Using Power Analysis to Choose the Unit of Randomization, Outcome, and Approach for Subgroup Analysis for a Multilevel Randomized Controlled Clinical Trial to Reduce Disparities in Cardiovascular Health.

We give examples of three features in the design of randomized controlled clinical trials which can increase power and thus decrease sample size and costs. We consider an example multilevel trial with several levels of clustering. For a fixed number of independent sampling units, we show that power can vary widely with the choice of the level of randomization. We demonstrate that power and interpretability can improve by testing a multivariate outcome rather than an unweighted composite outcome. Finally, we show that using a pooled analytic approach, which analyzes data for all subgroups in a single model, improves power for testing the intervention effect compared to a stratified analysis, which analyzes data for each subgroup in a separate model. The power results are computed for a proposed prevention research study. The trial plans to randomize adults to either telehealth (intervention) or in-person treatment (control) to reduce cardiovascular risk factors. The trial outcomes will be measures of the Essential Eight, a set of scores for cardiovascular health developed by the American Heart Association which can be combined into a single composite score. The proposed trial is a multilevel study, with outcomes measured on participants, participants treated by the same provider, providers nested within clinics, and clinics nested within hospitals. Investigators suspect that the intervention effect will be greater in rural participants, who live farther from clinics than urban participants. The results use published, exact analytic methods for power calculations with continuous outcomes. We provide example code for power analyses using validated software.

Stratified split sampling of electronic health records.

Although superficially similar to data from clinical research, data extracted from electronic health records may require fundamentally different approaches for model building and analysis. Because electronic health record data is designed for clinical, rather than scientific use, researchers must first provide clear definitions of outcome and predictor variables. Yet an iterative process of defining outcomes and predictors, assessing association, and then repeating the process may increase Type I error rates, and thus decrease the chance of replicability, defined by the National Academy of Sciences as the chance of "obtaining consistent results across studies aimed at answering the same scientific question, each of which has obtained its own data."[1] In addition, failure to account for subgroups may mask heterogeneous associations between predictor and outcome by subgroups, and decrease the generalizability of the findings. To increase chances of replicability and generalizability, we recommend using a stratified split sample approach for studies using electronic health records. A split sample approach divides the data randomly into an exploratory set for iterative variable definition, iterative analyses of association, and consideration of subgroups. The confirmatory set is used only to replicate results found in the first set. The addition of the word 'stratified' indicates that rare subgroups are oversampled randomly by including them in the exploratory sample at higher rates than appear in the population. The stratified sampling provides a sufficient sample size for assessing heterogeneity of association by testing for effect modification by group membership. An electronic health record study of the associations between socio-demographic factors and uptake of hepatic cancer screening, and potential heterogeneity of association in subgroups defined by gender, self-identified race and ethnicity, census-tract level poverty and insurance type illustrates the recommended approach.

Power for balanced linear mixed models with complex missing data processes.

When designing repeated measures studies, both the amount and the pattern of missing outcome data can affect power. The chance that an observation is missing may vary across measurements, and missingness may be correlated across measurements. For example, in a physiotherapy study of patients with Parkinson's disease, increasing intermittent dropout over time yielded missing measurements of physical function. In this example, we assume data are missing completely at random, since the chance that a data point was missing appears to be unrelated to either outcomes or covariates. For data missing completely at random, we propose noncentral F power approximations for the Wald test for balanced linear mixed models with Gaussian responses. The power approximations are based on moments of missing data summary statistics. The moments were derived assuming a conditional linear missingness process. The approach provides approximate power for both complete-case analyses, which include independent sampling units where all measurements are present, and observed-case analyses, which include all independent sampling units with at least one measurement. Monte Carlo simulations demonstrate the accuracy of the method in small samples. We illustrate the utility of the method by computing power for proposed replications of the Parkinson's study.

Power and sample size analysis for longitudinal mixed models of health in populations exposed to environmental contaminants: a tutorial.

