An Adaptive Behavioral Intervention for Weight Loss Management: A Randomized Clinical Trial.

Importance: Lifestyle interventions for weight loss are difficult to implement in clinical practice. Self-managed mobile health implementations without or with added support after unsuccessful weight loss attempts could offer effective population-level obesity management. Objective: To test whether a wireless feedback system (WFS) yields noninferior weight loss vs WFS plus telephone coaching and whether participants who do not respond to initial treatment achieve greater weight loss with more vs less vigorous step-up interventions. Design, Setting, and Participants: In this noninferiority randomized trial, 400 adults aged 18 to 60 years with a body mass index of 27 to 45 were randomized in a 1:1 ratio to undergo 3 months of treatment initially with WFS or WFS plus coaching at a US academic medical center between June 2017 and March 2021. Participants attaining suboptimal weight loss were rerandomized to undergo modest or vigorous step-up intervention. Interventions: The WFS included a Wi-Fi activity tracker and scale transmitting data to a smartphone app to provide daily feedback on progress in lifestyle change and weight loss, and WFS plus coaching added 12 weekly 10- to 15-minute supportive coaching calls delivered by bachelor's degree-level health promotionists viewing participants' self-monitoring data on a dashboard; step-up interventions included supportive messaging via mobile device screen notifications (app-based screen alerts) without or with coaching or powdered meal replacement. Participants and staff were unblinded and outcome assessors were blinded to treatment randomization. Main Outcomes and Measures: The primary outcome was the between-group difference in 6-month weight change, with the noninferiority margin defined as a difference in weight change of -2.5 kg; secondary outcomes included between-group differences for all participants in weight change at 3 and 12 months and between-group 6-month weight change difference among nonresponders exposed to modest vs vigorous step-up interventions. Results: Among 400 participants (mean [SD] age, 40.5 [11.2] years; 305 [76.3%] women; 81 participants were Black and 266 were White; mean [SD] body mass index, 34.4 [4.3]) randomized to undergo WFS (n = 199) vs WFS plus coaching (n = 201), outcome data were available for 342 participants (85.5%) at 6 months. Six-month weight loss was -2.8 kg (95% CI, -3.5 to -2.0) for the WFS group and -4.8 kg (95% CI, -5.5 to -4.1) for participants in the WFS plus coaching group (difference in weight change, -2.0 kg [90% CI, -2.9 to -1.1]; P < .001); the 90% CI included the noninferiority margin of -2.5 kg. Weight change differences were comparable at 3 and 12 months and, among nonresponders, at 6 months, with no difference by step-up therapy. Conclusions and Relevance: A wireless feedback system (Wi-Fi activity tracker and scale with smartphone app to provide daily feedback) was not noninferior to the same system with added coaching. Continued efforts are needed to identify strategies for weight loss management and to accurately select interventions for different individuals to achieve weight loss goals. Trial Registration: ClinicalTrials.gov Identifier: NCT02997943.

Joint modeling the frequency and duration of accelerometer-measured physical activity from a lifestyle intervention trial.

Physical activity (PA) guidelines recommend that PA be accumulated in bouts of 10 minutes or more in duration. Recently, researchers have sought to better understand how participants in PA interventions increase their activity. Participants can increase their daily PA by increasing the number of PA bouts per day while keeping the duration of the bouts constant; they can keep the number of bouts constant but increase the duration of each bout; or participants can increase both the number of bouts and their duration. We propose a novel joint modeling framework for modeling PA bouts and their duration over time. Our joint model is comprised of two sub-models: a mixed-effects Poisson hurdle sub-model for the number of bouts per day and a mixed-effects location scale gamma regression sub-model to characterize the duration of the bouts and their variance. The model allows us to estimate how daily PA bouts and their duration vary together over the course of an intervention and by treatment condition and is specifically designed to capture the unique distributional features of bouted PA as measured by accelerometer: frequent measurements, zero-inflated bouts, and skewed bout durations. We apply our methods to the Make Better Choices study, a longitudinal lifestyle intervention trial to increase PA. We perform a simulation study to evaluate how well our model is able to estimate relationships between outcomes.

