2111

OBJECTIVES/SPECIFIC AIMS: Nonalcoholic fatty liver disease (NAFLD) is the most common cause of liver disease in the United States and increases risk for cirrhosis and liver cancer. Identifying modifiable risk factors for NAFLD could allow better targeting of prevention programs. Insulin resistance (IR) plays a significant role in the development and progression of NAFLD. IR is also an important precursor to the development of type 2 diabetes (T2DM). However, the development and duration of IR during young adulthood and its association with NAFLD and T2DM in midlife is unclear. To test whether trajectories of IR using homeostatic model assessment (HOMA-IR) change throughout early adulthood are associated with risk of prevalent NAFLD and T2DM among persons with NAFLD in midlife independent of current or baseline HOMA-IR. METHODS/STUDY POPULATION: Participants from the CARDIA study, a prospective multicenter population-based biracial cohort of adults (baseline age 18–30 years), underwent HOMA-IR measurement (≥8 h fasting and not pregnant) at baseline (1985–1986) and follow-up exam years 7, 10, 15, 20, and 25. At Year 25 (Y25, 2010–2011), liver fat was assessed by noncontrast computed tomography (CT). NAFLD was defined as CT liver attenuation <51 Hounsfield Units after exclusion of other causes of liver fat (alcohol/hepatitis/medications). Latent mixture modeling was used to identify 25-year trajectories in HOMA-IR over time. Multivariable logistic regression models were used to assess associations between HOMA-IR trajectory groups and prevalent NAFLD with adjustment for baseline or Y25 HOMA-IR. RESULTS/ANTICIPATED RESULTS: Among 3060 participants, we identified 3 distinct trajectory groups for HOMA-IR for individuals free from diabetes in middle adulthood: qualitatively low-stable (46.7% of the cohort), moderate-increasing (42.0%), and high-increasing (11.3%) with a NAFLD prevalence at Y25 of: 8.3%, 33.4%, and 63.5%, respectively (p-trend<0.0001). After adjustment for confounders (baseline smoking status, alcohol use, body mass index, physical activity score, systolic blood pressure, antihypertensive medication use, and total/HDL cholesterol ratio) and baseline HOMA-IR, increasing HOMA-IR trajectories were associated with greater NAFLD prevalence compared with the low-stable trajectory group [odds ratio (95% CI): 5.8 (4.3–7.9) and 22.3 (14.2–34.9) for moderate and high, respectively]. These associations were attenuated, but remained significant, even after controlling for current Y25 HOMA-IR [OR=3.6 (2.6–5.0) for moderate and 5.9 (3.4–10.3) for high (referent: low)]. Among participants with NAFLD (n=511), high-increasing HOMA-IR trajectory was associated with greater prevalent [OR=6.5 (1.6–25.7)] and incident [OR=8.7 (2.2–34.4)] T2DM at Y25 independent of confounders and Y25 HOMA-IR (referent: low-stable). DISCUSSION/SIGNIFICANCE OF IMPACT: In this community-based sample of individuals free from diabetes at baseline, an increasing HOMA-IR trajectory through young adulthood was associated with greater NAFLD prevalence in midlife. Knowledge of changes in IR throughout adulthood provides new information on the risk of T2DM among persons with NAFLD in midlife independent of current level of IR. These findings highlight early identification of increasing IR as a potential target for primary prevention of T2DM in the setting of NAFLD.