BACKGROUND: When evaluating the impact of environmental exposures on human health, study designs often include a series of repeated measurements. The goal is to determine whether populations have different trajectories of the environmental exposure over time. Power analyses for longitudinal mixed models require multiple inputs, including clinically significant differences, standard deviations, and correlations of measurements. Further, methods for power analyses of longitudinal mixed models are complex and often challenging for the non-statistician. We discuss methods for extracting clinically relevant inputs from literature, and explain how to conduct a power analysis that appropriately accounts for longitudinal repeated measures. Finally, we provide careful recommendations for describing complex power analyses in a concise and clear manner. METHODS: For longitudinal studies of health outcomes from environmental exposures, we show how to [1] conduct a power analysis that aligns with the planned mixed model data analysis, [2] gather the inputs required for the power analysis, and [3] conduct repeated measures power analysis with a highly-cited, validated, free, point-and-click, web-based, open source software platform which was developed specifically for scientists. RESULTS: As an example, we describe the power analysis for a proposed study of repeated measures of per- and polyfluoroalkyl substances (PFAS) in human blood. We show how to align data analysis and power analysis plan to account for within-participant correlation across repeated measures. We illustrate how to perform a literature review to find inputs for the power analysis. We emphasize the need to examine the sensitivity of the power values by considering standard deviations and differences in means that are smaller and larger than the speculated, literature-based values. Finally, we provide an example power calculation and a summary checklist for describing power and sample size analysis. CONCLUSIONS: This paper provides a detailed roadmap for conducting and describing power analyses for longitudinal studies of environmental exposures. It provides a template and checklist for those seeking to write power analyses for grant applications.

Florida Medicaid Children's Receipt of First-Line Psychosocial Care Prior to Antipsychotic Initiation.

BACKGROUND AND OBJECTIVE: First-line, nonpharmacological therapy is recommended for many pediatric mental health (MH) conditions prior to initiating antipsychotic prescription therapies. Many children do not receive these recommended services, despite the known association between antipsychotic medications and metabolic dysfunction. The main objective of this study was to quantify the association among children's MH diagnosis categories, sociodemographic characteristics and receipt of first-line psychosocial care among children in Florida Medicaid METHODS: Florida Medicaid enrollment, healthcare and pharmacy claims were used for this multivariate analysis. Children were assigned to condition clusters wherein related diagnoses were grouped into clinically relevant categories. A total of 7704 children were included in the final analysis. RESULTS: Twenty-four percent of children in Florida Medicaid do not receive first-line, nonpharmacological psychosocial care. Age was significantly associated with not receiving psychosocial services, with older children less likely to receive. Non-Hispanic White children as well as those living in rural areas had lower odds of receiving behavioral intervention prior to initiating antipsychotics. Children with mood-disorders, behavior problems, anxiety and stress related disorders were more likely to receive first-line psychosocial care. CONCLUSIONS: This study provides an important understanding of the variability in receipt of first-line psychosocial care before antipsychotic medication initiation among children in Medicaid based on sociodemographic and MH health characteristics. These analyses can be used to develop quality improvement initiatives targeted toward children that are most vulnerable for not receiving recommended care.

Enhancing Quality Measurement With Clinical Information: A Use Case of Body Mass Index Change Among Children Taking Second Generation Antipsychotics.

OBJECTIVE: We sought to examine the extent to which body mass index (BMI) was available in electronic health records for Florida Medicaid recipients aged 5 to 18 years taking Second-Generation Antipsychotics (SGAP). We also sought to illustrate how clinical data can be used to identify children most at-risk for SGAP-induced weight gain, which cannot be done using process-focused measures. METHODS: Electronic health record (EHR) data and Medicaid claims were linked from 2013 to 2019. We quantified sociodemographic differences between children with and without pre- and post-BMI values. We developed a linear regression model of post-BMI to examine pre-post changes in BMI among 4 groups: 1) BH/SGAP+ children had behavioral health conditions and were taking SGAP; 2) BH/SGAP- children had behavioral health conditions without taking SGAP; 3) children with asthma; and 4) healthy children. RESULTS: Of 363,360 EHR-Medicaid linked children, 18,726 were BH/SGAP+. Roughly 4% of linked children and 8% of BH/SGAP+ children had both pre and post values of BMI required to assess quality of SGAP monitoring. The percentage varied with gender and race-ethnicity. The R2 for the regression model with all predictors was 0.865. Pre-post change in BMI differed significantly (P < .0001) among the groups, with more BMI gain among those taking SGAP, particularly those with higher baseline BMI. CONCLUSION: Meeting the 2030 Centers for Medicare and Medicaid Services goal of digital monitoring of quality of care will require continuing expansion of clinical encounter data capture to provide the data needed for digital quality monitoring. Using linked EHR and claims data allows identifying children at higher risk for SGAP-induced weight gain.