Effect of the Fit2Thrive Intervention on Patient-reported Outcomes in Breast Cancer Survivors: A Randomized Full Factorial Trial.

BACKGROUND: Moderate to vigorous physical activity (MVPA) interventions improve patient-reported outcomes (PROs) of physical and psychological health among breast cancer survivors (BCS); however, the effects of specific intervention components on PROs are unknown. PURPOSE: To use the Multiphase Optimization Strategy (MOST) to examine overall effects of the Fit2Thrive MVPA promotion intervention on PROs in BCS and explore whether there are intervention component-specific effects on PROs. METHODS: Physically inactive BCS [n = 269; Mage = 52.5 (SD = 9.9)] received a core intervention (Fitbit + Fit2Thrive smartphone app) and were randomly assigned to one of 32 conditions in a full factorial experiment of five components ("on" vs. "off"): (i) support calls, (ii) deluxe app, (iii) text messages, (iv) online gym, and (v) buddy. Patient-Reported Outcomes Measurement Information System (PROMIS) questionnaires assessed anxiety, depression, fatigue, physical functioning, sleep disturbance and sleep-related impairment at baseline, post-intervention (12-week), and 24-week follow-up. Main effects for all components at each time point were examined using an intention to treat mixed-effects model. RESULTS: All PROMIS measures except sleep disturbance significantly improved (p's < .008 for all) from baseline to 12-weeks. Effects were maintained at 24-weeks. The "on" level of each component did not result in significantly greater improvements on any PROMIS measure compared to the "off" level. CONCLUSIONS: Participation in Fit2Thrive was associated with improved PROs in BCS, but improvements did not differ for "on" vs. "off" levels for any component tested. The low-resource Fit2Thrive core intervention is a potential strategy to improve PROs among BCS. Future studies should test the core in an RCT and examine various intervention component effects in BCS with clinically elevated PROs.

Optimization of a technology-supported physical activity promotion intervention for breast cancer survivors: Results from Fit2Thrive.

BACKGROUND: The benefits of moderate to vigorous physical activity (MVPA) for breast cancer survivors are well established. However, most are insufficiently active. Fit2Thrive used the Multiphase Optimization Strategy methodology to determine the effect of 5 intervention components on MVPA in this population. METHODS: Two hundred sixty-nine participants (mean age, 52.5 years; SD, 9.9 years) received a core intervention (the Fit2Thrive self-monitoring app and Fitbit) and were randomly assigned to 5 intervention components set to on/off in a full factorial experiment: support calls, deluxe app, buddy, online gym, and text messages. The intervention was delivered over 12 weeks with a 12-week follow-up. MVPA was measured via accelerometry at the baseline (T1), at 12 weeks (T2), and at 24 weeks (T3). The main effects and interaction effects at each time point were examined for all components. RESULTS: Trial retention was high: 91.8% had valid accelerometer data at T2 or T3. Across all conditions, there were significant increases in MVPA (+53.6 min/wk; P < .001) and in the proportion of survivors meeting MVPA guidelines (+22.3%; P < .001) at T2 that were maintained but attenuated at T3 (MVPA, +24.6 min/wk; P < .001; meeting guidelines, +12.6%; P < .001). No individual components significantly improved MVPA, although increases were greater for the on level versus the off level for support calls, buddy, and text messages at T2 and T3. CONCLUSIONS: The Fit2Thrive core intervention (the self-monitoring app and Fitbit) is promising for increasing MVPA in breast cancer survivors, but the components provided no additional increases in MVPA. Future research should evaluate the core intervention in a randomized trial and determine what components optimize MVPA behaviors in breast cancer survivors.

Social cognitive variables and physical activity during chemotherapy for breast cancer: An intensive longitudinal examination.

