performance of several estimators in an example concerning mental health service utilization in children. Using an artificial dataset, we compare the performance of several estimators when auxiliary data are available.

An alternative parameterization of the general linear mixture model for longitudinal data with non-ignorable drop-outs.

This paper considers the mixture model methodology for handling non-ignorable drop-outs in longitudinal studies with continuous outcomes. Recently, Hogan and Laird have developed a mixture model for non-ignorable drop-outs which is a standard linear mixed effects model except that the parameters which characterize change over time depend also upon time of drop-out. That is, the mean response is linear in time, other covariates and drop-out time, and their interactions. One of the key attractions of the mixture modelling approach to drop-outs is that it is relatively easy to explore the sensitivity of results to model specification. However, the main drawback of mixture models is that the parameters that are ordinarily of interest are not immediately available, but require marginalization of the distribution of outcome over drop-out times. Furthermore, although a linear model is assumed for the conditional mean of the outcome vector given time of drop out, after marginalization, the unconditional mean of the outcome vector is not, in general, linear in the regression parameters. As a result, it is not possible to parsimoniously describe the effects of covariates on the marginal distribution of the outcome in terms of regression coefficients. The need to explicitly average over the distribution of the drop-out times and the absence of regression coefficients that describe the effects of covariates on the outcome are two unappealing features of the mixture modelling approach. In this paper we describe a particular parameterization of the general linear mixture model that circumvents both of these problems.

Multivariate logistic regression for familial aggregation of two disorders. I. Development of models and methods.

The question of whether two disorders cluster together, or coaggregate, within families often arises. This paper considers how to analyze familial aggregation of two disorders and presents two multivariate logistic regression methods that model both disorder outcomes simultaneously. The first, a proband predictive model, predicts a relative's outcomes (the presence or absence of each of the two disorders) by using the proband's disorder status. The second, a family predictive model derived from the quadratic exponential model, predicts a family member's outcomes by using all of the remaining family members' disorder statuses. The models are more realistic, flexible, and powerful than univariate models. Methods for estimation and testing account for the correlation of outcomes among family members and can be implemented by using commercial software.

Multivariate logistic regression for familial aggregation of two disorders. II. Analysis of studies of eating and mood disorders.

Family studies have suggested that eating disorders and mood disorders may coaggregate within families. Previous studies, however, have been limited by use of univariate modeling techniques and failure to account for the correlation of observations within families. To provide a more efficient analysis and to illustrate multivariate logistic regression models for familial aggregation of two disorders, the authors analyzed pooled data from two previously published family studies (conducted in Massachusetts in 1984-1986 and 1986-1987) by using multivariate proband predictive and family predictive models. Both models demonstrated a significant familial aggregation of mood disorders and familial coaggregation of eating and mood disorders. The magnitude of the coaggregation between eating and mood disorders was similar to that of the aggregation of mood disorders. Similar results were obtained with alternative models, including a traditional univariate proband predictive model. In comparison with the univariate model, the multivariate models provided greater flexibility, improved precision, and wider generality for interpreting aggregation effects.

Tutorial in Biostatistics: Evaluating the impact of 'critical periods' in longitudinal studies of growth using piecewise mixed effects models.

Recent developments in modern multivariate methods provide applied researchers with the means to address many important research questions that arise in studies with repeated measures data collected on individuals over time. One such area of applied research is focused on studying change associated with some event or critical period in human development. This tutorial deals with the use of the general linear mixed model for regression analysis of correlated data with a two-piece linear function of time corresponding to the pre- and post-event trends. The model assumes a continuous outcome is linearly related to a set of explanatory variables, but allows for the trend after the event to be different from the trend before it. This task can be accomplished using a piecewise linear random effects model for longitudinal data where the response depends upon time of the event. A detailed example that examines the influence of menarche on changes in body fat accretion will be presented using data from a prospective study of 162 girls measured annually from approximately age 10 until 4 years post menarche.

Family-based tests of association in the presence of linkage.

Linkage analysis may not provide the necessary resolution for identification of the genes underlying phenotypic variation. This is especially true for gene-mapping studies that focus on complex diseases that do not exhibit Mendelian inheritance patterns. One positional genomic strategy involves application of association methodology to areas of identified linkage. Detection of association in the presence of linkage localizes the gene(s) of interest to more-refined regions in the genome than is possible through linkage analysis alone. This strategy introduces a statistical complexity when family-based association tests are used: the marker genotypes among siblings are correlated in linked regions. Ignoring this correlation will compromise the size of the statistical hypothesis test, thus clouding the interpretation of test results. We present a method for computing the expectation of a wide range of association test statistics under the null hypothesis that there is linkage but no association. To standardize the test statistic, an empirical variance-covariance estimator that is robust to the sibling marker-genotype correlation is used. This method is widely applicable: any type of phenotypic measure or family configuration can be used. For example, we analyze a deletion in the A2M gene at the 5' splice site of "exon II" of the bait region in Alzheimer disease (AD) discordant sibships. Since the A2M gene lies in a chromosomal region (chromosome 12p) that consistently has been linked to AD, association tests should be conducted under the null hypothesis that there is linkage but no association.

[Validity and reproducibility of a questionnaire on physical activity and non-activity for school children in Mexico City]. / Validez y reproducibilidad de un cuestionario de actividad e inactividad física para escolares de la ciudad de México.

OBJECTIVE: To assess the validity and reproducibility of a self-reported questionnaire on physical activity and inactivity, developed for children aged 10-14 in Mexico City. MATERIAL AND METHODS: Between May and December 1996, a self-reported physical activity and inactivity questionnaire was developed and applied twice to a sample of 114 students aged 10 to 14, from a low and middle income population of Mexico City. The children's mothers completed the same questionnaire, and two 24-hour recalls of physical activity were used for comparison. Statistical analysis consisted of central tendency and dispersion measures and Pearson's correlation coefficient. RESULTS: Correlations between hours per day spent in physical activity and inactivity from the children's questionnaire and the 24-hour recall data, were 0.03 for moderate activity, 0.15 for vigorous activity, and 0.51 (p = 0.001) for watching television, adjusted by age, gender, town, and illness prior to the administration of the questionnaire. Compared to the 24-hour recall data, the questionnaire overestimated the time spent watching television, reading or participating in vigorous activity, and underestimated the time engaged in moderate activity. Statistically significant (p < 0.05) six-month reproducibility values were observed for watching television (r = 0.53), sleeping (r = 0.40), moderate (r = 0.38), and vigorous activity (r = 0.55). CONCLUSIONS: Among children of Mexico City aged 10-14, the questionnaire showed acceptable validity in estimating the time watching television, and acceptable reproducibility of the time watching television, vigorous and moderate activity.