rate of change in eGFR during defined portions of the follow-up period of the trial, following initiation of the randomized treatment interventions. The second class includes composite endpoints defined by the time from randomization until the occurrence of a designated decline in eGFR or kidney failure. The true clinical endpoint is considered to be the time from randomization until kidney failure, irrespective of the trajectory in eGFR measurements prior to kidney failure. We apply statistical simulation to determine conditions under which alternative endpoints within the 2 classes are (1) valid surrogate endpoints, in the sense of preserving a low probability of rejecting the null hypothesis of no treatment effect on the surrogate endpoint when there is no treatment effect on the clinical endpoints and are also (2) useful surrogate endpoints, in the sense of providing increased statistical power that allows significant reductions in sample size and/or duration of follow-up. Input parameters for the simulations include (a) characteristics of the joint distribution of the longitudinal eGFR measurements and the time to occurrence of renal failure, (b) characteristics of the short-term and long-term effects of the treatment, and (c) design parameters, including the duration of accrual and follow-up and the spacing of eGFR measurements during the follow-up period. We use joint analyses of 19 treatment comparisons across 13 previous clinical trials of CKD patients to guide the selection of input parameters for the simulations. We apply longitudinal mixed effects models for analysis of endpoints based on eGFR slope, and Cox regression for analyses of the composite time-to-event endpoints. RESULTS/ANTICIPATED RESULTS: We have previously shown that surrogate endpoints defined by eGFR declines of 30% or 40% can provide valid and useful alternative endpoints in CKD clinical trials for interventions that do not produce short-term effects on eGFR which differ from the longer-term effects of the interventions. Other factors influencing the validity and utility of these endpoints include the average baseline eGFR, the mean rate of change in eGFR, and the extent to which the size of the treatment effect depends on the patient's underling rate of eGFR decline. We will extend these results by presenting preliminary results describing conditions under which outcomes based on eGFR slope provide valid and useful alternatives to the clinical endpoint of time until occurrence of kidney failure. DISCUSSION/SIGNIFI-CANCE OF IMPACT: The statistical simulation strategy described in this research can be used during the design of clinical trials of chronic kidney disease to assist in the selection of endpoints that maximize savings in sample size and duration of follow-up while retaining a low risk of producing a false positive conclusion in the absence of a true effect of the treatment on the time until kidney failure.

2093
Utilization of an ICD-coded electronic health records (EHR) database to characterize the epidemiology of prosopagnosia Christina Pressl, Caroline Jiang, Joel Correa da Rosa, Maximilian Friedrich, Winrich Freiwald and Jonathan Tobin OBJECTIVES/SPECIFIC AIMS: We aim to examine the epidemiological characteristics of prosopagnosia by querying and analyzing a large deidentified clinical data set from 12 New York City-based hospitals and Federally Qualified Health Centers (FQHCs). The PCORI-funded New York City Clinical Data Research Network (NYC-CDRN) contains~4.5 million deidentified ICDcoded electronic health records (EHRs) with comprehensive longitudinal information on demographics, patient visits, clinical conditions/diagnoses, laboratory and radiology results, medications, and clinical procedures. The NYC-CDRN will be expanded to include other data sources, including insurance claims, social determinant of health, patient reported outcomes, and patient generated data. The central hypothesis was that systematic mining of this database would reveal new epidemiological information about prosopagnosia. We developed a computable phenotype for prosopagnosia, using the International Classification of Diseases version 9 (ICD-9). The computable phenotype consisted of the diagnostic code for the condition under study, prosopagnosia (ICD-9 code 368.16), as well as the codes for known surrogate diagnoses. We expected to identify cases of acquired prosopagnosia, where the condition occurs only after brain damage, due to stroke, trauma, or meningitis for example, and cases of developmental prosopagnosia, where the condition is present from an early age, with no history of brain damage. The goals of this project were to provide new information about the condition's prevalence rate in the New York City area, which could be furthermore translated into wider geographical areas and to yield novel details about its antecedents and comorbid conditions. METHODS/STUDY POPULATION: To determine the presence of the diagnosis of interest, prosopagnosia, and common co-occurring conditions among a New York City-based study population, we investigated a large database in collaboration with the NYC-CDRN. At the time the large database was mined it contained~4 million ICD-9 coded EHRs. We first created a search paradigm; applicable for screening the database that consists of ICD-9 coded EHRs. We generated a list of ICD-9 codes indicative for the patients' difficulties with the perception of faces (368.16), which indicates the presence of the condition as part of the psychophysical visual disturbances complex, and this code identified 871 patients. Furthermore, we collected codes that indicate the presence of conditions that are known to be surrogate diagnoses of prosopagnosia. ICD-9 codes for surrogate diagnoses included for example, 854.