Research

Profiling/Assessment of Health Care Providers. Profiling analysis aims to evaluate health care providers, such as hospitals, nursing homes, or dialysis facilities, with respect to a patient outcome. Various common patient outcomes of interest include hospital-wide or cause-specific in-hospital mortality and 30-day unplanned hospital readmission. For example, the Center for Medicare and Medicaid Services (CMS) has implement various measures of quality of care at Hospital Compare. Our research focuses on elucidating profiling models (e.g., fixed and random effects models, time-dynamic profiling models) for various patient outcomes, with applications to assessment of US dialysis facilities.

Selected Publications

  • Estes JP, Chen Y, Senturk D, Rhee CM, Kurum E, You AS, Streja E, Kalantar-Zadeh K, Nguyen DV (2020) Profiling dialysis facilities for adverse recurrent events. Statistics in Medicine, 39:9, 1374-1389.
  • Estes JP, Senturk D, Kurum E, Rhee CM, Nguyen DV (2021) Fixed effects high-dimensional profiling models in low information context. International Journal of Statistics in Medical Research, 10, 118-131.
  • Chen Y, Senturk D, Estes JP, Campos LF, Rhee CM, Dalrymple LS, Kalantar-Zadeh K, Nguyen DV (2021) Performance characteristics of profiling methods and the impact of inadequate case-mix adjustment. Communications in Statistics – Simulation and Computation, 50, 1854-1871.
  • Senturk D, Chen Y, Estes JP, Campos LF, Rhee CM, Kalantar-Zadeh K, Nguyen DV (2020) Impact of case-mix measurement error on estimation and inference in profiling of health care providers. Communications in Statistics – Simulation and Computation, 49:8, 2206-2224.
  • Chen Y, Rhee CM, Senturk D, Kurum E, Campos LF, Li Y, Kalantar-Zadeh K, Nguyen DV (2019) Association of U.S. dialysis facility staffing with profiling of hospital-wide 30-day unplanned readmission. Kidney Diseases, 5(3):153-162.
  • Estes JP, Nguyen DV, Chen Y, Dalrymple LS, Rhee CM, Kalantar-Zedeh K, Senturk D (2018) Time-dynamic profiling with application to hospital readmission among patients on dialysis (with discussion). Biometrics, Dec;74(4):1383-1394.
  • Estes JP, Nguyen DV, Chen Y, Dalrymple LS, Rhee CM, Kalantar-Zedeh K, Senturk D (2018) Rejoinder: Time-dynamic profiling with application to hospital readmission among patients on dialysis. Biometrics, Dec;74(4):1383-1394.

Models for Time Varying Effects. Methods to assess the effects of risk factors associated with patient outcome, where the effects are not constant, but are dynamic and vary over  the duration of follow-up (time) or by patient characteristics (e.g., age, disease severity) are important for longitudinal studies.  For example, understanding how cardiovascular (CV) risk evolves during the course of dialysis treatment and how it changes following critical  events like infection-related hospitalization may inform better patient care. These methods will also allow identification of the time periods  of increased outcome (e.g., CV) risk; this knowledge is potentially useful for formulation of CV risk reduction strategies.  Generalized varying coefficient models (GVCMs) depart from traditional simplistic modeling approaches that assume a static or constant effect size for risk factors,  e.g., “patients with baseline diabetes have 20% increased CV outcome risk.” Clearly, although such a simplification is useful in some  studies, it cannot be used to quantify how the effects of individual risk factors vary depending on age at the start of dialysis, for instance. Our group and collaborators have developed novel GVCMs and extensions that allow for flexible models of time varying effects.

