Grants & Funded Projects


1 R01 HL121330-01 (Dr. Donald Hedeker PI)
National Institute of Health (NHLBI) 05/01/2014 - 04/30/2018

Novel Statistical Models for EMA Studies of Physical Activity

This proposal responds to PAR-12-198 ("Improving Diet and Physical Activity Assessment"). It will develop and test novel multilevel statistical methods to examine the effects of subject-level parameters (variance and slope) of time-varying variables in ecological momentary assessment (EMA) studies of physical activity. Low level of physical activity heightens the risk of numerous deadly diseases (e.g., heart disease, stroke, cancer, diabetes) throughout the life course.

The use of EMA in physical activity research is growing rapidly because real-time data capture methods supply novel insights into determinants of this behavior. In EMA studies, it is common to have up to thirty or forty observations per subject, and this allows us to model subject-level parameters such as variances (e.g., how erratic is a subject’s mood?) and slopes (e.g., how much does a subject’s mood change across contexts?) of time-varying variables. For example, in our recent EMA work, we have found that more physically active children have greater positive and negative emotional stability than children who are less physically active. However, current multilevel modeling strategies are restricted to treating subject-level variances and slopes as outcomes. As a consequence, statistical models do not have the ability to test whether subject-level variance and slope parameters have predictive, mediating, and moderating effects on physical and sedentary activity. For example, we are unable to ask important research questions such as whether erratic mood mediates the effects of depression on physical activity, or whether the effects of living in a highly walkable neighborhood on physical activity are attenuated for individuals with unstable self-efficacy beliefs. This modeling restriction severely limits our ability to capitalize on the full potential of the time-varying nature of EMA data to enhance physical activity research.

To address this critical methodological gap, we propose to develop multilevel models, software, and strategies to test for the effects of these parameters in EMA studies. We will apply these modeling strategies to secondary analyses of pooled data from five federally- and foundation-supported EMA studies of physical activity with a combined sample size of N = 553 participants (including children and adults). The primary aims are:

(1) To develop novel multilevel modeling strategies and software to test whether subject-level variance and slope parameters have predictive, mediating, and moderating effects on subject-level physical and sedentary activity outcomes

(2) To apply these novel modeling strategies and software in secondary analyses of existing EMA datasets to examine the effects of subject-level variance and slopes of time-varying variables such as safety, stress, fatigue, and self-efficacy on physical and sedentary activity.

This study has the potential to make novel methodological and substantive contributions for analysis of EMA data in physical activity research. The methods to be developed can easily generalize to a variety of chronic disease-relevant research areas.

1 R01 MH100155-01 (Dr. Robert D. Gibbons PI) National Institute of Mental Health 05/27/2013 - 04/30/2018
A New Statistical Paradigm for Measuring Psychopathology Dimensions in Youth

Computerized adaptive testing (CAT), which was developed originally for educational measurement, offers extremely important advantages to the measurement of psychopathology. Traditional psychiatric measurement fixes the number of items and allows measurement precision to vary from subject to subject. In CAT, the numbers of items and the specific items that are administered are allowed to vary across individuals, but the precision of measurement is fixed. When used with a large bank of items, CAT dynamically selects a small and optimal set of items for each individual until a high and predefined level of measurement precision is achieved. This paradigm shift in measurement can achieve both substantially increased measurement precision and greatly decreased assessment times.

We propose to develop, test, and apply a new CAT approach to measuring severity of depression, anxiety, mania, disruptive behavior, and attention-deficit/hyperactivity disorders in children and adolescents (9-17 years). Building on our success in developing a CAT-based measure for assessing adult psychopathology1, this proposal contributes both methodologically and scientifically to research on the assessment of pediatric psychopathology. The proposed work will advance mental health research, improve psychiatric screening and monitoring in primary care.

The methodologic work proposed in this application is also driven by a fundamental scientific challenge that has limited progress in measuring psychopathology in pediatric populations. We need to understand how the measurement of psychopathology in youth changes from childhood through adolescence. Our proposed work includes new statistical methodology for CAT based on multidimensional Item Response Theory (IRT) that allows us to tailor measurement process to each child’s developmental level (vertical scaling). This methodologic advance will enable us to extend our accomplishments in measuring psychopathology in adults to youth. The overarching aim of this application is to develop a CAT for children and adolescents (Y-CAT-MH) that achieves the following goals:

Aim 1: Provides dimensional severity scores for depression, mania, anxiety, disruptive behavioral disorders (DBDs), and attention-deficit/hyperactivity disorder (ADHD).