* (coding for personal history of traumatic brain injury, n = 1,409), 434.01, 434.11, and 434.91 (coding for cerebral thrombosis, embolus and artery occlusion unspecified with cerebral infarction, n = 19,409), and 191.2 (coding for malignant neoplasm of the temporal lobe, n = 566). In October 2015, coding was changed to the new ICD-10 coding system. No additional patients were revealed from the data set when the cohort was searched for the presence of corresponding ICD-10 codes, as institutions are currently in transition from ICD-9 to ICD-10. Using this search query with the large database, we extracted novel information about the epidemiological and demographical distribution of prosopagnosia and furthermore, gained new knowledge about commonly associated diseases. The fact that it must be presumed that the majority of diagnoses of prosopagnosia have been made on the basis of patients' selfreports and clinicians' judgments represents a limiting factor in this study. We are currently exploring machine-learning strategies to identify potential falsenegative cases among the patients with surrogate diagnoses. RESULTS/ ANTICIPATED RESULTS: Investigations and application of our search query revealed a total number of n = 129,549 patients carrying either the diagnosis code for prosopagnosia or the codes for the known surrogate diagnoses. There were 871 patients who carried the ICD-9 code 368.16, indicating the potential presence of prosopagnosia among other visual disturbances. Remaining patients (n = 128,678) carried codes for known surrogate diagnoses, contained in the search query. Statistical analyses revealed elevated odds ratios for men (OR = 1.55, 95% CI: 1.36, 1.77, p < 0.0001), and for Black/African Americans Versus White individuals (OR = 2.09, 95% CI: 1.74, 2.51, p < 0.0001). DISCUSSION/SIGNIFICANCE OF IMPACT: Currently, the prevalence of prosopagnosia remains unknown. Face blind individuals are struggling to recognize their social contacts by their face only in every day life and are therefore prone to experience reduced quality of life. We searched the large NYC-based clinical database, containing more than 4.5 million deidentified ICDcoded health records, for cases of prosopagnosia to shed light into its prevalence and epidemiological characteristics. We furthermore, mined the database for cases carrying known surrogate diagnoses to explore the magnitude and characteristics of individuals potentially under increased risk. Our efforts address a great healthcare need, as they revealed new epidemiological knowledge of a vulnerable and understudied population. The results of this project reveal new insights into the epidemiological characteristics of prosopagnosia and its surrogate diagnoses, and demonstrate the feasibility of mining large clinical databases to identify rare clinical populations. Our results suggest the need for a more targeted diagnostic assessment of face perception abilities in populations under increased risk.

2111
Insulin resistance patterns over 25-years of adulthood and nonalcoholic fatty liver disease in middle age: The Coronary Artery Risk Development in Young Adults (CARDIA) Study Lisa B. VanWagner, Michael Bancks, Hongyan Ning, Juned Siddique, Cora Lewis, John Jeffrey Carr, Miriam Vos, Elizabeth Speliotes, Norah Terrault, Mary E. Rinella, Norrina B. Allen and Donald Lloyd-Jones Northwestern University, Evanston, IL, USA OBJECTIVES/SPECIFIC AIMS: Nonalcoholic fatty liver disease (NAFLD) is the most common cause of liver disease in the United States and increases risk for cirrhosis and liver cancer. Identifying modifiable risk factors for NAFLD could allow better targeting of prevention programs. Insulin resistance (IR) plays a significant role in the development and progression of NAFLD. IR is also an important precursor to the development of type 2 diabetes (T2DM). However, the development and duration of IR during young adulthood and its association with NAFLD and T2DM in midlife is unclear. To test whether trajectories of IR using homeostatic model assessment (HOMA-IR) change throughout early adulthood are associated with risk of prevalent NAFLD and T2DM among persons with NAFLD in midlife independent of current or baseline HOMA-IR. METHODS/STUDY POPULATION: Participants from the CARDIA study, a prospective multicenter population-based biracial cohort of adults (baseline age 18-30 years), underwent HOMA-IR measurement (≥8 h fasting and not pregnant) at baseline (1985)(1986)) and follow-up exam years 7, 10, 15, 20, and 25. At Year 25 (Y25, 2010-2011), liver fat was assessed by noncontrast computed tomography (CT). NAFLD was defined as CT liver attenuation <51 cambridge.org/jcts Hounsfield Units after exclusion of other causes of liver fat (alcohol/hepatitis/ medications). Latent mixture modeling was used to identify 25-year trajectories in HOMA-IR over time. Multivariable logistic