OBJECTIVE: Although physical activity is associated with better health outcomes in breast cancer survivors (BCS), activity often declines during cancer treatment. Social cognitive theory (SCT) constructs have been associated with physical activity in post-treatment BCS, but little is known about the relation between these constructs and physical activity during chemotherapy. METHODS: BCS (n = 67; Mage  = 48.6 [SD = 10.3]) undergoing chemotherapy wore accelerometers and completed prompts in the morning and at night assessing same-day and next-day exercise self-efficacy, physical and psychological outcome expectations, and goal-setting for 10 consecutive days (3 days pre-, day of, and 6 days post-chemotherapy dose) at three time points (beginning, middle, and end of chemotherapy). Separate mixed models assessed between- and within-person associations of each of the SCT constructs associations with same- and next-day moderate to vigorous physical activity (MVPA) and light physical activity (LPA), independently. RESULTS: Within-person differences in all SCT variables were statistically significantly related to same-day MVPA (p's < 0.001) and LPA (p's < 0.001). Every one-point increase in SCT construct related to an increase in MVPA ranging from (a) 3.70 (self-efficacy) to 8.02 (physical outcome expectations) minute increase in MVPA and (b) 12.72 (self-efficacy) to 20.38 (physical outcome expectations) increase in LPA that day. No same-day between-person effects nor any next-day effects were significant. CONCLUSION: MVPA and LPA were related to same-day within-person differences in SCT variables. Interventions targeted at increasing or mitigating chemotherapy-related declines in daily within-person changes in SCT constructs could help to increase physical activity among BCS during chemotherapy.

Using ecological momentary assessment to understand associations between daily physical activity and symptoms in breast cancer patients undergoing chemotherapy.

PURPOSE: Understanding real-time relationships between physical activity (PA) and symptoms during chemotherapy (CT) could have important implications for intervention. This study used ecological momentary assessment to examine the relationship between objective PA and symptoms during CT. METHODS: Breast cancers patients (n = 67; Mage = 48.6 (SD = 10.3)) participated in data collection at three time points during CT: beginning, middle, and end. At each time point, participants answered four prompts assessing symptoms and wore an accelerometer for 10 days (3 days pre-CT, day of CT, and 6 days post-CT). Multilevel linear regression models examined the between- and within-person associations between moderate to vigorous (MVPA) and light-intensity physical activity (LPA) and same and next-day symptom ratings controlling for covariates. RESULTS: On days when individuals engaged in more LPA or MVPA, separately, they reported improved affect, anxiety, fatigue, physical functioning (walking and activities of daily living), pain, and cognition that day (p < 0.001 for all). Findings were consistent for next-day symptom ratings with the exception that only previous day LPA was related to next-day fatigue and neither LPA nor MVPA were related to next-day cognition (p < 0.001 for all). No between-person effects were found. CONCLUSIONS: Within person higher than usual PA on a given day, regardless of intensity, is associated with improved symptoms ratings on the current and next day. IMPLICATIONS FOR CANCER SURVIVORS: Encouraging breast cancer patients undergoing CT to engage in daily PA could help manage CT-associated symptoms.

Development and Validation of a Large Synthetic Cohort for the Study of Cardiovascular Health Across the Life Span.

We developed and validated a synthetic cohort approach to examine numbers of cardiovascular risk factors (CRFs) and adverse clinical events, including incident cardiovascular disease and all-cause mortality, across the life span from ages 20 years to 90 years. The current analysis included 40,875 participants from 7 large, population-based longitudinal epidemiologic studies (1948-2016). On the basis of a joint multilevel imputation model, we multiply imputed each participant's life-span numbers of CRFs and events using available records. To validate the imputed values, we partially removed the observed data and then compared the imputed and observed values. The complete life-span synthetic data set reflected the original observed data trends well. In our validation sample, the distributions of imputed CRFs and events were close to the observed distributions but with less variability. Bland-Altman plots indicated that there was a slightly negative trend in general, and the agreement bias was relatively small for the continuous CRFs. The hypothetical linear regression model suggested that the relationships between the CRFs and events were preserved in the imputed data set. This approach generated valid estimates of CRFs and events across the life span for African-American and White participants. The synthetic cohort may be sufficiently accurate to be useful in assessing the origins and timing of accumulating cardiovascular risk that can inform efforts to avoid cardiovascular disease development.

Generalizing randomized trial findings to a target population using complex survey population data.

