A SEMIPARAMETRIC MULTIPLE IMPUTATION APPROACH TO FULLY SYNTHETIC DATA FOR COMPLEX SURVEYS.

Data synthesis is an effective statistical approach for reducing data disclosure risk. Generating fully synthetic data might minimize such risk, but its modeling and application can be difficult for data from large, complex surveys. This article extended the two-stage imputation to simultaneously impute item missing values and generate fully synthetic data. A new combining rule for making inferences using data generated in this manner was developed. Two semiparametric missing data imputation models were adapted to generate fully synthetic data for skewed continuous variable and sparse binary variable, respectively. The proposed approach was evaluated using simulated data and real longitudinal data from the Health and Retirement Study. The proposed approach was also compared with two existing synthesis approaches: (1) parametric regressions models as implemented in IVEware; and (2) nonparametric Classification and Regression Trees as implemented in synthpop package for R using real data. The results show that high data utility is maintained for a wide variety of descriptive and model-based statistics using the proposed strategy. The proposed strategy also performs better than existing methods for sophisticated analyses such as factor analysis.

Small Area Estimation of Cancer Risk Factors and Screening Behaviors in US Counties by Combining Two Large National Health Surveys.

BACKGROUND: National health surveys, such as the National Health Interview Survey (NHIS) and the Behavioral Risk Factor Surveillance System (BRFSS), collect data on cancer screening and smoking-related measures in the US noninstitutionalized population. These surveys are designed to produce reliable estimates at the national and state levels. However, county-level data are often needed for cancer surveillance and related research. METHODS: To use the large sample sizes of BRFSS and the high response rates and better coverage of NHIS, we applied multilevel models that combined information from both surveys. We also used relevant sources such as census and administrative records. By using these methods, we generated estimates for several cancer risk factors and screening behaviors that are more precise than design-based estimates. RESULTS: We produced reliable, modeled estimates for 11 outcomes related to smoking and to screening for female breast cancer, cervical cancer, and colorectal cancer. The estimates were produced for 3,112 counties in the United States for the data period from 2008 through 2010. CONCLUSION: The modeled estimates corrected for potential noncoverage bias and nonresponse bias in the BRFSS and reduced the variability in NHIS estimates that is attributable to small sample size. The small area estimates produced in this study can serve as a useful resource to the cancer surveillance community.

Explaining The Slowdown In Medical Spending Growth Among The Elderly, 1999-2012.

We examined trends in per capita spending for Medicare beneficiaries ages sixty-five and older in the United States in the period 1999-2012 to determine why spending growth has been declining since around 2005. Decomposing spending by condition, we found that half of the spending slowdown was attributable to slower growth in spending for cardiovascular diseases. Spending growth also slowed for dementia, renal and genitourinary diseases, and aftercare for people with acute illnesses. Using estimates from the medical literature of the impact of pharmaceuticals on acute disease, we found that roughly half of the reduction in major cardiovascular events was attributable to medications controlling cardiovascular risk factors. Despite this substantial cost-saving improvement in cardiovascular health, additional opportunities remain to lower spending through disease prevention and control.

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.

Effect of Group-Administered Behavioral Treatment on Urinary Incontinence in Older Women: A Randomized Clinical Trial.

Importance: Urinary incontinence (UI) guidelines recommend behavioral interventions as first-line treatment using individualized approaches. A one-time, group-administered behavioral treatment (GBT) could enhance access to behavioral treatment. Objective: To compare the effectiveness, cost, and cost-effectiveness of GBT with no treatment for UI in older women. Design, Setting, and Participants: Multisite randomized clinical trial (the Group Learning Achieves Decreased Incidents of Lower Urinary Symptoms [GLADIOLUS] study), conducted from July 7, 2014, to December 31, 2016. The setting was outpatient practices at 3 academic medical centers. Community-dwelling women 55 years or older with UI were recruited by mail and screened for eligibility, including a score of 3 or higher on the International Consultation on Incontinence Questionnaire-Short Form (ICIQ-SF), symptoms of at least 3 months' duration, and absence of medical conditions or treatments that could affect continence status. Of 2171 mail respondents, 1125 were invited for clinical screening; 463 were eligible and randomized; 398 completed the 12-month study. Interventions: The GBT group received a one-time 2-hour bladder health class, supported by written materials and an audio CD. Main Outcomes and Measures: Outcomes were measured at in-person visits (at 3 and 12 months) and by mail or telephone (at 6 and 9 months). The primary outcome was the change in the ICIQ-SF score. Secondary outcome measures assessed UI severity, quality of life, perceptions of improvement, pelvic floor muscle strength, and costs. Evaluators were masked to group assignment. Results: Participants (232 in the GBT group and 231 in the control group) were aged 55 to 91 years (mean [SD] age, 64 [7] years), and 46.2% (214 of 463) were African American. In intent-to-treat analyses, the ICIQ-SF scores for GBT were consistently lower than control across all time points but did not achieve the projected 3-point difference. At 3 months, the difference in differences was 0.96 points (95% CI, -1.51 to -0.41 points), which was statistically significant but clinically modest. The mean (SE) treatment effects at 6, 9, and 12 months were 1.36 (0.32), 2.13 (0.33), and 1.77 (0.31), respectively. Significant group differences were found at all time points in favor of GBT on all secondary outcomes except pelvic floor muscle strength. The incremental cost to achieve a treatment success was $723 at 3 months; GBT dominated at 12 months. Conclusions and Relevance: The GLADIOLUS study shows that a novel one-time GBT program is modestly effective and cost-effective for reducing UI frequency, severity, and bother and improving quality of life. Group-administered behavioral treatment is a promising first-line approach to enhancing access to noninvasive behavioral treatment for older women with UI. Trial Registration: ClinicalTrials.gov identifier: NCT02001714.

