Joint inference for competing risks data using multiple endpoints.

Competing risks data are commonly encountered in randomized clinical trials and observational studies. This paper considers the situation where the ending statuses of competing events have different clinical interpretations and/or are of simultaneous interest. In clinical trials, often more than one competing event has meaningful clinical interpretations even though the trial effects of different events could be different or even opposite to each other. In this paper, we develop estimation procedures and inferential properties for the joint use of multiple cumulative incidence functions (CIFs). Additionally, by incorporating longitudinal marker information, we develop estimation and inference procedures for weighted CIFs and related metrics. The proposed methods are applied to a COVID-19 in-patient treatment clinical trial, where the outcomes of COVID-19 hospitalization are either death or discharge from the hospital, two competing events with completely different clinical implications.

Simultaneous hypothesis testing for multiple competing risks in comparative clinical trials.

Competing risks data are commonly encountered in randomized clinical trials or observational studies. Ignoring competing risks in survival analysis leads to biased risk estimates and improper conclusions. Often, one of the competing events is of primary interest and the rest competing events are handled as nuisances. These approaches can be inadequate when multiple competing events have important clinical interpretations and thus of equal interest. For example, in COVID-19 in-patient treatment trials, the outcomes of COVID-19 related hospitalization are either death or discharge from hospital, which have completely different clinical implications and are of equal interest, especially during the pandemic. In this paper we develop nonparametric estimation and simultaneous inferential methods for multiple cumulative incidence functions (CIFs) and corresponding restricted mean times. Based on Monte Carlo simulations and a data analysis of COVID-19 in-patient treatment clinical trial, we demonstrate that the proposed method provides global insights of the treatment effects across multiple endpoints.

Obtaining optimal cutoff values for tree classifiers using multiple biomarkers.

In biomedical practices, multiple biomarkers are often combined using a prespecified classification rule with tree structure for diagnostic decisions. The classification structure and cutoff point at each node of a tree are usually chosen on an ad hoc basis, depending on decision makers' experience. There is a lack of analytical approaches that lead to optimal prediction performance, and that guide the choice of optimal cutoff points in a pre-specified classification tree. In this paper, we propose to search for and estimate the optimal decision rule through an approach of rank correlation maximization. The proposed method is flexible, theoretically sound, and computationally feasible when many biomarkers are available for classification or prediction. Using the proposed approach, for a prespecified tree-structured classification rule, we can guide the choice of optimal cutoff points at tree nodes and estimate optimal prediction performance from multiple biomarkers combined.

Association of Lifestyle Activities with Functional Brain Connectivity and Relationship to Cognitive Decline among Older Adults.

This study examines the relationship of engagement in different lifestyle activities to connectivity in large-scale functional brain networks, and whether network connectivity modifies cognitive decline, independent of brain amyloid levels. Participants (N = 153, mean age = 69 years, including N = 126 with amyloid imaging) were cognitively normal when they completed resting-state functional magnetic resonance imaging, a lifestyle activity questionnaire, and cognitive testing. They were followed with annual cognitive tests up to 5 years (mean = 3.3 years). Linear regressions showed positive relationships between cognitive activity engagement and connectivity within the dorsal attention network, and between physical activity levels and connectivity within the default-mode, limbic, and frontoparietal control networks, and global within-network connectivity. Additionally, higher cognitive and physical activity levels were independently associated with higher network modularity, a measure of functional network specialization. These associations were largely independent of APOE4 genotype, amyloid burden, global brain atrophy, vascular risk, and level of cognitive reserve. Moreover, higher connectivity in the dorsal attention, default-mode, and limbic networks, and greater global connectivity and modularity were associated with reduced cognitive decline, independent of APOE4 genotype and amyloid burden. These findings suggest that changes in functional brain connectivity may be one mechanism by which lifestyle activity engagement reduces cognitive decline.

Development of Severe COVID-19 Adaptive Risk Predictor (SCARP), a Calculator to Predict Severe Disease or Death in Hospitalized Patients With COVID-19.