Do patients at high risk for Hepatitis C receive recommended testing? A retrospective cohort study of statewide Medicaid claims linked with OneFlorida clinical data.

ABSTRACT: Hepatitis C virus (HCV) infection is a leading risk factor for hepatocellular carcinoma.We employed a retrospective cohort study design and analyzed 2012-2018 Medicaid claims linked with electronic health records data from the OneFlorida Data Trust, a statewide data repository containing electronic health records data for 15.07 million Floridians from 11 health care systems. Only adult patients at high-risk for HCV (n = 30,113), defined by diagnosis of: HIV/AIDS (20%), substance use disorder (64%), or sexually transmitted infections (22%) were included. Logistic regression examined factors associated with meeting the recommended sequence of HCV testing.Overall, 44.1% received an HCV test. The odds of receiving an initial test were significantly higher for pregnant females (odds ratio [OR]1.99; 95% confidence interval [CI] 1.86-2.12; P < .001) and increased with age (OR 1.01; 95% CI 1.00-1.01; P < .001).Among patients with low Charlson comorbidity index (CCI = 1), non-Hispanic (NH) black patients (OR 0.86; 95% CI 0.81-0.9; P < .001) had lower odds of getting an HCV test; however, NH black patients with CCI = 10 had higher odds (OR 1.41; 95% CI 1.21-1.66; P < .001) of receiving a test. Of those who tested negative during initial testing, 17% received a second recommended test after 6 to 24 months. Medicaid-Medicare dual eligible patients, those with high CCI (OR 1.14; 95% CI 1.11-1.17; P < .001), NH blacks (OR 1.93; 95% CI 1.61-2.32; P < .001), and Hispanics (OR 1.49; 95% CI 1.08-2.06; P = .02) were significantly more likely to have received a second HCV test, while pregnant females (OR 0.71; 95% CI 0.57-0.89; P = .003), had lower odds of receiving it. The majority of patients who tested positive during the initial test (97%) received subsequent testing.We observed suboptimal adherence to the recommended HCV testing among high-risk patients underscoring the need for tailored interventions aimed at successfully navigating high-risk individuals through the HCV screening process. Future interventional studies targeting multilevel factors, including patients, clinicians and health systems are needed to increase HCV screening rates for high-risk populations.

A power approximation for the Kenward and Roger Wald test in the linear mixed model.

We derive a noncentral [Formula: see text] power approximation for the Kenward and Roger test. We use a method of moments approach to form an approximate distribution for the Kenward and Roger scaled Wald statistic, under the alternative. The result depends on the approximate moments of the unscaled Wald statistic. Via Monte Carlo simulation, we demonstrate that the new power approximation is accurate for cluster randomized trials and longitudinal study designs. The method retains accuracy for small sample sizes, even in the presence of missing data. We illustrate the method with a power calculation for an unbalanced group-randomized trial in oral cancer prevention.

The Wellness Incentive and Navigation intervention improved health-related quality of life among Medicaid enrollees: A randomized pragmatic clinical trial.

OBJECTIVE: To examine whether the Wellness Incentive and Navigation (WIN) intervention can improve health-related quality of life (HRQOL) among Medicaid enrollees with co-occurring physical and behavioral health conditions. DATA SOURCES: Annual telephone survey data from 2013 to 2016, linked with claims data. STUDY DESIGN: We recruited 1259 participants from the Texas STAR + PLUS managed care program and randomized them into an intervention group that received flexible wellness accounts and navigator services or a control group that received standard care. We conducted 4 waves of telephone surveys to collect data on HRQOL, patient activation, and other participant demographic and clinical characteristics. DATA COLLECTION/EXTRACTION METHODS: The 3M Clinical Risk Grouping Software was used to extract variables from claims data and group participants based on disease severity. PRINCIPAL FINDINGS: Our results showed that the WIN intervention was effective in increasing patient activation and HRQOL among Medicaid enrollees with co-occurring physical and behavioral health conditions. Furthermore, we found that this intervention effect on HRQOL was partially mediated by patient activation. CONCLUSIONS: Providing navigator support with wellness account is effective in improving HRQOL among Medicaid enrollees. The pragmatic nature of the trial maximizes the chance of successfully implementing it in state Medicaid programs.