Selected Publications

  • Li Y, Nguyen DV, Kurum E, Rhee CM, Chen Y, Kalantar-Zadeh K, Senturk D (2020) A multilevel mixed effects varying coefficient model with multilevel predictors and random effects for modeling hospitalization risk in patients on dialysis. Biometrics, in-press.
  • Li Y, Nguyen DV, Chen Y, Rhee CM, Kalantar-Zedeh, Senturk D (2018) Modeling time-varying effects of multilevel risk factors of hospitalizations in patients on dialysis. Statistics in Medicine, 30;37(30):4707-4720.
  • Estes JP, Nguyen DV, Chen Y, Dalrymple LS, Rhee CM, Kalantar-Zedeh K, Senturk D (2018) Time-dynamic profiling with application to hospital readmission among patients on dialysis (with discussion). Biometrics, Dec;74(4):1383-1394.
  • Estes JP, Nguyen DV, Dalrymple LS, Mu Y, Senturk D (2016) Time-varying effect modeling with longitudinal data truncated by death: Conditional models, interpretations and inference. Statistics in Medicine, 35(11):1834-47.
  • Estes J, Nguyen DV, Dalrymple DS, Mu Y, Senturk D (2014) Cardiovascular event risk dynamics over time in older patients on dialysis: A generalized multiple-index varying coefficient model approach. Biometrics, 70, 754–764.
  • Senturk D, Ghosh S, Nguyen DV (2014) Exploratory time varying lagged regression: Modeling association of cognitive and functional trajectories with expected clinic visits in older adults. Computational Statistics and Data Analysis, 73, 1-15.
  • Senturk D, Dalrymple DS, Mohammed SM, Kaysen GA, Nguyen DV (2013) Modeling time varying effects with generalized and unsynchronized longitudinal data. Statistics in Medicine, 32, 2971-2987.
  • Senturk D, Nguyen DV (2011) Varying coefficient models for sparse noise-contaminated longitudinal data. Statistica Sinica, 21, 1831-1856.

Self-Controlled Case Series Method, Exposure Onset Error, Risk Period Misspecification. The self-controlled case series (SCCS) method is an approach to study the relationship between time-varying exposures and adverse events (AEs), such as AEs following vaccination or other acute exposures.  The SCCS design requires only subjects with one or more events. This aspect of the SCCS design is particularly useful for large longitudinal database applications, such as pharmacovigilance (drug safety monitoring), where the SCCS analysis efficiently utilizes only AEs. Another major advantage of the SCCS method is that it controls for all measured and unmeasured baseline confounders and is self-matched. Thus, the SCCS estimate of the relative incidence of events is not confounded by baseline differences in individual factors, such as socioeconomic status, underlying genetics, and baseline health status or comorbidities, which are difficult to accurately ascertain between exposure groups (e.g., vaccinated and unvaccinated individuals; patients on dialysis who do and do not acquire infections). Our work in this areas currently focuses on  extending the SCCS method to studies where the exposure onset time (e.g., infection time) is not known precisely. We refer to this as “exposure onset  measurement error”. Our second area of research in the SCCS method is to develop unbiased estimates when the risk period in the SCCS is misspecified, which is typically the case in practice since we do not know the true risk time period following exposure.

Selected Publications

  • Campos LF, Senturk D, Chen Y, Nguyen DV (2017) Bias and estimation under misspecification of the risk period in self-controlled case series studies. Stat, 6(1), 373-389 DOI: 10.1002/sta4.166.
  • Mohammed SM, Dalrymple DS, Senturk D, Nguyen DV (2013) Naïve hypothesis testing for case series models with time-varying exposure onset measurement error: Inference for infection-cardiovascular risk in patients on dialysis. Biometrics, 69, 520-529.
  • Mohammed SM, Dalrymple DS, Senturk D, Nguyen DV (2013) Design considerations for case series models with exposure onset measurement error. Statistics in Medicine, 28, 772-786.
    • *** ONLINE TOOL for study design and sample size calculation: Explore tool
  • Mohammed SM, Senturk D, Dalrymple DS, Nguyen DV (2012) Measurement error case series models with application to infection-cardiovascular risk in older patients on dialysis. Journal of the American Statistical Association, 107, 1310-1323.
  • Dalrymple, LS, Mohammed SM, Mu Y, Johansen KL, Chertow GM, Grimes B, Kaysen GA, Nguyen DV (2011) The risk of cardiovascular-related events following infection-related hospitalizations in older patients on dialysis. Clinical Journal of the American Society of Nephrology, 6, 1708-1713.

Development and Validation of Patient Outcome Risk Prediction. Risk prediction tools to inform patients, heath care providers, and researchers on the success of a treatment (e.g., surgery, kidney transplant, extra-corporeal membrane oxygenation [ECMO]) is an important component of the planning and initiation of treatment. Risk prediction tools, rigorously developed and validated, allow for more informed decision making at the individual patient level. Our group and collaborators have developed prediction tools for diverse patient populations, including patients on dialysis and neonates with congenital diaphragmatic hernia (CDH).