Aim 2: Identifies children and adolescents who have symptom severity associated with functional impairment who would potentially benefit from a more extensive diagnostic assessment to evaluate the need for treatment.

Aim 3: Uses differential item functioning to identify a set of items that optimally discriminate high and low levels of severity for each of psychopathology dimension equally well for parent and child ratings of that dimension.

Aim 4: Accurately predicts DSM categorical diagnoses of major depressive disorder (MDD), ADHD, oppositional defiant disorder (ODD), conduct disorder (CD), anxiety disorders (AD; generalized anxiety disorder, separation anxiety disorder, social phobia, specific phobia), and bipolar disorder (BD).

Exploratory Aim: Using the same powerful psychometric strategies, we will take several important steps toward developing and testing of a parallel CAT measure of the core biopsychological processes identified in the Research Domain Criteria (RDoC).

To achieve these aims, we will develop a bank of items addressing at different developmental levels symptoms of depression (including a subdomain of suicidality), mania, anxiety, DBDs and ADHD, as well as positive and negative valence domains. In Phase 1, we will administer subsets of the items to a sample of 600 psychiatric outpatients and 200 control children not in psychiatric treatment, and to their primary caregivers. These data will be used to develop the Y-CAT-MH for parent and youth informants. In Phase 2, the Y-CAT-MH will be administered to a sample of 600 children with their primary caregiver (each of the 5 diagnostic domains represented by 100 youth, plus 100 control children). Phase 2 will include a psychiatric diagnostic interview and symptom severity assessment using validated instruments. Children with comorbid disorders will be included and dealt with in the analysis. In adults, the CAT-MH depression test required 2.7 minutes to administer an average of 12 items to maintain a correlation of r=0.95 with the bank of 389 depression items. Live CAT testing found sensitivity of 0.92 and specificity of 0.88 for a clinician-based SCID/DSM diagnosis of MDD. The test yields a severity score and a precise estimate of the probability of a DSM diagnosis of MDD.

1 R01 MH080122-01A2 (Dr. Robert D. Gibbons PI)
National Institute of Mental Health 05/01/2008 - 06/30/2013

Antidepressant Treatment and Suicidality: Biostatistical/Methodological Solutions

The enormous human cost of suicide makes research and prevention a national priority. Traditional approaches to drug safety have proved inadequate to address current public health issues. Providing statistical and methodological advances for the active surveillance of Drug-AE interactions, such as antidepressants and suicide, is a necessary first step in the establishment of a science-based health policy system, and is the primary goal of this proposal.

The purpose of this proposal is to develop, test, and apply new statistical methodologies that can be used to identify low base rate drug – adverse event (AE) interactions. These new methods will then be applied to a wide range of existing nonexperimental datasets to examine the relationship between antidepressants and suicide attempts and completion. We have designed this research project as an integral collaboration between biostatisticians, research psychiatrists and clinicians, economists, and pharmacoepidemiologists, working with large ecological and medical records/claims electronic patient databases covering years where antidepressant use is varying dramatically.

Three major specific aims are proposed:
(a) the development of new statistical surveillance methods for detecting drug-AE interactions,
(b) the development and application of analytic methods for ecological and population data, and
(c) the development of and application of statistical methods for analysis of person-level data
from large scale medical records/claims databases both in the U.S. and in Europe.

The datasets that we have been granted access to include the FDA spontaneous reporting system (SRS/AERS/MedWatch), electronic medical record/claims databases in the U.S. (VA, and PHARMetrics), and similar data from the Netherlands (PHARMO).

The work in this proposal will be carried out by a research consortium that will study national and international drug safety issues. The multidisciplinary group includes the areas of statistics (Drs. Gibbons, (PI), Brown (co-PI), Bhaumik, Hur, Marcus, and Rosenbaum), psychiatry from adult (Mann), and child (Brent) perspectives, health economics/econometrics (Heckman), and pharmacoepidemiology (Valuck). Collaboration with members of the VA (Cunningham), and PHARMO (Erkens, Herings) is an integral part of the proposal.