BACKGROUND: Predicting the clinical trajectory of individual patients hospitalized with coronavirus disease 2019 (COVID-19) is challenging but necessary to inform clinical care. The majority of COVID-19 prognostic tools use only data present upon admission and do not incorporate changes occurring after admission. OBJECTIVE: To develop the Severe COVID-19 Adaptive Risk Predictor (SCARP) (https://rsconnect.biostat.jhsph.edu/covid_trajectory/), a novel tool that can provide dynamic risk predictions for progression from moderate disease to severe illness or death in patients with COVID-19 at any time within the first 14 days of their hospitalization. DESIGN: Retrospective observational cohort study. SETTINGS: Five hospitals in Maryland and Washington, D.C. PATIENTS: Patients who were hospitalized between 5 March and 4 December 2020 with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) confirmed by nucleic acid test and symptomatic disease. MEASUREMENTS: A clinical registry for patients hospitalized with COVID-19 was the primary data source; data included demographic characteristics, admission source, comorbid conditions, time-varying vital signs, laboratory measurements, and clinical severity. Random forest for survival, longitudinal, and multivariate (RF-SLAM) data analysis was applied to predict the 1-day and 7-day risks for progression to severe disease or death for any given day during the first 14 days of hospitalization. RESULTS: Among 3163 patients admitted with moderate COVID-19, 228 (7%) became severely ill or died in the next 24 hours; an additional 355 (11%) became severely ill or died in the next 7 days. The area under the receiver-operating characteristic curve (AUC) for 1-day risk predictions for progression to severe disease or death was 0.89 (95% CI, 0.88 to 0.90) and 0.89 (CI, 0.87 to 0.91) during the first and second weeks of hospitalization, respectively. The AUC for 7-day risk predictions for progression to severe disease or death was 0.83 (CI, 0.83 to 0.84) and 0.87 (CI, 0.86 to 0.89) during the first and second weeks of hospitalization, respectively. LIMITATION: The SCARP tool was developed by using data from a single health system. CONCLUSION: Using the predictive power of RF-SLAM and longitudinal data from more than 3000 patients hospitalized with COVID-19, an interactive tool was developed that rapidly and accurately provides the probability of an individual patient's progression to severe illness or death on the basis of readily available clinical information. PRIMARY FUNDING SOURCE: Hopkins inHealth and COVID-19 Administrative Supplement for the HHS Region 3 Treatment Center from the Office of the Assistant Secretary for Preparedness and Response.

Patient Trajectories Among Persons Hospitalized for COVID-19 : A Cohort Study.

BACKGROUND: Risk factors for progression of coronavirus disease 2019 (COVID-19) to severe disease or death are underexplored in U.S. cohorts. OBJECTIVE: To determine the factors on hospital admission that are predictive of severe disease or death from COVID-19. DESIGN: Retrospective cohort analysis. SETTING: Five hospitals in the Maryland and Washington, DC, area. PATIENTS: 832 consecutive COVID-19 admissions from 4 March to 24 April 2020, with follow-up through 27 June 2020. MEASUREMENTS: Patient trajectories and outcomes, categorized by using the World Health Organization COVID-19 disease severity scale. Primary outcomes were death and a composite of severe disease or death. RESULTS: Median patient age was 64 years (range, 1 to 108 years); 47% were women, 40% were Black, 16% were Latinx, and 21% were nursing home residents. Among all patients, 131 (16%) died and 694 (83%) were discharged (523 [63%] had mild to moderate disease and 171 [20%] had severe disease). Of deaths, 66 (50%) were nursing home residents. Of 787 patients admitted with mild to moderate disease, 302 (38%) progressed to severe disease or death: 181 (60%) by day 2 and 238 (79%) by day 4. Patients had markedly different probabilities of disease progression on the basis of age, nursing home residence, comorbid conditions, obesity, respiratory symptoms, respiratory rate, fever, absolute lymphocyte count, hypoalbuminemia, troponin level, and C-reactive protein level and the interactions among these factors. Using only factors present on admission, a model to predict in-hospital disease progression had an area under the curve of 0.85, 0.79, and 0.79 at days 2, 4, and 7, respectively. LIMITATION: The study was done in a single health care system. CONCLUSION: A combination of demographic and clinical variables is strongly associated with severe COVID-19 disease or death and their early onset. The COVID-19 Inpatient Risk Calculator (CIRC), using factors present on admission, can inform clinical and resource allocation decisions. PRIMARY FUNDING SOURCE: Hopkins inHealth and COVID-19 Administrative Supplement for the HHS Region 3 Treatment Center from the Office of the Assistant Secretary for Preparedness and Response.

Bias correction via outcome reassignment for cross-sectional data with binary disease outcome.