regression models were used to assess associations between HOMA-IR trajectory groups and prevalent NAFLD with adjustment for baseline or Y25 HOMA-IR. RESULTS/ANTICIPATED RESULTS: Among 3060 participants, we identified 3 distinct trajectory groups for HOMA-IR for individuals free from diabetes in middle adulthood: qualitatively low-stable (46.7% of the cohort), moderate-increasing (42.0%), and high-increasing (11.3%) with a NAFLD prevalence at Y25 of: 8.3%, 33.4%, and 63.5%, respectively (p-trend < 0.0001). After adjustment for confounders (baseline smoking status, alcohol use, body mass index, physical activity score, systolic blood pressure, antihypertensive medication use, and total/HDL cholesterol ratio) and baseline HOMA-IR, increasing HOMA-IR trajectories were associated with greater NAFLD prevalence compared with the low-stable trajectory group [odds ratio (95% CI): 5.8 (4.3-7.9) and 22.3 (14.2-34.9) for moderate and high, respectively]. These associations were attenuated, but remained significant, even after controlling for current Y25 HOMA-IR [OR = 3.6 (2.6-5.0) for moderate and 5.9 (3.4-10.3) for high (referent: low)]. Among participants with NAFLD (n = 511), high-increasing HOMA-IR trajectory was associated with greater prevalent [OR = 6.5 (1.6-25.7)] and incident )] T2DM at Y25 independent of confounders and Y25 HOMA-IR (referent: low-stable). DISCUSSION/SIGNIFICANCE OF IMPACT: In this community-based sample of individuals free from diabetes at baseline, an increasing HOMA-IR trajectory through young adulthood was associated with greater NAFLD prevalence in midlife. Knowledge of changes in IR throughout adulthood provides new information on the risk of T2DM among persons with NAFLD in midlife independent of current level of IR. These findings highlight early identification of increasing IR as a potential target for primary prevention of T2DM in the setting of NAFLD.

2115
Post-traumatic stress disorder associated with Hurricane Katrina predicts cardiovascular disease events among elderly adults Zachary Lenane, Erin Peacock and Marie Krousel-Wood Tulane University School of Medicine -LA CaTS, New Orleans, LA, USA OBJECTIVES/SPECIFIC AIMS: Cardiovascular disease (CVD) is the leading cause of death among US adults and its prevalence is increasing, despite efforts to identify, and address risk factors. Post-traumatic stress disorder (PTSD) has been identified as a potential risk factor for CVD, though the results to date have focused on male veterans with combat-related PTSD. To our knowledge, there are no prospective analyses/reports among older community-dwelling adults following Hurricane Katrina. The purpose of this study was to explore the link between PTSD associated with Hurricane Katrina and incident CVD among elderly adults using data from the Cohort Study of Medication Adherence among Older Adults (CoSMO). METHODS/STUDY POPULA-TION: PTSD associated with Katrina and incident CVD events were assessed among 2075 hypertensive participants age ≥65 who were enrolled in a managed care organization in southeastern Louisiana. Baseline surveys were conducted between August 2006 and September 2007. Baseline surveys were conducted between August 2006 and September 2007. PTSD was assessed using the civilian PTSD CheckList (PCL-17) and 2 cut-off points, ≥37 and ≥44, for primary and secondary analyses, respectively. Participants were followed through February 2011 for a composite CVD outcome of MI, stroke, CHF, or CVD death. Multivariable logistic regression was performed with 13 covariates identified in bivariate analysis: age, sex, race, marital status, education, hypertension knowledge, comorbidities, number of antihypertensive medication classes, dissatisfaction with healthcare, reduced medications due to cost, number of visits to healthcare provider in last year, depression, and coping. RESULTS/ANTICIPATED RESULTS: Participants were 59.8% female and 30.4% black, with a mean age of 75 years. The prevalence of PTSD using the primary and secondary cut points was 6.1% and 4.2%, respectively. In total, 240 (11.5%) participants had a CVD event during a median 3.8 year follow-up. After multivariable adjustment, the odds ratios and 95% confidence intervals (CI) for CVD event for the primary and secondary analyses were 1.90 (95% CI: 1.17, 3.09) and 3.74 (95% CI: 2.05, 6.81), respectively. DISCUSSION/SIGNIFI-CANCE OF IMPACT: PTSD was associated with an increased risk of incident CVD events among elderly adults. This finding from a prospective cohort study supports earlier reports suggesting PTSD is an independent risk factor for CVD. To our knowledge, this association has not been previously reported among a cohort of elderly community-dwelling adults. This study included hypertensive, elderly, insured participants living in southeastern Louisiana following Hurricane Katrina and may not be generalizable to all people with PTSD. Strengths of this study include its longitudinal design, the identification of incident CVD, the diversity of the study population with respect to gender, race and CV risk, and reduced confounding due to access to care and insurance status. Future research is needed to confirm this finding in other populations and to assess if efforts to minimize the impact of PTSD following disasters reduce CVD risk and premature CVD events among older adults.