1P50HD055751-01 (Dr. Robert D. Gibbons PI of Statistical Core)
National Institute of Mental Health 08/06/2007 - 07/31/2012

Autism Center of Excellence: Translational Studies of Insistence of Sameness in Autism

The UIC ACE will focus over the next 5 years on the genetics, neurobiology, cognitive and affective processes, and pharmacology of insistence on sameness (IS) in autism spectrum disorders (ASD). A large sample of children with self-reported autism spectrum disorder will be screened by the Assessment Core for further screening by administration of the ADI-R to the parents. Profanes meeting ADI-R criteria for autistic disorder will be recruited for further study if they are also classified by the ADI-R IS items as high (N=150) or low IS (N=100). In addition, high IS subjects will need to score 15 or more on the sum of two IS factors on the RBS-R to avoid floor effects for the pharmacogenetic trial.

These 250 subjects will all be included in project I, Genetics of Serotonin in Autism: Neurochemical and Clinical Endophenotypes, along with 225 previously studied subjects and their parents for a total of 475 trios. This project will study 25 serotonin- related genes for association with autism and with IS more specifically. Resequencing of strong candidate genes will be conducted with all of the subjects in the pharmacogenetic Project III and with the low IS subjects in Project II. In addition, the 250 subjects will have serotonin measures collected for analysis with genetic and phenotype measures.

In Project II: Translational Studies of Cognitive, Affective and Neurochemical Processes Underlying Insistence on Sameness in Autism, fMRI studies of IS will be conducted on 50 high IS subjects also in Project III, 50 low IS subjects (also in Project I) and 50 control subjects. In addition, rat studies in which parallel behavioral and neurochemical approaches will be used.

Project III: The Pharmacogenetics of Treatment for Insistence on Sameness in Autism has been designed to replicate and extend a preliminary study of escitalopram treatment of IS related irritability in ASD.

Project IV: Autism-Associated Serotonin Transporter (SERT) Mutations will provide characterization of mutations previously found to be associated with high IS behaviors in subjects with autism. The UIC ACE is an exciting center that brings together experts in a diverse set of disciplines to comprehensively study IS, one of the two cardinal features described by Kanner in 1943.

U18HS016973 (Dr. Robert D. Gibbons PI of Statistical Core)
National Institute of Health (AHRQ - CERT) 09/01/2007 - 08/31/2011

Tools for Optimizing Prescribing, Monitoring and Education

Efforts to maximize the benefits and minimize the risks associated with drugs continue to be impeded by suboptimal prescribing, inadequate monitoring of patients' outcomes, and inadequate prescriber education and support to overcome these limitations. As a result, the US healthcare system suffers from persistent problems of underuse, overuse and misuse of drugs, with unacceptably high rates of preventable errors, adverse effects and suboptimal patient health outcomes. To address these problems, we propose a CERT organized around the following theme: Tools for Optimizing Prescribing, Monitoring and Education (TOP-MED).

The long term objective of our CERT is to improve the safety, efficacy and cost-effectiveness of drug therapy by increasing the appropriateness of prescribing and the quality of monitoring. The short term objective is to develop, redesign, refine, integrate, test, deploy and disseminate tools and training materials in five key areas: drug formularies, drug utilization review, lab-pharmacy linkages, N-of-1 trials and pharmacoeconomics. To achieve these objectives, the TOP-MED CERT will pursue the following specific aims:
1. Revitalize the drug formulary as an evidenced based tool for directing drug therapy decisions.
2. Re-engineer drug usage review (DUR) systems and processes so that data analysis is easier, more timely and more likely to yield valid generalizations.
3. Reduce prescribing errors and enhance recognition of adverse drug effects in high hazard contexts by linking lab and pharmacy information systems and generating clinical alerts when problems are detected.
4. Develop, deploy and evaluate an N-of-1 trial service, integrated into a formulary restriction program, in order to support the goal of individualized therapy without succumbing to the unsafe, unscientific experimentation that is often now the norm.
5. Implement and study the impact of pharmacoeconomic support to enhance formulary decision making, as well as evaluate the cost-effectiveness of other interventions.