Longitudinal study of adolescent tobacco use and tobacco control policies in India.

BACKGROUND: This project will use a multilevel longitudinal cohort study design to assess whether changes in Community Tobacco Environmental (CTE) factors, measured as community compliance with tobacco control policies and community density of tobacco vendors and tobacco advertisements, are associated with adolescent tobacco use in urban India. India's tobacco control policies regulate secondhand smoke exposure, access to tobacco products and exposure to tobacco marketing. Research data about the association between community level compliance with tobacco control policies and youth tobacco use are largely unavailable, and are needed to inform policy enforcement, implementation and development. METHODS: The geographic scope will include Mumbai and Kolkata, India. The study protocol calls for an annual comprehensive longitudinal population-based tobacco use risk and protective factors survey in a cohort of 1820 adolescents ages 12-14 years (and their parent) from baseline (Wave 1) to 36-month follow-up (Wave 4). Geographic Information Systems data collection will be used to map tobacco vendors, tobacco advertisements, availability of e-cigarettes, COTPA defined public places, and compliance with tobacco sale, point-of-sale and smoke-free laws. Finally, we will estimate the longitudinal associations between CTE factors and adolescent tobacco use, and assess whether the associations are moderated by family level factors, and mediated by individual level factors. DISCUSSION: India experiences a high burden of disease and mortality from tobacco use. To address this burden, significant long-term prevention and control activities need to include the joint impact of policy, community and family factors on adolescent tobacco use onset. The findings from this study can be used to guide the development and implementation of future tobacco control policy designed to minimize adolescent tobacco use.

FPCA-based method to select optimal sampling schedules that capture between-subject variability in longitudinal studies.

A critical component of longitudinal study design involves determining the sampling schedule. Criteria for optimal design often focus on accurate estimation of the mean profile, although capturing the between-subject variance of the longitudinal process is also important since variance patterns may be associated with covariates of interest or predict future outcomes. Existing design approaches have limited applicability when one wishes to optimize sampling schedules to capture between-individual variability. We propose an approach to derive optimal sampling schedules based on functional principal component analysis (FPCA), which separately characterizes the mean and the variability of longitudinal profiles and leads to a parsimonious representation of the temporal pattern of the variability. Simulation studies show that the new design approach performs equally well compared to an existing approach based on parametric mixed model (PMM) when a PMM is adequate for the data, and outperforms the PMM-based approach otherwise. We use the methods to design studies aiming to characterize daily salivary cortisol profiles and identify the optimal days within the menstrual cycle when urinary progesterone should be measured.

Multiple Imputation in Two-Stage Cluster Samples Using The Weighted Finite Population Bayesian Bootstrap.

Multistage sampling is often employed in survey samples for cost and convenience. However, accounting for clustering features when generating datasets for multiple imputation is a nontrivial task, particularly when, as is often the case, cluster sampling is accompanied by unequal probabilities of selection, necessitating case weights. Thus, multiple imputation often ignores complex sample designs and assumes simple random sampling when generating imputations, even though failing to account for complex sample design features is known to yield biased estimates and confidence intervals that have incorrect nominal coverage. In this article, we extend a recently developed, weighted, finite-population Bayesian bootstrap procedure to generate synthetic populations conditional on complex sample design data that can be treated as simple random samples at the imputation stage, obviating the need to directly model design features for imputation. We develop two forms of this method: one where the probabilities of selection are known at the first and second stages of the design, and the other, more common in public use files, where only the final weight based on the product of the two probabilities is known. We show that this method has advantages in terms of bias, mean square error, and coverage properties over methods where sample designs are ignored, with little loss in efficiency, even when compared with correct fully parametric models. An application is made using the National Automotive Sampling System Crashworthiness Data System, a multistage, unequal probability sample of U.S. passenger vehicle crashes, which suffers from a substantial amount of missing data in "Delta-V," a key crash severity measure.