2122
Factors associated with unintended pregnancy in cancer survivors after cancer Participants were interviewed about their reproductive histories. The prediagnosis analysis included all women who completed the interview; the postdiagnosis analysis excluded those who had a hysterectomy, bilateral oophorectomy, or tubal ligation by cancer diagnosis. RESULTS/ANTICIPATED RESULTS: Of the 1282 survivors interviewed, 57.5% reported at least 1 pregnancy before cancer diagnosis; of which, 44.5% were unintended. Of the 1088 survivors included in the postdiagnosis analysis, 36.9% reported a postcancer pregnancy. Among those who had a pregnancy after cancer diagnosis, 38.6% reported at least 1 pregnancy was unintended. Of the 80 breast cancer survivors who had a pregnancy after diagnosis, 52.5% of them were unintended. Predictors of unintended pregnancy in cancer survivors included being younger than 30 years at diagnosis [odds ratio (OR) 2.1; 95% confidence interval (CI) 1.4, 2.9], identifying as Black (OR 1.6, 95% CI 1.1, 2.3, comparison: White), and having resumption of menses after cancer treatment (OR 8.1, 95% CI 2.0, 33.0). Compared with being <4 years from cancer diagnosis, those who were farther from diagnosis at the time of the interview also had increased odds of unintended pregnancy (4-7 years: OR 1.5, 95% CI 0.9, 2.7; 8-10 years: OR 1.3, 95% CI 0.7, 2.4; >10 years: OR 2.7, 95% CI 1.6, 4.7). DISCUSSION/ SIGNIFICANCE OF IMPACT: Despite being at higher risk of infertility, cancer survivors may still be at considerable risk of unintended pregnancy. Women with certain types of cancer that are more likely to be hormone responsive, such as some types of breast cancer, may be hesitant to use hormonal birth control and thus be at higher risk of unintended pregnancy. Counseling for cancer survivors should include a discussion of the risk of unintended pregnancy and contraceptive options.

2156
A confounder assessment of patient frailty in the relationship between antidiabetic medication and heart failure Caroline Presley, Marie Griffin, Jea Young Min, Robert Greevy and Christianne Roumie Vanderbilt University, Nashville, TN, USA OBJECTIVES/SPECIFIC AIMS: This study is part of a parent grant evaluating antidiabetic medications and risk for heart failure in an observational cohort of Veterans with type 2 diabetes (T2DM). Confounding by indication remains a concern in many observational studies of medications because difficult to measure confounders such as frailty may influence prescribing of different medications based on patient characteristics. Frailty is a multidimensional syndrome of loss of reserves (energy, physical ability, cognition, health) that gives rise to vulnerability to adverse outcomes. The objective of this study is to determine if frailty is a potential confounder in Veterans with T2DM, that is, independently associated with exposure to a specific antidiabetic medications and hospitalization for decompensated heart failure. METHODS/STUDY POPULATION: We conducted a cross-sectional study of patients with diabetes who were hospitalized within the Veterans Health Administration (VHA) Tennessee Valley Healthcare System from 2002 to 2012. Inclusion criteria were: age 18 years or older, receive regular VHA care (prescription fill or visit at least once every 180 d), a diagnosis of T2DM. A probability sample of HF and non-HF hospitalizations was collected. HF hospitalizations were selected on the basis of meeting either a primary diagnosis code (ICD-9) and/or disease related group (DRG) code for HF. For each hospitalization using a standardized chart abstraction tool, data was abstracted on: antidiabetic medication(s), patient