Randomized trials are considered the gold standard for estimating causal effects. Trial findings are often used to inform policy and programming efforts, yet their results may not generalize well to a relevant target population due to potential differences in effect moderators between the trial and population. Statistical methods have been developed to improve generalizability by combining trials and population data, and weighting the trial to resemble the population on baseline covariates. Large-scale surveys in fields such as health and education with complex survey designs are a logical source for population data; however, there is currently no best practice for incorporating survey weights when generalizing trial findings to a complex survey. We propose and investigate ways to incorporate survey weights in this context. We examine the performance of our proposed estimator through simulations in comparison to estimators that ignore the complex survey design. We then apply the methods to generalize findings from two trials-a lifestyle intervention for blood pressure reduction and a web-based intervention to treat substance use disorders-to their respective target populations using population data from complex surveys. The work highlights the importance in properly accounting for the complex survey design when generalizing trial findings to a population represented by a complex survey sample.

Effects of differential measurement error in self-reported diet in longitudinal lifestyle intervention studies.

BACKGROUND: Lifestyle intervention studies often use self-reported measures of diet as an outcome variable to measure changes in dietary intake. The presence of measurement error in self-reported diet due to participant failure to accurately report their diet is well known. Less familiar to researchers is differential measurement error, where the nature of measurement error differs by treatment group and/or time. Differential measurement error is often present in intervention studies and can result in biased estimates of the treatment effect and reduced power to detect treatment effects. Investigators need to be aware of the impact of differential measurement error when designing intervention studies that use self-reported measures. METHODS: We use simulation to assess the consequences of differential measurement error on the ability to estimate treatment effects in a two-arm randomized trial with two time points. We simulate data under a variety of scenarios, focusing on how different factors affect power to detect a treatment effect, bias of the treatment effect, and coverage of the 95% confidence interval of the treatment effect. Simulations use realistic scenarios based on data from the Trials of Hypertension Prevention Study. Simulated sample sizes ranged from 110-380 per group. RESULTS: Realistic differential measurement error seen in lifestyle intervention studies can require an increased sample size to achieve 80% power to detect a treatment effect and may result in a biased estimate of the treatment effect. CONCLUSIONS: Investigators designing intervention studies that use self-reported measures should take differential measurement error into account by increasing their sample size, incorporating an internal validation study, and/or identifying statistical methods to correct for differential measurement error.

Bidirectional 10-year associations of accelerometer-measured sedentary behavior and activity categories with weight among middle-aged adults.

BACKGROUND: Although higher sedentary behavior (SB) with low light intensity (LPA) and moderate-to-vigorous intensity physical activity (MVPA) are thought to increase risk for obesity, other data suggest excess weight may precede these behaviors in the causal pathway. We aimed to investigate 10-year bidirectional associations between SB and activity with weight. METHODS: Analysis included 886 CARDIA participants (aged 38-50 years, 62% female, 38% black) with weight and accelerometry ( ≥ 4 days with ≥ 10 h/day) collected in 2005-6 (ActiGraph 7164) and 2015-6 (ActiGraph wGT3X-BT). Accelerometer data were calibrated, harmonized, and expressed as counts per minute (cpm) and time-dependent intensity categories (min/day of SB, LPA, and MVPA; SB and MVPA were also separated into long-bout and short-bout categories). Linear regression models were constructed to estimate adjusted associations of baseline activity with 10-year change in weight and vice versa. When activity categories were the independent variables, standardized regression coefficients (ßstd.) estimated associations of replacing SB with a one SD increase in other categories, adjusted for accelerometer wear time. RESULTS: Over 10-years, weight increased by a mean 2.55 ± 8.05 kg and mean total activity decreased by 50 ± 153 cpm. In adjusted models, one SD higher baseline mean total activity (ßstd. = -1.4 kg, p < 0.001), LPA (ßstd. = -0.80 kg, p = 0.013), total MVPA (ßstd. = -1.07 kg, p = 0.001), and long-bout MVPA (ßstd. = -1.20 kg, p < 0.001) were associated with attenuated 10-year weight gain. Conversely, a one SD higher baseline weight was associated with unfavorable 10-year changes in daily activity profile including increases in SB (ßstd. = 12.0 min, p < 0.001) and decreases in mean total activity (ßstd. = 14.9 cpm, p = 0.004), LPA (ßstd. = 8.9, p = 0.002), and MVPA (ßstd. = 3.5 min, p = 0.001). Associations varied by race and gender. CONCLUSIONS: Higher SB with lower activity and body weight were bidirectionally related. Interventions that work simultaneously to replace SB with LPA and long-bout MVPA while also using other methods to address excess weight may be optimal.