Cross-sectionally sampled data with binary disease outcome are commonly analyzed in observational studies to identify the relationship between covariates and disease outcome. A cross-sectional population is defined as a population of living individuals at the sampling or observational time. It is generally understood that binary disease outcome from cross-sectional data contains less information than longitudinally collected time-to-event data, but there is insufficient understanding as to whether bias can possibly exist in cross-sectional data and how the bias is related to the population risk of interest. Wang and Yang (2021) presented the complexity and bias in cross-sectional data with binary disease outcome with detailed analytical explorations into the data structure. As the distribution of the cross-sectional binary outcome is quite different from the population risk distribution, bias can arise when using cross-sectional data analysis to draw inference for population risk. In this paper we argue that the commonly adopted age-specific risk probability is biased for the estimation of population risk and propose an outcome reassignment approach which reassigns a portion of the observed binary outcome, 0 or 1, to the other disease category. A sign test and a semiparametric pseudo-likelihood method are developed for analyzing cross-sectional data using the OR approach. Simulations and an analysis based on Alzheimer's Disease data are presented to illustrate the proposed methods.

Analyzing wearable device data using marked point processes.

This paper introduces two sets of measures as exploratory tools to study physical activity patterns: active-to-sedentary/sedentary-to-active rate function (ASRF/SARF) and active/sedentary rate function (ARF/SRF). These two sets of measures are complementary to each other and can be effectively used together to understand physical activity patterns. The specific features are illustrated by an analysis of wearable device data from National Health and Nutrition Examination Survey (NHANES). A two-level semiparametric regression model for ARF and the associated activity magnitude is developed under a unified framework using the marked point process formulation. The inactive and active states measured by accelerometers are treated as a 0-1 point process, and the activity magnitude measured at each active state is defined as a marked variable. The commonly encountered missing data problem due to device nonwear is referred to as "window censoring," which is handled by a proper estimation approach that adopts techniques from recurrent event data. Large sample properties of the estimator and comparison between two regression models as measurement frequency increases are studied. Simulation and NHANES data analysis results are presented. The statistical inference and analysis results suggest that ASRF/SARF and ARF/SRF provide useful analytical tools to practitioners for future research on wearable device data.

Complexity and bias in cross-sectional data with binary disease outcome in observational studies.

A cross sectional population is defined as a population of living individuals at the sampling or observational time. Cross-sectionally sampled data with binary disease outcome are commonly analyzed in observational studies for identifying how covariates correlate with disease occurrence. It is generally understood that cross-sectional binary outcome is not as informative as longitudinally collected time-to-event data, but there is insufficient understanding as to whether bias can possibly exist in cross-sectional data and how the bias is related to the population risk of interest. As the progression of a disease typically involves both time and disease status, we consider how the binary disease outcome from the cross-sectional population is connected to birth-illness-death process in the target population. We argue that the distribution of cross-sectional binary outcome is different from the risk distribution from the target population and that bias would typically arise when using cross-sectional data to draw inference for population risk. In general, the cross-sectional risk probability is determined jointly by the population risk probability and the ratio of duration of diseased state to the duration of disease-free state. Through explicit formulas we conclude that bias can almost never be avoided from cross-sectional data. We present age-specific risk probability (ARP) and argue that models based on ARP offers a compromised but still biased approach to understand the population risk. An analysis based on Alzheimer's disease data is presented to illustrate the ARP model and possible critiques for the analysis results.

Semiparametric modelling and estimation of covariate-adjusted dependence between bivariate recurrent events.

A time-dependent measure, termed the rate ratio, was proposed to assess the local dependence between two types of recurrent event processes in one-sample settings. However, the one-sample work does not consider modeling the dependence by covariates such as subject characteristics and treatments received. The focus of this paper is to understand how and in what magnitude the covariates influence the dependence strength for bivariate recurrent events. We propose the covariate-adjusted rate ratio, a measure of covariate-adjusted dependence. We propose a semiparametric regression model for jointly modeling the frequency and dependence of bivariate recurrent events: the first level is a proportional rates model for the marginal rates and the second level is a proportional rate ratio model for the dependence structure. We develop a pseudo-partial likelihood to estimate the parameters in the proportional rate ratio model. We establish the asymptotic properties of the estimators and evaluate the finite sample performance via simulation studies. We illustrate the proposed models and methods using a soft tissue sarcoma study that examines the effects of initial treatments on the marginal frequencies of local/distant sarcoma recurrence and the dependence structure between the two types of cancer recurrence.