Selected Publications

  • Obi Y, Nguyen DV, Zhou H, Soohoo M, Zhang L, Chen Y, Streja E, Sim JJ, Molnar MZ, Rhee CM, Abbott KC, Jacobsen SJ, Kovesdy CP, Kalantar-Zadeh K (2018) Development and validation of prediction scores for early mortality upon transition to dialysis. Mayo Clinic Proceedings, 93(9):1224-1235.
  • Guner YS, Nguyen DV, Zhang L, Chen Y, Harting MT, Rycus P, Barbaro R, Di Nardo M, Brogan TV, Cleary J, Yu PT (2018) Development and validation of pre and on-ECMO mortality-risk models for congenital diaphragmatic hernia. American Society of Artificial Internal Organ Journal, 64(6):785-794.
  • Molnar MZ, Nguyen DV, Chen Y, Ravel V, Streja E, Krishnan M, Kovesdy CP, Mehrotra R, Kalantar-Zadeh K (2017) Predictive score for post-transplantation outcomes. Transplantation, 101(6):1353-1364.

Fragile X Spectrum of Disorders. The fragile X mental retardation 1 (FMR1) gene premutation, with 55-200 CGG repeats, is present in about 1/130-260 females and 1/250-810 males in the general population. When premutation alleles are maternally transmitted, they may expand into the full mutation range, over 200 CGG repeats. Full mutation (>200 CGG repeats) leads to gene silencing and results in the absence or severe deficiency of the FMR1 protein (FMRP) and is the cause of fragile X syndrome (FXS). FXS is the most common inherited cause of intellectual disability and the most common single gene cause of autism. Prevalence estimates are 1 in about 4000-8000, however, the full mutation allele frequency may be as high as 1 in about 2500 in some populations. The phenotype associated with FXS includes both behavioral and cognitive deficits in addition to physical features, such as prominent ears, hyperextensible finger joints and macroorchidism, which begins at puberty. The behavioral phenotype typically includes attention deficit hyperactivity disorder (ADHD), anxiety and intermittent aggression, which can cause significant difficulties for the families. Premutation carriers have elevated levels of FMR1 mRNA levels and two clinical conditions associated with premutation carriers are primary ovarian insufficiency, which can occur in up to 20% of female carriers, and fragile X-associated tremor/ataxia syndrome (FXTAS), a late onset neneurodegenerative disease with a spectrum of clinical characteristics (including intention tremor, ataxia, parkinsonism, peripheral neuropathy, autonomic dysfunction, psychiatric manifestations, and cognitive impairment).

Selected Publications – Treatment Studies, RCTs

  • Hessl D, Schweitzer J, Nguyen DV, McLennan YA, Johnston C, Shickman R, Chen Y (2019) Cognitive training for children and adolescents with fragile X syndrome: A randomized controlled trial of Cogmed. Journal of Neurodevelopmental Disorders, 11(1):4, 1-14.
  • Potter LA, Scholze DA, Biag H, Schneider A, Chen Y, Nguyen DV, Rajaratnam A, Rivera SM, Dwyer PS, Tassone F, Choudhary NS, Salcedo-Arellano JM, Hagerman RJ (2019) A randomized controlled trial of sertraline in young children with autism spectrum disorder. Frontiers in Psychiatry, 10:810.
  • Ligsay A, Dijck AV, Nguyen DV, Lozano R, Chen Y, Bickel E, Hessl DR, Schneider A, Angkustsiri K, Tassone F, Kooy RF, Hagerman RJ (2017) A randomized double-blind, placebo-controlled trial of ganaxolone in children and adolescents with fragile X syndrome. Journal of Neurodevelopmental Disorders, 9(1):26.
  • Seritan AL, Nguyen DV, Mu Y, Tassone F, Bourgeois JA, Schneider A, Cogswell J, Cook K, Leehey M, Grigsby J, Olichney JM, Adams P, Legg W, Zhang L, Hagerman PJ, Hagerman RJ (2014) Memantine for fragile X-associated tremor/ataxia syndrome (FXTAS): a randomized, double-blind, placebo-control trial. Journal of Clinical Psychiatry, 75, 264-271.
  • Leigh MJ, Nguyen DV, Mu Y, Winarni TI, Hessl DR, Rivera SM, Chechi T, Polussa J, Tassone F, Hagerman RJ (2013) A randomized double blind, placebo controlled trial of minocycline in children and adolescents with fragile X syndrome. Journal of Developmental and Behavioral Pediatrics, 34, 147-155.
  • Berry-Kravis E, Hessl D, Rathmell B, Zarevics P, Cherubini M, Walton-Bowen K, Mu Y, Nguyen DV, Gonzales-Heydrich I, Wang P, Carpenter R, Bear M, Hagerman RJ (2012) Effects of STX209 (arbaclofen) on neurobehavioral function in children and adults with fragile X syndrome: A randomized, controlled, Phase 2 trial. Science Translational Medicine, 4, 152:127.
  • Berry-Kravis E, Hessl D, Coffey S, Hervey C, Schneider A, Yuhas J, Hutchinson J, Snape M, Tranfaglia M, Nguyen DV, Hagerman RJ (2009) A pilot open-label single-dose trial of fenobam in adults with fragile X syndrome. Journal of Medical Genetics, 46, 266-271.