Propensity Score-Based Estimators With Multiple Error-Prone Covariates.

Propensity score methods are an important tool to help reduce confounding in nonexperimental studies. Most propensity score methods assume that covariates are measured without error. However, covariates are often measured with error, which leads to biased causal effect estimates if the true underlying covariates are the actual confounders. Although some groups have investigated the impact of a single mismeasured covariate on estimating a causal effect and proposed methods for handling the measurement error, fewer have investigated the case where multiple covariates are mismeasured, and we found none that discussed correlated measurement errors. In this study, we examined the consequences of multiple error-prone covariates when estimating causal effects using propensity score-based estimators via extensive simulation studies and real data analyses. We found that causal effect estimates are less biased when the propensity score model includes mismeasured covariates whose true underlying values are strongly correlated with each other. However, when the measurement errors are correlated with each other, additional bias is introduced. In addition, it is beneficial to include correctly measured auxiliary variables that are correlated with confounders whose true underlying values are mismeasured in the propensity score model.

Sedentary behavior after breast cancer: motivational, demographic, disease, and health status correlates of sitting time in breast cancer survivors.

PURPOSE: Sedentary behavior is associated with poor health outcomes including obesity, lower quality of life, and mortality in breast cancer survivors. This study sought to identify motivational, demographic, and disease characteristics of breast cancer survivors who engage in greater amounts of sedentary behavior. METHODS: Multivariate linear regression models estimated associations between demographic, disease, and health characteristics with reported sitting in breast cancer survivors (n = 279; Mage = 60.7 (± 9.7) years). Regression models estimated associations between motivational factors and reported sitting adjusted for demographic and disease and health covariates. RESULTS: Working at least part-time and marital status were associated various sitting domains including weekday and non-leisure sitting. Higher BMI was associated with more average daily, weekend, and weekday sitting. High income was additionally associated with less non-leisure sitting. The belief that sedentary behavior is bad for health, physical function, and self-evaluative OE, and lifestyle self-efficacy were associated with multiple sitting domains in both univariate and covariate-adjusted models. CONCLUSIONS: Future work should examine the relationships between motivational, demographic, and disease predictors and objectively measured sedentary behavior over time and across different sedentary behavior domains. Understanding activity changes during and after treatment is needed to identify intervention targets and develop effective interventions.

Measurement error correction and sensitivity analysis in longitudinal dietary intervention studies using an external validation study.

In lifestyle intervention trials, where the goal is to change a participant's weight or modify their eating behavior, self-reported diet is a longitudinal outcome variable that is subject to measurement error. We propose a statistical framework for correcting for measurement error in longitudinal self-reported dietary data by combining intervention data with auxiliary data from an external biomarker validation study where both self-reported and recovery biomarkers of dietary intake are available. In this setting, dietary intake measured without error in the intervention trial is missing data and multiple imputation is used to fill in the missing measurements. Since most validation studies are cross-sectional, they do not contain information on whether the nature of the measurement error changes over time or differs between treatment and control groups. We use sensitivity analyses to address the influence of these unverifiable assumptions involving the measurement error process and how they affect inferences regarding the effect of treatment. We apply our methods to self-reported sodium intake from the PREMIER study, a multi-component lifestyle intervention trial.

Feasibility and acceptability of intensive longitudinal data collection of activity and patient-reported outcomes during chemotherapy for breast cancer.