ROC-guided survival trees and ensembles.

Tree-based methods are popular nonparametric tools in studying time-to-event outcomes. In this article, we introduce a novel framework for survival trees and ensembles, where the trees partition the dynamic survivor population and can handle time-dependent covariates. Using the idea of randomized tests, we develop generalized time-dependent receiver operating characteristic (ROC) curves for evaluating the performance of survival trees. The tree-building algorithm is guided by decision-theoretic criteria based on ROC, targeting specifically for prediction accuracy. To address the instability issue of a single tree, we propose a novel ensemble procedure based on averaging martingale estimating equations, which is different from existing methods that average the predicted survival or cumulative hazard functions from individual trees. Extensive simulation studies are conducted to examine the performance of the proposed methods. We apply the methods to a study on AIDS for illustration.

In utero exposure to mercury and childhood overweight or obesity: counteracting effect of maternal folate status.

BACKGROUND: Low-dose mercury (Hg) exposure has been associated with cardiovascular diseases, diabetes, and obesity in adults, but it is unknown the metabolic consequence of in utero Hg exposure. This study aimed to investigate the association between in utero Hg exposure and child overweight or obesity (OWO) and to explore if adequate maternal folate can mitigate Hg toxicity. METHODS: This prospective study included 1442 mother-child pairs recruited at birth and followed up to age 15 years. Maternal Hg in red blood cells and plasma folate levels were measured in samples collected 1-3 days after delivery (a proxy for third trimester exposure). Adequate folate was defined as plasma folate ≥ 20.4 nmol/L. Childhood OWO was defined as body mass index ≥ 85% percentile for age and sex. RESULTS: The median (interquartile range) of maternal Hg levels were 2.11 (1.04-3.70) µg/L. Geometric mean (95% CI) of maternal folate levels were 31.1 (30.1-32.1) nmol/L. Maternal Hg levels were positively associated with child OWO from age 2-15 years, independent of maternal pre-pregnancy OWO, diabetes, and other covariates. The relative risk (RR = 1.24, 95% CI 1.05-1.47) of child OWO associated with the highest quartile of Hg exposure was 24% higher than those with the lowest quartile. Maternal pre-pregnancy OWO and/or diabetes additively enhanced Hg toxicity. The highest risk of child OWO was found among children of OWO and diabetic mothers in the top Hg quartile (RR = 2.06; 95% CI 1.56-2.71) compared to their counterparts. Furthermore, adequate maternal folate status mitigated Hg toxicity. Given top quartile Hg exposure, adequate maternal folate was associated with a 34% reduction in child OWO risk (RR = 0.66, 95% CI 0.51-0.85) as compared with insufficient maternal folate. There was a suggestive interaction between maternal Hg and folate levels on child OWO risk (p for interaction = 0.086). CONCLUSIONS: In this US urban, multi-ethnic population, elevated in utero Hg exposure was associated with a higher risk of OWO in childhood, and such risk was enhanced by maternal OWO and/or diabetes and reduced by adequate maternal folate. These findings underscore the need to screen for Hg and to optimize maternal folate status, especially among mothers with OWO and/or diabetes.

Estimations of the joint distribution of failure time and failure type with dependent truncation.

In biomedical studies involving survival data, the observation of failure times is sometimes accompanied by a variable which describes the type of failure event (Kalbeisch and Prentice, 2002). This paper considers two specific challenges which are encountered in the joint analysis of failure time and failure type. First, because the observation of failure times is subject to left truncation, the sampling bias extends to the failure type which is associated with the failure time. An analytical challenge is to deal with such sampling bias. Second, in case that the joint distribution of failure time and failure type is allowed to have a temporal trend, it is of interest to estimate the joint distribution of failure time and failure type nonparametrically. This paper develops statistical approaches to address these two analytical challenges on the basis of prevalent survival data. The proposed approaches are examined through simulation studies and illustrated by using a real data set.

Self-reported Lifestyle Activities in Relation to Longitudinal Cognitive Trajectories.