Cutting across these five aims will be initiatives to disseminate and deploy these tools in a more rigorous and far-reaching fashion; to provide support to make the tools easier to use; to enhance the buy-in, motivation and enthusiasm of prescribers to use the tools; to link the tools with each other in order to gain greater synergies, and to share the knowledge gained and lessons learned across multiple institutions. Headquartered at the University of Illinois at Chicago (UIC), and leveraging collaborations with the Cook County Bureau of Health Services, Northwestern Memorial Hospital, University of Washington, the VA Center for Medication Safety, Advocate Health Care, the Illinois Hospital Association and the Illinois Department of Public Health, the TOPMED CERT will be able to focus on the needs of multiple AHRQ priority populations.

R01 MH66302-05 (Dr. Robert D. Gibbons PI)
National Institute of Mental Health 09/20/2006 - 08/31/2011

Mental Health Computerized Adaptive Testing - Competitive Renewal

The aim of this investigation is to develop and evaluate computerized adaptive testing programs and algorithms for the assessment of depression. In the original study we demonstrated the feasibility of item response theory (IRT), and computerized adaptive testing (CAT) in the development and administration of a large (626 item) mental health rating scale. Using an item bank of 626 mood and anxiety disorder symptom items, we found that
(a) the majority of the items in the item bank (90%) were discriminating of high and low levels of mood disorders,
(b) the bi-factor IRT model did an excellent job of accounting for the clustering of items within symptom domains,
(c) on average, CAT administration of the test resulted in a 95% reduction in the number of items administered to an individual subject (24 out of 626 items using simulated CAT and 31 items for live CAT testing), and
(d) the correlation between the CAT based impairment rating and the score based on all 626 items was r=0.93.

Based on these very encouraging preliminary results, this competitive renewal proposes to use IRT and CAT to develop a CAT Depression Inventory (CAT-DI). The specific objectives of the renewal are
(1) create a depression item bank by collecting items from a review of approximately 100 existing depression scales and depression items previously identified as a part of the PROMIS network,
(2) calibrate the depression item bank using a variety of IRT models (unidimensional, bi-factor, multidimensional) using a balanced incomplete blocks (BIB) design administered to 800 depressed patients and 200 non-depressed controls,
(3) obtain a new sample of 300 subjects (200 depressed, 100 non-depressed) that take all of the items in the bank, perform a simulated CAT, and optimize the tuning parameters of the CAT,
(4) obtain a new sample 300 subjects (200 depressed, 100 non- depressed) for live CAT testing,
(5) apply the final CAT-DI to a community sample of 700 patients (approximately 200 meeting criteria for major depressive disorder - MOD) to test validity (comparison of impairment estimates in patients with and without MOD), predicting MOD, and establishing normative ranges for patient screening, and
(6) conduct 20 cognitive interviews of patients from a behavioral health clinic who have taken the CAT-DI as a qualitative research approach to beta-testing of the instrument.

R56 MH078580-01 (Dr. Robert D. Gibbons PI)
National Institute of Mental Health 09/30/2006 - 08/31/2008

Antidepressant Treatment and Suicidality: Methodological and Biostatistical Solutions

The purpose of this proposal is to develop, test, and apply new statistical design and analytical methodologies that can be used to identify low base rate drug - adverse event (AE) interactions. These new methods will then be applied to a wide range of existing non-experimental datasets to examine the relationship between SSRIs and suicide ideation, attempts, and completion. We have designed this research project as an integral collaboration between biostatisticians, research psychiatrists and clinicians, economists, and pharmacoepidemiologists, working with large ecological and electronic patient databases covering years where antidepressant use is varying dramatically.

The first set of aims will lead to the development of new biostatistical methods for making inferences from spontaneous reporting and electronic medical record databases. Specifically, we will
1. develop new statistical designs and analyses for identifying drug-AE interactions using both spontaneous (SRS) and active reporting systems (ARS);
2. develop biostatistical methods to address selection and reporting bias in electronic medical record databases;
3. develop statistical methods for large-scale drug-AE screening.