PURPOSE: Ecological momentary assessment (EMA) may help us better understand biopsychosocial determinants and outcomes of physical activity during chemotherapy, but may be burdensome for patients. The purpose of this study was to determine the feasibility and acceptability of using EMA to assess activity, symptoms, and motivation among early-stage breast cancer patients undergoing chemotherapy. METHODS: Women were instructed to wear an accelerometer 24/7 (hip during day and wrist overnight). Text message prompts were sent 4 times/day concerning patient-reported symptoms and motivational factors for 10 consecutive days (3 days pre-, day of, and 6 days post-chemotherapy dose). These measures occurred at the beginning, middle, and end of a full course of chemotherapy. At study conclusion, participants reported on perceived study acceptability, burden, and reactivity. RESULTS: Of the 75 women who consented to participate, 63 (84%) completed all 3 assessment time points. Participants responded to 86% of total text prompts and had valid accelerometer data on 82% of study days. Compliance was similar across all time points. The majority (78%) rated their study experience as positive; 100% were confident in their ability to use study technology. Reactivity varied with 27% indicating answering symptom questions did not affect how they felt and 44% and 68% indicated answering questions and wearing the accelerometer, respectively, made them want to increase activity. CONCLUSIONS: Findings indicate EMA methods are feasible for breast cancer patients undergoing chemotherapy. EMA may help us better understand the biopsychosocial processes underlying breast cancer patients' activity in the context of daily life.

Assessment of the influence of food attributes on meal choice selection by socioeconomic status and race/ethnicity among women living in Chicago, USA: A discrete choice experiment.

Large and persistent obesity disparities exist in the US by socioeconomic status (SES) and race/ethnicity, and weight loss interventions have traditionally been less effective in these populations. Thus, a better understanding is needed of the behavioral, economic, and geographic factors that influence obesity risk factors such as eating behaviors. We used a discrete choice experiment to evaluate the impact of different meal attributes on meal choice and to test whether the relative importance of these attributes varied by SES and race/ethnicity. Study participants (n = 228) were given a series of 10 choice tasks and asked to choose among 4 meals, each rated based on the following attributes: taste; healthfulness; preparation time; travel time to food outlet for meal/ingredients; and price. SES was measured using education and self-reported difficulty paying for basics. Race/ethnicity was categorized as Hispanic/Latina, non-Hispanic black, non-Hispanic white, and non-Hispanic other. Data were analyzed using mixed logit regression models with interaction terms to determine whether meal attributes influenced meal choices differentially by SES and race/ethnicity. Healthfulness and taste were the most important attributes for all participants. Price was a more important attribute among those in the lowest SES group compared with those in the higher SES groups. Travel was the least important attribute for low SES participants, and it was not significantly related to meal choice in these groups. Discrete choice experiments as illustrated here may help pinpoint the most salient targets for interventions to improve eating behaviors and reduce obesity disparities. Specifically, our findings suggest interventions should incorporate strategies to target the pricing of healthy and unhealthy food options.

Evaluating Community-Based Translational Interventions Using Historical Controls: Propensity Score vs. Disease Risk Score Approach.

Many community-based translations of evidence-based interventions are designed as one-arm studies due to ethical and other considerations. Evaluating the impacts of such programs is challenging. Here, we examine the effectiveness of the lifestyle intervention implemented by the Special Diabetes Program for Indians Diabetes Prevention (SDPI-DP) demonstration project, a translational lifestyle intervention among American Indian and Alaska Native communities. Data from the landmark Diabetes Prevention Program placebo group was used as a historical control. We compared the use of propensity score (PS) and disease risk score (DRS) matching to adjust for potential confounder imbalance between groups. The unadjusted hazard ratio (HR) for diabetes risk was 0.35 for SDPI-DP lifestyle intervention vs. control. However, when relevant diabetes risk factors were considered, the adjusted HR estimates were attenuated toward 1, ranging from 0.56 (95% CI 0.44-0.71) to 0.69 (95% CI 0.56-0.96). The differences in estimated HRs using the PS and DRS approaches were relatively small but DRS matching resulted in more participants being matched and smaller standard errors of effect estimates. Carefully employed, publicly available randomized clinical trial data can be used as a historical control to evaluate the intervention effectiveness of one-arm community translational initiatives. It is critical to use a proper statistical method to balance the distributions of potential confounders between comparison groups in this kind of evaluations.