BACKGROUND: Few studies have examined the relationship between lifestyle activity engagement and cognitive trajectories among individuals who were cognitively normal at baseline. OBJECTIVE: To examine the relationship of current engagement in lifestyle activities to previous cognitive performance among individuals who were cognitively normal at baseline, and whether this relationship differed for individuals who subsequently developed mild cognitive impairment (MCI), or by APOE-4 genotype, age, and level of cognitive reserve. METHODS: Participants (N=189) were primarily middle-aged (M=56.6 y) at baseline and have been prospectively followed with annual assessments (M follow-up=14.3 y). Engagement in physical, cognitive, and social activities was measured by the CHAMPS activity questionnaire. Longitudinal cognitive performance was measured by a global composite score. RESULTS: Among individuals who progressed to MCI (n=27), higher lifestyle activity engagement was associated with less decline in prior cognitive performance. In contrast, among individuals who remained cognitively normal, lifestyle activity engagement was not associated with prior cognitive trajectories. These effects were largely independent of APOE-4 genotype, age, and cognitive reserve. CONCLUSIONS: Greater engagement in lifestyle activities may modify the rate of cognitive decline among those who develop symptoms of MCI, but these findings need to be confirmed in prospective studies.

Predicting progression from normal cognition to mild cognitive impairment for individuals at 5 years.

Recent evidence indicates that measures from cerebrospinal fluid, MRI scans and cognitive testing obtained from cognitively normal individuals can be used to predict likelihood of progression to mild cognitive impairment several years later, for groups of individuals. However, it remains unclear whether these measures are useful for predicting likelihood of progression for an individual. The increasing focus on early intervention in clinical trials for Alzheimer's disease emphasizes the importance of improving the ability to identify which cognitively normal individuals are more likely to progress over time, thus allowing researchers to efficiently screen participants, as well as determine the efficacy of any treatment intervention. The goal of this study was to determine which measures, obtained when individuals were cognitively normal, predict on an individual basis, the onset of clinical symptoms associated with a diagnosis of mild cognitive impairment due to Alzheimer's disease. Cognitively normal participants (n = 224, mean baseline age = 57 years) were evaluated with a range of measures, including: cerebrospinal fluid amyloid-ß and phosphorylated-tau, hippocampal and entorhinal cortex volume, cognitive tests scores and APOE genotype. They were then followed to determine which individuals developed mild cognitive impairment over time (mean follow-up = 11 years). The primary outcome was progression from normal cognition to the onset of clinical symptoms of mild cognitive impairment due to Alzheimer's disease at 5 years post-baseline. Time-dependent receiver operating characteristic analyses examined the sensitivity and specificity of individual measures, and combinations of measures, as predictors of the outcome. Six measures, in combination, were the most parsimonious predictors of transition to mild cognitive impairment 5 years after baseline (area under the curve = 0.85; sensitivity = 0.80, specificity = 0.75). The addition of variables from each domain significantly improved the accuracy of prediction. The incremental accuracy of prediction achieved by adding individual measures or sets of measures successively to one another was also examined, as might be done when enrolling individuals in a clinical trial. The results indicate that biomarkers obtained when individuals are cognitively normal can be used to predict which individuals are likely to develop clinical symptoms at 5 years post-baseline. As a number of the measures included in the study could also be used as subject selection criteria in a clinical trial, the findings also provide information about measures that would be useful for screening in a clinical trial aimed at individuals with preclinical Alzheimer's disease.

Depressive symptoms in relation to clinical symptom onset of mild cognitive impairment.

ABSTRACTObjective:There is increasing evidence of an association between depressive symptoms and mild cognitive impairment (MCI) in cross-sectional studies, but the longitudinal association between depressive symptoms and risk of MCI onset is less clear. The authors investigated whether baseline symptom severity of depression was predictive of time to onset of symptoms of MCI. METHOD: These analyses included 300 participants from the BIOCARD study, a cohort of individuals who were cognitively normal at baseline (mean age = 57.4 years) and followed for up to 20 years (mean follow-up = 2.5 years). Depression symptom severity was measured using the Hamilton Depression Scale (HAM-D). The authors assessed the association between dichotomous and continuous HAM-D and time to onset of MCI within 7 years versus after 7 years from baseline (reflecting the mean time from baseline to onset of clinical symptoms in the cohort) using Cox regression models adjusted for gender, age, and education. RESULTS: At baseline, subjects had a mean HAM-D score of 2.2 (SD = 2.8). Higher baseline HAM-D scores were associated with an increased risk of progression from normal cognition to clinical symptom onset ≤ 7 years from baseline (p = 0.043), but not with progression > 7 years from baseline (p = 0.194). These findings remained significant after adjustment for baseline cognition. CONCLUSIONS: These results suggest that low levels of depressive symptoms may be predictive of clinical symptom onset within approximately 7 years among cognitively normal individuals and may be useful in identifying persons at risk for MCI due to Alzheimer's disease.