Selected Publications – Clinical Involvements in Premutation Carriers

  • Hamlin A, Lui Y, Nguyen DV, Tassone F, Zhang L, Hagerman RJ (2011) Sleep apnea in premutation carriers with and without FXTAS. American Journal of Medical Genetics: Neuropsychiatric Genetics, 156, 923-928.
  • Bourgeois JA, Seritan A, Casillas EM, Hessl D, Schneider A, Yang Y, Kaur I, Cogswell JB, Nguyen DV, Hagerman RJ (2011) Lifetime prevalence of mood and anxiety disorders in fragile X premutation carriers. Journal of Clinical Psychiatry, 72, 175-182.
  • Coffey SM, Cook K, Tartaglia N, Tassone F, Nguyen DV, Pan R, Bronksy HE, Yuhas J, Borodyanskaya M, Grigsby J, Doerflinger M, Hagerman PJ, Hagerman RJ (2008) Expanded clinical phenotype of women with the FMR1 premutation. American Journal of Medical Genetics, 146A, 1009-1016.

Covariate Adjusted Regression and Correlation. Covariate adjusted regression (CAR) is a semiparametric adjustment methodology for modeling general (unknown) confounding or multiplicative distortion in both the response/outcome variable and covariates. These are new measurement error models that we have proposed in both cross-sectional and longitudinal data settings. These models allow for general (unknown) “confounding” effects (in regression and correlation modeling), which may be multiplicative, additive, nonlinear or null effects. No a priori assumptions are made on the type or form of confounding, therefore providing a flexible modeling framework. An illustration of our proposed method is the following application to understand fragile X mental retardation (FMR1) gene expression in female premutation carriers. Motivated by molecular data on female premutation carriers of the FMR1 gene, we developed a new method of covariate adjusted correlation analysis to examine the association of messenger RNA (mRNA) and number of CGG repeat expansion in the promoter region of the FMR1 gene. The association between the molecular variables in female carriers needs to adjust for activation ratio, a measure which accounts for the protective effects of one normal X chromosome in female fragile X premutation carriers.

Selected Publications

  • Senturk D, Nguyen DV, Tassone F, Hagerman RJ, Carroll RJ, Hagerman PJ (2009) Covariate adjusted correlation analysis with application to FMR1 premutation female carrier data. Biometrics, 65, 781-792.
  • Nguyen DV, Senturk D (2009) Covariate-adjusted regression for longitudinal data incorporating correlation between repeated measurements. Australian and New Zealand Journal of Statistics, 51, 319-333
  • Nguyen DV, Senturk D, Carroll RJ (2008) Covariate-adjusted linear mixed effects models with an application to longitudinal data. Journal of Nonparametric Statistics, 20, 459-481.
  • Senturk D, Nguyen DV (2006) Estimation in covariate-adjusted regression. Computational Statistics and Data Analysis, 50, 3294-3310.

High-Dimensional Data, Genomics, Classification, Partial Least Squares. We have developed several innovative approaches to the analysis of high-dimensional data (e.g., genomics data). At the start of the “-omics” era in late 1990’s, we were among the first to develop dimension reduction methods for classification and prediction where the sample size n << p, where p is the number of predictor variables (e.g., genomics expression data). The phrase “small n and large p,” now in common usage, characterizes this class of data analysis problems. In a sequence of papers, we proposed and studied partial least squares (PLS) dimension reduction for binary and multi-class classification, as well as the analysis of (censored) survival data, based on genomics expression data as predictors. Applications of PLS dimension reduction to cancer classification based on genomics profiles were provided. Since our introduction of PLS as a dimension reduction method, it has been applied to other types of high-dimensional data (e.g., proteomics, metabolomics etc.) with applications beyond cancer classification. In the context of high-dimensional data, we have also investigated the performance of the false discovery rate as an error control, data quality issues, and approaches to missing data among others.