The second set of aims involves the critical application of these methods to existing large scale databases in order to examine the role of antidepressants in suicidality among different populations. The datasets range from the spontaneous reporting system (MedWatch), to electronic medical record databases (e.g., VA, PHARMetrics, Kaiser, and PHARMO, Indian Health Service), to the synthesis of information from randomized clinical trials (RCTs). The work in this proposal will be carried out by a research consortium that will study national and international drug safety issues. The multidisciplinary group includes the areas of statistics (Drs. Gibbons, (PI), Brown (co-Pi), Bhaumik, Duan, Hur, Marcus, SahaRay), psychiatry from adult, child, and genetic perspectives (Brent, Mann, Reynolds, Tsuang), health economics/econometrics (Meltzer, Heckman), pediatrics (Leslie), and pharmacoepidemiology (Valuck). Collaboration with members of the VA (Cunningham, Valenstein), Kaiser Permanente (Clarke, Gullion), PHARMO (Erkens) and the Indian Health Service (Perez) is an integral part of the proposal.

RO1 MH069353-03 (Dr. Dulal Bhaumik PI)
National Institute of Mental Health 05/01/2005 - 03/31/2008

Statistical Testing for Generalized Mixed-effects Models

Over the last decade, mental health services researchers have made widespread use of generalized mixed-effects regression models for analysis of clustered and longitudinal data. Much of the work in this area has involved the development of efficient methods of statistical estimation, based on maximum marginal likelihood, empirical Bayes, and fully Bayesian estimation strategies. Generalization of the original model for continuous and normally distributed data to the case of non-linear mixed-effects regression models for binary, ordinal, nominal, and Poisson distributions are now generally available and enjoy widespread use.

Furthermore, computer software has now been developed and is either freely available over the Internet or commercially available. With the speed of this development and acceptance by the research community, it is therefore somewhat surprising that so little research has been conducted on the issue of hypothesis testing for generalized mixed-effects regression models. Indeed, traditional approaches of large sample tests based on likelihood ratios and Wald-type statistics are all that are generally available. These approaches are limited due to their large sample properties in addition to well-known limitations for testing models with varying numbers of random effects.

In addition to the absence of an arsenal of tools for statistical testing, the literature is also quite limited with respect to statistically rigorous approaches to computing statistical power for clustered and longitudinal designs. For non-linear mixed-models (e.g., binary and ordinal cases), the literature on statistical power is virtually nonexistent, and gross oversimplification of the study design, estimation, and testing procedures must be used to obtain any estimates of the number of measurements needed at each level of nesting that are required to test a hypothesis with a reasonable balance of Type I and II errors.

The primary goal of this proposal is to fill this void by
(1) studying the large and small sample properties of various existing and new tests suitable for generalized linear and non-linear mixed-effects regression models,
(2) developing statistically rigorous approaches to computing statistical power for this class of models that is now so widely used by behavioral, social, and biological scientists in general, and health and mental health services researchers in particular, and
(3) developing a computer program for computing statistical power for linear and non-linear mixed-effects regression models (MIXPWR), and to incorporate these new tests into the existing programs (MIXREG, MIXOR, MIXPREG, MIXNO), which are distributed freely from the MIXREG/MIXOR homepage.

Preliminary results reveal that the new small sample tests that we have derived provide the ability to detect dramatically smaller effects in small samples and increased statistical power over traditional large sample tests even when sample sizes are large. The net result is the ability to use rigorous statistical methods for analysis of longitudinal and clustered data, even in small and difficult to recruit populations such as minorities, homeless, and those at high risk for suicide.

R01 MH67198-01 (Dr. Hua Yun Chen PI)
National Institute of Mental Health 07/01/2004 - 04/30/2008

Multivariate Probit Model for Health Services Research

In this study, we will develop a general mixed-effects multivariate probit regression model for the simultaneous analysis of repeatedly measured multivariate binary data. Correlations between multiple binary measures at a single point in time are modeled as a factor analytic process, and correlation among the repeated measurements over time are modeled as a random-effects process. The net result is that we can now model the effects of design variables (e.g., changes in the health care delivery system) and case mix variables (e.g., age, sex, and race) on multivariate utilization patterns.

Generalizations of the model will include extension to ordinal response data (e.g., no use, mild use, moderate use, high use, or 0 visits, 1 visit, 2 visits, 3 or more visits), mixtures of discrete and continuous responses (e.g., the joint analysis of service utilization and cost), and extension to a multivariate logistic regression model. An integral part of the project will be to both explore and develop alternative approaches to likelihood evaluation (fixed-point and adaptive quadrature, Laplace approximation, and Monte Carlo integration), parameter estimation (Newton Raphson, Fisher scoring, and the EM algorithm), and hypothesis testing. A large-scale simulation study will be conducted to study the statistical properties of the general model and various alternative formulations.