Transformation of a face-to-face workshop into a Massive Open Online Course (MOOC): A design and development case.

The purpose of this design and development case is to share our experiences in the transformation of a face-to-face workshop into a Massive Open Online Course (MOOC) for a prominent MOOC platform. The goal of the workshop and MOOC is to teach learners how to conduct appropriate power and sample size analysis for multilevel and longitudinal studies in social and behavioral health research. Learners include people from across the biomedical research spectrum, from students to full professors. We first describe the design and development frameworks and processes used to create the three-day, face-to-face workshop. Then, we detail the design and development approach to transform this face-to-face workshop into a MOOC. At a macro-design level, we employed backward design (Wiggins & McTighe, 1998) as an instructional design framework. At a micro-design level, we used a combination of the first principles of instruction, the cognitive theory of multimedia learning, the nine events of instruction, and design recommendations for MOOCs found in the literature. We report the results of a formative evaluation of the MOOC. Finally, we provide closing remarks, lessons learned, and the next steps for the instructional program.

Power and Sample Size for Fixed-Effects Inference in Reversible Linear Mixed Models.

Despite the popularity of the general linear mixed model for data analysis, power and sample size methods and software are not generally available for commonly used test statistics and reference distributions. Statisticians resort to simulations with homegrown and uncertified programs or rough approximations which are misaligned with the data analysis. For a wide range of designs with longitudinal and clustering features, we provide accurate power and sample size approximations for inference about fixed effects in linear models we call reversible. We show that under widely applicable conditions, the general linear mixed-model Wald test has non-central distributions equivalent to well-studied multivariate tests. In turn, exact and approximate power and sample size results for the multivariate Hotelling-Lawley test provide exact and approximate power and sample size results for the mixed-model Wald test. The calculations are easily computed with a free, open-source product that requires only a web browser to use. Commercial software can be used for a smaller range of reversible models. Simple approximations allow accounting for modest amounts of missing data. A real-world example illustrates the methods. Sample size results are presented for a multicenter study on pregnancy. The proposed study, an extension of a funded project, has clustering within clinic. Exchangeability among participants allows averaging across them to remove the clustering structure. The resulting simplified design is a single level longitudinal study. Multivariate methods for power provide an approximate sample size. All proofs and inputs for the example are in the Supplementary Materials (available online).

Trajectories of Nevus Development From Age 3 to 16 Years in the Colorado Kids Sun Care Program Cohort.

Importance: Nevi are a risk factor for melanoma and other forms of skin cancer, and many of the same factors confer risk for both. Understanding childhood nevus development may provide clues to possible causes and prevention of melanoma. Objectives: To describe nevus acquisition from the ages of 3 to 16 years among white youths and evaluate variation by sex, Hispanic ethnicity, and body sites that are chronically vs intermittently exposed to the sun. Design, Setting, and Participants: This annual longitudinal observational cohort study of nevus development was conducted between June 1, 2001, and October 31, 2014, among 1085 Colorado youths. Data analysis was conducted between February 1, 2015, and August 31, 2017. Main Outcomes and Measures: Total nevus counts on all body sites and on sites chronically and intermittently exposed to the sun separately. Results: A total of 557 girls and 528 boys (150 [13.8%] Hispanic participants) born in 1998 were included in this study. Median total body nevus counts increased linearly among non-Hispanic white boys and girls between the age of 3 years (boys, 6.31; 95% CI, 5.66-7.03; and girls, 6.61; 95% CI, 5.96-7.33) and the age of 16 years (boys, 81.30; 95% CI, 75.95-87.03; and girls, 77.58; 95% CI, 72.68-82.81). Median total body nevus counts were lower among Hispanic white children (boys aged 16 years, 51.45; 95% CI, 44.01-60.15; and girls aged 16 years, 53.75; 95% CI, 45.40-63.62) compared with non-Hispanic white children, but they followed a largely linear trend that varied by sex. Nevus counts on body sites chronically exposed to the sun increased over time but leveled off by the age of 16 years. Nevus counts on sites intermittently exposed to the sun followed a strong linear pattern through the age of 16 years. Hispanic white boys and girls had similar nevus counts on sites intermittently exposed to the sun through the age of 10 years, but increases thereafter were steeper for girls, with nevus counts surpassing those of boys aged 11 to 16 years. Conclusions and Relevance: Youths are at risk for nevus development beginning in early childhood and continuing through midadolescence. Patterns of nevus acquisition differ between boys and girls, Hispanic and non-Hispanic white youths, and body sites that are chronically exposed to the sun and body sites that are intermittently exposed to the sun. Exposure to UV light during this period should be reduced, particularly on body sites intermittently exposed to the sun, where nevi accumulate through midadolescence in all children. Increased attention to sun protection appears to be merited for boys, in general, because they accumulated more nevi overall, and for girls, specifically, during the adolescent years.