Alcohol Use and Cardiovascular Disease Risk in Patients With Nonalcoholic Fatty Liver Disease.

BACKGROUND & AIMS: Cardiovascular disease (CVD) is the leading cause of death among patients with nonalcoholic fatty liver disease (NAFLD). Moderate drinking (vs abstinence) is associated with lower risk of CVD in the general population. We assessed whether alcohol use is associated with CVD risk in patients with NAFLD. METHODS: We analyzed data from participants in the Coronary Artery Risk Development in Young Adults longitudinal cohort study of 5115 black and white young adults, 18-30 years old, recruited from 4 cities in the United States from 1985 through 1986. Participants self-reported alcohol use at study entry and then again after 15, 20, and 25 years. At year 25 (2010-2011), participants underwent computed tomography examination of the thorax and abdomen and tissue Doppler echocardiography with myocardial strain measured by speckle tracking. Coronary artery calcification was defined as an Agatston score above 0. NAFLD was defined as liver attenuation <51 Hounsfield Units after exclusions. Drinkers reported 1-21 (men) or 1-14 (women) standard drinks/week at years 15, 20, or 25. Nondrinkers reported no alcohol use at years 15, 20, and 25. RESULTS: Of the 570 participants with NAFLD (mean age, 50 years; 54% black; 46% female), 332 (58%) were drinkers; significantly higher proportions of drinkers were white, male, and with higher levels of education compared with nondrinkers (P < .05 for all). Higher proportions of drinkers had obesity, diabetes, and metabolic syndrome compared with nondrinkers (P < .01). There was no difference in liver attenuation between groups (P = .12). After multivariable adjustment, there was no association between alcohol use and CVD risk factors (diabetes, hypertension, hyperlipidemia) or subclinical CVD measures (coronary artery calcification, early transmitral velocity/late (atrial) transmitral velocity (E/A) ratio, global longitudinal strain). CONCLUSIONS: In a population-based sample of individuals with NAFLD in midlife, prospectively assessed alcohol use is not associated with significant differences in risk factors for CVD or markers of subclinical CVD. In contrast to general population findings, alcohol use may not reduce the risk of CVD in patients with NAFLD.

Body mass index trajectories in young adulthood predict non-alcoholic fatty liver disease in middle age: The CARDIA cohort study.

BACKGROUND & AIMS: Non-alcoholic fatty liver disease is an epidemic. Identifying modifiable risk factors for non-alcoholic fatty liver disease development is essential to design effective prevention programmes. We tested whether 25-year patterns of body mass index change are associated with midlife non-alcoholic fatty liver disease. METHODS: In all, 4423 participants from Coronary Artery Risk Development in Young Adults, a prospective population-based biracial cohort (age 18-30), underwent body mass index measurement at baseline (1985-1986) and 3 or more times over 25 years. At Year 25, 3115 had liver fat assessed by non-contrast computed tomography. Non-alcoholic fatty liver disease was defined as liver attenuation ≤40 Hounsfield Units after exclusions. Latent mixture modelling identified 25-year trajectories in body mass index per cent change (%Δ) from baseline. RESULTS: We identified four distinct trajectories of BMI%Δ: stable (26.2% of cohort, 25-year BMI %Δ = 3.1%), moderate increase (46.0%, BMI%Δ = 21.7%), high increase (20.9%, BMI%Δ = 41.9%) and extreme increase (6.9%, BMI%Δ = 65.9%). Y25 non-alcoholic fatty liver disease prevalence was higher in groups with greater BMI %Δ: 4.1%, 9.3%, 13.0%, and 17.6%, respectively (P-trend <.0001). In multivariable analyses, participants with increasing BMI%Δ had increasingly greater odds of non-alcoholic fatty liver disease compared to the stable group: OR: 3.35 (95% CI: 2.07-5.42), 7.80 (4.60-13.23) and 12.68 (6.68-24.09) for moderate, high and extreme body mass index increase, respectively. Associations were only moderately attenuated when adjusted for baseline or Y25 body mass index. CONCLUSIONS: Trajectories of weight gain during young adulthood are associated with greater non-alcoholic fatty liver disease prevalence in midlife independent of metabolic covariates and baseline or concurrent body mass index highlighting the importance of weight maintenance throughout adulthood as a target for primary non-alcoholic fatty liver disease prevention.