Synthetic Multiple-Imputation Procedure for Multistage Complex Samples.

Multiple imputation (MI) is commonly used when item-level missing data are present. However, MI requires that survey design information be built into the imputation models. For multistage stratified clustered designs, this requires dummy variables to represent strata as well as primary sampling units (PSUs) nested within each stratum in the imputation model. Such a modeling strategy is not only operationally burdensome but also inferentially inefficient when there are many strata in the sample design. Complexity only increases when sampling weights need to be modeled. This article develops a general-purpose analytic strategy for population inference from complex sample designs with item-level missingness. In a simulation study, the proposed procedures demonstrate efficient estimation and good coverage properties. We also consider an application to accommodate missing body mass index (BMI) data in the analysis of BMI percentiles using National Health and Nutrition Examination Survey (NHANES) III data. We argue that the proposed methods offer an easy-to-implement solution to problems that are not well-handled by current MI techniques. Note that, while the proposed method borrows from the MI framework to develop its inferential methods, it is not designed as an alternative strategy to release multiply imputed datasets for complex sample design data, but rather as an analytic strategy in and of itself.

Erythrocyte very long-chain saturated fatty acids associated with lower risk of incident sudden cardiac arrest.

Prior studies suggest that circulating n-3 and trans-fatty acids influence the risk of sudden cardiac arrest (SCA). Yet, while other fatty acids also differ in their membrane properties and biological activities which may influence SCA, little is known about the associations of other circulating fatty acids with SCA. The aim of this study was to investigate the associations of 17 erythrocyte membrane fatty acids with SCA risk. We used data from a population-based case-control study of SCA in the greater Seattle, Washington, area. Cases, aged 25-74 years, were out-of-hospital SCA patients, attended by paramedics (n=265). Controls, matched to cases by age, sex and calendar year, were randomly identified from the community (n=415). All participants were free of prior clinically-diagnosed heart disease. Blood was obtained at the time of cardiac arrest by attending paramedics (cases) or at the time of an interview (controls). Higher levels of erythrocyte very long-chain saturated fatty acids (VLSFA) were associated with lower risk of SCA. After adjustment for risk factors and levels of n-3 and trans-fatty acids, higher levels of 20:0 corresponding to 1 SD were associated with 30% lower SCA risk (13-43%, p=0.001). Higher levels of 22:0 and 24:0 were associated with similar lower SCA risk (ORs for 1 SD-difference: 0.71 [95% CI: 0.57-0.88, p=0.002] for 22:0; and 0.79 [95% CI: 0.63-0.98, p=0.04] for 24:0). These novel findings support the need for investigation of biologic effects of circulating VLSFA and their determinants.

A nonparametric method to generate synthetic populations to adjust for complex sampling design features.

Outside of the survey sampling literature, samples are often assumed to be generated by a simple random sampling process that produces independent and identically distributed (IID) samples. Many statistical methods are developed largely in this IID world. Application of these methods to data from complex sample surveys without making allowance for the survey design features can lead to erroneous inferences. Hence, much time and effort have been devoted to develop the statistical methods to analyze complex survey data and account for the sample design. This issue is particularly important when generating synthetic populations using finite population Bayesian inference, as is often done in missing data or disclosure risk settings, or when combining data from multiple surveys. By extending previous work in finite population Bayesian bootstrap literature, we propose a method to generate synthetic populations from a posterior predictive distribution in a fashion inverts the complex sampling design features and generates simple random samples from a superpopulation point of view, making adjustment on the complex data so that they can be analyzed as simple random samples. We consider a simulation study with a stratified, clustered unequal-probability of selection sample design, and use the proposed nonparametric method to generate synthetic populations for the 2006 National Health Interview Survey (NHIS), and the Medical Expenditure Panel Survey (MEPS), which are stratified, clustered unequal-probability of selection sample designs.

Combining information from multiple complex surveys.