Selected Publications

  • Nguyen DV (2005) Partial least squares dimension reduction for microarray gene expression data with a censored response. Mathematical Biosciences, 193, 119-137.
  • Nguyen DV, Wang N, Carroll RJ (2004) Evaluation of missing value estimation for microarray data. Journal of Data Science, 2, 347-370.
  • Nguyen DV (2004) On estimating the proportion of true null hypotheses for false discovery rate controlling procedures in exploratory DNA microarray studies. Computational Statistics and Data Analysis, 47, 611-637.
  • Nguyen DV, Arpat AB, Wang N, Carroll RJ (2002) DNA microarray experiments: biological and technological aspects. Biometrics, 58, 701-717.
  • Nguyen DV, Rocke DM (2002) Partial least squares proportional hazard regression for application to DNA microarray survival data. Bioinformatics, 18, 1625-1632.
  • Nguyen DV, Rocke DM (2002) Multi-class cancer classification via partial least squares using gene expression profiles. Bioinformatics, 18, 1216-1226.
  • Nguyen DV, Rocke DM (2002) Tumor classification by partial least squares using microarray gene expression data. Bioinformatics, 18, 39-50.

Regression Models for Joint Modeling of Disease Onset and Recurrence, Zero-Inflated Count Data. Cardiovascular disease remains one of the leading causes of hospitalization and death in the population of patients on dialysis. Our aim here is to develop methods to jointly model the relationship/association between covariates and (a) the probability of cardiovascular events (onset), a binary process and (b) the rate of events (recurrence) once the realization is positive – when the ‘hurdle’ is crossed – using a zero-truncated Poisson distribution. When the observation period or follow-up time, from the start of dialysis, varies among individuals the estimated probability of positive cardiovascular events during the study period will be biased. We develop strategies to eliminate this bias. In the context of zero-inflated count data, we are also developing functional linear model to be able to model functional predictors X(t) measured over time t, for instance.

Selected Publications

  • Senturk D, Dalrymple DS, Mu Y, Nguyen DV (2014) Weighted hurdle regression method for joint modeling of cardiovascular events likelihood and rate in the U.S. dialysis population. Statistics in Medicine, 33(25):4387-4401.
  • Senturk D, Dalrymple LS, Nguyen DV (2014) Functional linear models for zero-inflated count data with application to modeling hospitalizations in patients on dialysis. Statistics in Medicine, 33(27):4825-4840.

Lung Cancer: Invasive/Non-Invasive Staging, Disparities, Survival. Our group have collaborated on using large databases, including the Surveillance, Epidemiology and End Results (SEER) and National Inpatient Sample (NIS) databases, to examine outcomes in patients with lung cancer as well as disparities disease staging. For example, we examined stage migration in oncology and apparent improvement in survvival from 1994-2004. Positron emission tomography (PET) imaging emerged as a standard diagnostic imaging modality for the staging of lung cancer, offering enhanced sensitivity in detecting otherwise occult tumor spread and PET imaging was approval for reimbursement for lung cancer staging by Medicare in 1998. We explored how stage migration, particularly between stage III and IV due to more sensitive stage via PET, resulted in apparent improved survival for both patients with stage III and IV. This stage migration (“Will Rogers phenomenon”), occurs if moving an element from one set to another raises the average values of both sets. Using this same cohort, we also examined racial disparities the on the use of invasive and noninvasive staging in patients with non-small cell lung cancer.

Selected Publications

  • Launer H, Nguyen DV, Cooke DT (2013) National Perioperative Outcomes of Pulmonary Lobectomy for Cancer in the Obese Patient: A Propensity core Matched Analysis. Journal of Thoracic and Cardiovascular Surgery, 145, 1312-1318.
  • Maximus S, Nguyen DV, Mu Y, Calhoun RF, Cooke DT (2012) Size of Stage IIIA Primary Lung Cancers and Survival: a surveillance, epidemiology and end results database analysis. The American Surgeon, 78, 1232-1237.
  • Cooke DT, Nguyen DV, Yang Y, Chen SL, Yu C, Calhoun RF (2010) Survival comparison of adenosquamous, squamous cell, and adenocarcinoma of the lung after lobectomy. Annals of Thoracic Surgery, 90, 943-948.
  • Suga MJ, Nguyen DV, Mohammed SM, Brown M, Calhoun R, Yoneda K, Gandara, D.R., Lara, P.N. (2010) Racial disparities on the use of invasive and non-invasive staging in patients with non-small cell lung cancer. Journal of Thoracic Oncology, 5, 1772-1778.
  • Chee K G, Nguyen DV, Brown M, Gandara DR, Wun T, Lara PN (2008) Positron emission tomography and improved survival in lung cancer: The Will Rogers phenomenon revisited. Archives of Internal Medicine, 168, 1541-1549.