Finally, the model will be applied in the analysis of data collected by Dr. Margarita Alegria at the University of Puerto Rico on the effects of health care reform on longitudinal mental health services utilization. In addition to development of the statistical theory and estimation procedure, we propose to develop a WINDOWS based freeware computer program, MIXMVP, to be distributed from the MIXREG/MIXOR homepage. No general multivariate probit regression software is currently available.

R01 MH65556-03 (Gibbons, Hur, Bhaumik, Hedeker, DSI)
National Institute of Mental Health 07/19/2002 - 05/31/2007

Mixed-Effects ZIP Models for Mental Health Services Research

This project involves the development of a mixed-effects zero-inflated Poisson (ZIP) regression model for the analysis of health services utilization data. The ZIP model provides a method for simultaneously modelling the presence or absence of an event (e.g., service utilization) and the intensity of utilization conditional on its use. Statistically, the model is a mixture of a logistic (use vs no use) and Poisson (intensity of use) regression models. The same or different covariates can be related to the two outcomes. The purpose of the grant is to extend the model to the case of a mixture of fixed and random effects so that it can be used in analysis of clustered and/or longitudinal health services data.

R01 MH66302-03 (Gibbons, Bock, Bhaumik, Kupfer (WPIC), Frank (WPIC), Kessler (Harvard), Weiss (Minnesota), DSI)
MTA Follow-up Study 09/20/2002 - 07/31/2006

Mental Health Computerized Adaptive Testing

Mental health research relies heavily on antiquated systems of measurement. The construction of traditional mental health scales is based largely on subjective judgement, and at best, application of methods from classical test theory to determine a scale's psychometric properties. In this application we borrow strength from major advances in test construction and administration that have been developed in the fields of educational measurement and modern psychometric theory. In particular, we propose to use Item Response Theory (IRT) to calibrate a large item pool of 626 mood disorder items and then adaptively administer them such that a given subject can be evaluated on a small subset of the items to any practical degree of accuracy. The use of the IRT model allows us to evaluate the intensity of the mood disorder for different subjects who have taken potentially different numbers of items selected from the item pool. Using computerized adaptive testing (CAT) we can then adaptively select the most appropriate set of items for each subject based on his/her responses to previous items, beginning from a small screening set of items that characterize low to high levels of impairment. The net result is that large "item banks" can be developed that thoroughly characterize a particular disorder. Although it is not routinely possible for any one subject to be evaluated on all of the items, CAT permits each subject to be evaluated on a small subset of the total item pool, with minimal and controllable loss of information.

Contract (Gibbons, Marcus, Hur, Chen, and Bhaumik)
National Institute of Mental Health
01/01/2000 - 12/31/2007
Statistical Center for analysis of the Long-term MTA Follow-up Study

The NIMH Multimodal Treatment Study for Attention-Deficit/Hyperactivity Disorder (MTA) is the world's largest controlled study of the treament of children with ADHD. NIMH has recently contracted with the original investigators to provide a long-term naturalistic followup study of the children who were enrolled in the original randomized MTA study. Little is known regarding the analysis of data from naturalistic followups of randomized clinical trials. NIMH has contracted with the Center for Health Statistics at UIC to provide statistical research on this topic and to aid the MTA investigators in applying that work in analysis of the followup data. The statistical effort includes research in the areas of generalized mixed-effects regression models, analysis of observational data, propensity score matching, instrumental variables, and growth mixture models.

R01 MH056146-04 (Hedeker, Gibbons, Bhaumik, DSI)
National Institute of Mental Health
07/01/1999 - 07/31/2003
Statistical Models for Nested Service Utilization Data

The goal of this project is to generalize a variety of mixed-effects regression models to the case of three-level data (e.g., Clinics, subjects and measurement occasions). The project involves development of the statistical theory and use of the theory in modifying the MIXREG, MIXOR, MIXPREG, and MIXNO computer programs to the case of three-level designs.

  Last updated April 9, 2012. Copyright © 2003-2012 Center for Health Statistics. All Rights Reserved.

Center for Health Statistics Logo University of Chicago Home Center for Health Statistics Home