This manuscript describes the use of multiple imputation to combine information from multiple surveys of the same underlying population. We use a newly developed method to generate synthetic populations nonparametrically using a finite population Bayesian bootstrap that automatically accounting for complex sample designs. We then analyze each synthetic population with standard complete-data software for simple random samples and obtain valid inference by combining the point and variance estimates using extensions of existing combining rules for synthetic data. We illustrate the approach by combining data from the 2006 National Health Interview Survey (NHIS) and the 2006 Medical Expenditure Panel Survey (MEPS).

Modeling the salivary cortisol profile in population research: the multi-ethnic study of atherosclerosis.

In many studies, it has been hypothesized that stress and its biologic consequences may contribute to disparities in rates of cardiovascular disease. However, understanding of the most appropriate statistical methods to analyze biologic markers of stress, such as salivary cortisol, remains limited. The authors explore the utility of various statistical methods in modeling daily cortisol profiles in population-based studies. They demonstrate that the proposed methods allow additional insight into the cortisol profile compared with commonly used summaries of the profiles based on raw data. For instance, one can gain insights regarding the shape of the population average curve, characterize the types of individual-level departures from the average curve, and better understand the relation between covariates and attained cortisol levels or slopes at various points of the day, in addition to drawing inferences regarding common features of the cortisol profile, such as the cortisol awakening response and the area under the curve. The authors compare the inference and interpretations drawn from these methods and use data collected as part of the Multi-Ethnic Study of Atherosclerosis to illustrate them.

On Bayesian methods of exploring qualitative interactions for targeted treatment.

Providing personalized treatments designed to maximize benefits and minimizing harms is of tremendous current medical interest. One problem in this area is the evaluation of the interaction between the treatment and other predictor variables. Treatment effects in subgroups having the same direction but different magnitudes are called quantitative interactions, whereas those having opposite directions in subgroups are called qualitative interactions (QIs). Identifying QIs is challenging because they are rare and usually unknown among many potential biomarkers. Meanwhile, subgroup analysis reduces the power of hypothesis testing and multiple subgroup analyses inflate the type I error rate. We propose a new Bayesian approach to search for QI in a multiple regression setting with adaptive decision rules. We consider various regression models for the outcome. We illustrate this method in two examples of phase III clinical trials. The algorithm is straightforward and easy to implement using existing software packages. We provide a sample code in Appendix A.

A Bayesian model for time-to-event data with informative censoring.

Randomized trials with dropouts or censored data and discrete time-to-event type outcomes are frequently analyzed using the Kaplan-Meier or product limit (PL) estimation method. However, the PL method assumes that the censoring mechanism is noninformative and when this assumption is violated, the inferences may not be valid. We propose an expanded PL method using a Bayesian framework to incorporate informative censoring mechanism and perform sensitivity analysis on estimates of the cumulative incidence curves. The expanded method uses a model, which can be viewed as a pattern mixture model, where odds for having an event during the follow-up interval $$({t}_{k-1},{t}_{k}]$$, conditional on being at risk at $${t}_{k-1}$$, differ across the patterns of missing data. The sensitivity parameters relate the odds of an event, between subjects from a missing-data pattern with the observed subjects for each interval. The large number of the sensitivity parameters is reduced by considering them as random and assumed to follow a log-normal distribution with prespecified mean and variance. Then we vary the mean and variance to explore sensitivity of inferences. The missing at random (MAR) mechanism is a special case of the expanded model, thus allowing exploration of the sensitivity to inferences as departures from the inferences under the MAR assumption. The proposed approach is applied to data from the TRial Of Preventing HYpertension.

A functional multiple imputation approach to incomplete longitudinal data.

In designed longitudinal studies, information from the same set of subjects are collected repeatedly over time. The longitudinal measurements are often subject to missing data which impose an analytic challenge. We propose a functional multiple imputation approach modeling longitudinal response profiles as smooth curves of time under a functional mixed effects model. We develop a Gibbs sampling algorithm to draw model parameters and imputations for missing values, using a blocking technique for an increased computational efficiency. In an illustrative example, we apply a multiple imputation analysis to data from the Panel Study of Income Dynamics and the Child Development Supplement to investigate the gradient effect of family income on children's health status. Our simulation study demonstrates that this approach performs well under varying modeling assumptions on the time trajectory functions and missingness patterns.

Deeper and wider: income and mortality in the USA over three decades.