A two-stage model for wearable device data.

Recent advances of wearable computing technology have allowed continuous health monitoring in large observational studies and clinical trials. Examples of data collected by wearable devices include minute-by-minute physical activity proxies measured by accelerometers or heart rate. The analysis of data generated by wearable devices has so far been quite limited to crude summaries, for example, the mean activity count over the day. To better utilize the full data and account for the dynamics of activity level in the time domain, we introduce a two-stage regression model for the minute-by-minute physical activity proxy data. The model allows for both time-varying parameters and time-invariant parameters, which helps capture both the transition dynamics between active/inactive periods (Stage 1) and the activity intensity dynamics during active periods (Stage 2). The approach extends methods developed for zero-inflated Poisson data to account for the high-dimensionality and time-dependence of the high density data generated by wearable devices. Methods are motivated by and applied to the Baltimore Longitudinal Study of Aging.

Alternating event processes during lifetimes: population dynamics and statistical inference.

In the literature studying recurrent event data, a large amount of work has been focused on univariate recurrent event processes where the occurrence of each event is treated as a single point in time. There are many applications, however, in which univariate recurrent events are insufficient to characterize the feature of the process because patients experience nontrivial durations associated with each event. This results in an alternating event process where the disease status of a patient alternates between exacerbations and remissions. In this paper, we consider the dynamics of a chronic disease and its associated exacerbation-remission process over two time scales: calendar time and time-since-onset. In particular, over calendar time, we explore population dynamics and the relationship between incidence, prevalence and duration for such alternating event processes. We provide nonparametric estimation techniques for characteristic quantities of the process. In some settings, exacerbation processes are observed from an onset time until death; to account for the relationship between the survival and alternating event processes, nonparametric approaches are developed for estimating exacerbation process over lifetime. By understanding the population dynamics and within-process structure, the paper provide a new and general way to study alternating event processes.

Joint modeling of longitudinal, recurrent events and failure time data for survivor's population.

Recurrent events together with longitudinal measurements are commonly observed in follow-up studies where the observation is terminated by censoring or a primary failure event. In this article, we developed a joint model where the dependence of longitudinal measurements, recurrent event process and time to failure event is modeled through rescaling the time index. The general idea is that the trajectories of all biology processes of subjects in the survivors' population are elongated or shortened by the rate identified from a model for the failure event. To avoid making disputing assumptions on recurrent events or biomarkers after the failure event (such as death), the model is constructed on the basis of survivors' population. The model also possesses a specific feature that, by aligning failure events as time origins, the backward-in-time model of recurrent events and longitudinal measurements shares the same parameter values with the forward time model. The statistical properties, simulation studies and real data examples are conducted. The proposed method can be generalized to analyze left-truncated data.

The BIOCARD Index: A Summary Measure to Predict Onset of Mild Cognitive Impairment.

BACKGROUND: Changes in neuropsychological testing, neuroimaging, and cerebrospinal fluid may precede mild cognitive impairment (MCI). However, these markers are not routinely performed in outpatient clinical visits. OBJECTIVE: To evaluate whether a simple clinical index, consisting of questions given to patients and their informants, could predict the onset of symptoms of MCI among cognitively normal individuals. MATERIALS AND METHODS: Two hundred twenty-two participants in the BIOCARD study received a detailed history, physical examination, and neuropsychological testing annually. An index was calculated by including questions about memory problems, depression, age, education, history of cerebrovascular disease risk factors, and brain injury, family history of dementia, and the Mini-Mental State examination score. Cox regression analyses were used to determine if this index score was related to diagnosis of MCI. RESULTS: The BIOCARD Index score mean for individuals who progressed to MCI was 20.3 (SD=2.9), whereas the score for individuals who remained normal was 24.8 (SD=2.3) (P<0.001) [hazard ratio, SE for subsequent diagnosis of MCI=0.75 (0.67 to 0.84); P<0.001]. CONCLUSIONS: Lower BIOCARD Index score predicted symptoms of MCI several years before the MCI diagnosis. The BIOCARD Index can be easily used in clinics to identify cognitively normal older individuals who are at risk for deterioration.