Twenty-five-year trajectories of insulin resistance and pancreatic ß-cell response and diabetes risk in nonalcoholic fatty liver disease.

BACKGROUND & AIMS: Insulin resistance is a risk marker for non-alcoholic fatty liver disease, and a risk factor for liver disease progression. We assessed temporal trajectories of insulin resistance and ß-cell response to serum glucose concentration throughout adulthood and their association with diabetes risk in non-alcoholic fatty liver disease. METHODS: Three thousand and sixty participants from Coronary Artery Risk Development in Young Adults, a prospective bi-racial cohort of adults age 18-30 years at baseline (1985-1986; Y0) who completed up to 5 exams over 25 years and had fasting insulin and glucose measurement were included. At Y25 (2010-2011), non-alcoholic fatty liver disease was assessed by noncontrast computed tomography after exclusion of other liver fat causes. Latent mixture modelling identified 25-year trajectories in homeostatic model assessment insulin resistance and ß-cell response homeostatic model assessment-ß. RESULTS: Three distinct trajectories were identified, separately, for homeostatic model assessment insulin resistance (low-stable [47%]; moderate-increasing [42%]; and high-increasing [12%]) and homeostatic model assessment-ß (low-decreasing [16%]; moderate-decreasing [63%]; and high-decreasing [21%]). Y25 non-alcoholic fatty liver disease prevalence was 24.5%. Among non-alcoholic fatty liver disease, high-increasing homeostatic model assessment insulin resistance (referent: low-stable) was associated with greater prevalent (OR 95% CI = 8.0, 2.0-31.9) and incident (OR = 10.5, 2.6-32.8) diabetes after multivariable adjustment including Y0 or Y25 homeostatic model assessment insulin resistance. In contrast, non-alcoholic fatty liver disease participants with low-decreasing homeostatic model assessment-ß (referent: high-decreasing) had the highest odds of prevalent (OR = 14.1, 3.9-50.9) and incident (OR = 10.3, 2.7-39.3) diabetes. CONCLUSION: Trajectories of insulin resistance and ß-cell response during young and middle adulthood are robustly associated with diabetes risk in non-alcoholic fatty liver disease. Thus, how persons with non-alcoholic fatty liver disease develop resistance to insulin provides important information about risk of diabetes in midlife above and beyond degree of insulin resistance at the time of non-alcoholic fatty liver disease assessment.

Strategies for imputing missing covariates in accelerated failure time models.

Missing covariates often occur in biomedical studies with survival outcomes. Multiple imputation via chained equations (MICE) is a semi-parametric and flexible approach that imputes multivariate data by a series of conditional models, one for each incomplete variable. When applying MICE, practitioners tend to specify the conditional models in a simple fashion largely dictated by the software, which could lead to suboptimal results. Practical guidelines for specifying appropriate conditional models in MICE are lacking. Motivated by a study of time to hip fractures in the Women's Health Initiative Observational Study using accelerated failure time models, we propose and experiment with some rationales leading to appropriate MICE specifications. This strategy starts with specifying a joint model for the variables involved. We first derive the conditional distribution of each variable under the joint model, then approximate these conditional distributions to the extent which can be characterized by commonly used regression models. We propose to fit separate models to impute incomplete variables by the failure status, which is key to generating appropriate MICE specifications for survival outcomes. The proposed strategy can be conveniently implemented with all available imputation software that uses fully conditional specifications. Our simulation results show that some commonly used simple MICE specifications can produce suboptimal results, while those based on the proposed strategy appear to perform well and be robust toward model misspecifications. Hence, we warn against a mechanical use of MICE and suggest careful modeling of the conditional distributions of variables to ensure proper performance.