BACKGROUND: Literature on the socio-economic 'gradient' in health often asserts that income is associated with better health not only for the very poor, but also across the entire income distribution. In addition, changes in the shape of the association between incomes during a period of increasing economic inequality have not been previously studied. The goal of the current study was to estimate and compare the shape of the relationship between income and mortality in the USA for the 1970s, the 1980s and the 1990s. METHODS: Using income and mortality data from the Panel Study of Income Dynamics for respondents aged 35-64 years, we used a Bayesian Cox Model with regression splines to model the risk of mortality over three 10-year follow-up periods. To identify whether income was more strongly associated with mortality at different parts of the income distribution, we treated income as a linear spline with an unknown knot location. RESULTS: The shape of the association between income and mortality was quite non-linear, with a much stronger association at lower levels of income. The relationship between income and mortality in the USA was also not invariant across time, with the increased risk of death associated with lower income applying to an increasing proportion of the US population over time (9th percentile of income in 1970-79, 20th percentile in 1980-89 and 32nd percentile in 1990-99). CONCLUSIONS: Our analyses do not support the claim that income is associated with mortality throughout the income distribution, nor is the association between income and mortality the same across periods. Based on our analyses, a focus on the bottom 30% of the income distribution would seem to return the greatest benefits in reducing socio-economic inequalities in health.

Efficacy of a web-based, tailored, alcohol prevention/intervention program for college students: 3-month follow-up.

This study presents the results of an efficacy evaluation of a web-based brief motivational alcohol prevention/intervention program called Michigan Prevention and Alcohol Safety for Students (M-PASS). Four on-line sessions providing individually-tailored feedback were delivered to first-year college students over 9 weeks. Non- and low-risk drinking participants received risk prevention, while high-risk drinking participants received a risk-reduction intervention. Both intervention and control groups were surveyed at baseline and at a 3-month follow-up. Analysis showed positive effects for both men and women on stage of change, drinking behavior, drinking motivation and attitudes, and use of risk-reduction strategies. These results provided evidence of efficacy and found that M-PASS had both intervention and prevention effects, making it unique among currently developed brief alcohol interventions for college students.

State-based estimates of mammography screening rates based on information from two health surveys.

OBJECTIVES: We compared national and state-based estimates for the prevalence of mammography screening from the National Health Interview Survey (NHIS), the Behavioral Risk Factor Surveillance System (BRFSS), and a model-based approach that combines information from the two surveys. METHODS: At the state and national levels, we compared the three estimates of prevalence for two time periods (1997-1999 and 2000-2003) and the estimated difference between the periods. We included state-level covariates in the model-based approach through principal components. RESULTS: The national mammography screening prevalence estimate based on the BRFSS was substantially larger than the NHIS estimate for both time periods. This difference may have been due to nonresponse and noncoverage biases, response mode (telephone vs. in-person) differences, or other factors. However, the estimated change between the two periods was similar for the two surveys. Consistent with the model assumptions, the model-based estimates were more similar to the NHIS estimates than to the BRFSS prevalence estimates. The state-level covariates (through the principal components) were shown to be related to the mammography prevalence with the expected positive relationship for socioeconomic status and urbanicity. In addition, several principal components were significantly related to the difference between NHIS and BRFSS telephone prevalence estimates. CONCLUSIONS: Model-based estimates, based on information from the two surveys, are useful tools in representing combined information about mammography prevalence estimates from the two surveys. The model-based approach adjusts for the possible nonresponse and noncoverage biases of the telephone survey while using the large BRFSS state sample size to increase precision.

Type 2 diabetes mellitus and the risk of sudden cardiac arrest in the community.

The reduction of mortality from sudden cardiac arrest (SCA) in the setting of coronary heart disease (CHD) remains a major challenge, especially among patients with type 2 diabetes. Diabetes is associated with an increased risk of SCA, at least in part, from an increased presence and extent of coronary atherosclerosis (macrovascular disease). Diabetes also is associated with microvascular disease and autonomic neuropathy; and, these non-coronary atherosclerotic pathophysiologic processes also have the potential to increase the risk of SCA. In this report, we review the absolute and relative risk of SCA associated with diabetes. We summarize recent evidence that suggests that the increase in risk in patients with diabetes is not specific for SCA, as diabetes also is associated with a similar increase in risk for non-SCA CHD death and non-fatal myocardial infarction. These data are consistent with prior observations that coronary atherosclerosis is a major contributor to the increased SCA risk associated with diabetes. We also present previously published and unpublished data that demonstrates that both clinically-recognized microvascular and autonomic neuropathy also are associated with the risk of SCA among treated patients with diabetes, after taking into account prior clinically-recognized heart disease and other risk factors for SCA. We then discuss how these data might inform research and clinical efforts to prevent SCA. Although the prediction of SCA in this "high" risk population is likely to remain a challenge, as it is in other "high" risk clinical populations, we suggest that current recommendations for the prevention of SCA in the community, related to both lifestyle prescriptions and risk factor reduction, are likely to reduce mortality from SCA among patients with diabetes.