ON THE TRACE OF A WISHART.

A derivation based on spectral decomposition allows specifying the characteristic function of the trace of a singular or nonsingular, central or noncentral, true or pseudo-Wishart. The trace equals a weighted sum of noncentral chi-squared random variables and constants. We describe computational methods.

Internal pilot design for balanced repeated measures.

Repeated measures are common in clinical trials and epidemiological studies. Designing studies with repeated measures requires reasonably accurate specifications of the variances and correlations to select an appropriate sample size. Underspecifying the variances leads to a sample size that is inadequate to detect a meaningful scientific difference, while overspecifying the variances results in an unnecessarily large sample size. Both lead to wasting resources and placing study participants in unwarranted risk. An internal pilot design allows sample size recalculation based on estimates of the nuisance parameters in the covariance matrix. We provide the theoretical results that account for the stochastic nature of the final sample size in a common class of linear mixed models. The results are useful for designing studies with repeated measures and balanced design. Simulations examine the impact of misspecification of the covariance matrix and demonstrate the accuracy of the approximations in controlling the type I error rate and achieving the target power. The proposed methods are applied to a longitudinal study assessing early antiretroviral therapy for youth living with HIV.

Multivariate test power approximations for balanced linear mixed models in studies with missing data.

Multilevel and longitudinal studies are frequently subject to missing data. For example, biomarker studies for oral cancer may involve multiple assays for each participant. Assays may fail, resulting in missing data values that can be assumed to be missing completely at random. Catellier and Muller proposed a data analytic technique to account for data missing at random in multilevel and longitudinal studies. They suggested modifying the degrees of freedom for both the Hotelling-Lawley trace F statistic and its null case reference distribution. We propose parallel adjustments to approximate power for this multivariate test in studies with missing data. The power approximations use a modified non-central F statistic, which is a function of (i) the expected number of complete cases, (ii) the expected number of non-missing pairs of responses, or (iii) the trimmed sample size, which is the planned sample size reduced by the anticipated proportion of missing data. The accuracy of the method is assessed by comparing the theoretical results to the Monte Carlo simulated power for the Catellier and Muller multivariate test. Over all experimental conditions, the closest approximation to the empirical power of the Catellier and Muller multivariate test is obtained by adjusting power calculations with the expected number of complete cases. The utility of the method is demonstrated with a multivariate power analysis for a hypothetical oral cancer biomarkers study. We describe how to implement the method using standard, commercially available software products and give example code. Copyright © 2015 John Wiley & Sons, Ltd.

Quality of Care for Chronic Conditions Among Disabled Medicaid Enrollees: An Evaluation of a 1915 (b) and (c) Waiver Program.

IMPORTANCE: Examining the impact of Medicaid-managed care home-based and community-based service (HCBS) alternatives to institutional care is critical given the recent rapid expansion of these models nationally. OBJECTIVE: We analyzed the effects of STAR+PLUS, a Texas Medicaid-managed care HCBS waiver program for adults with disabilities on the quality of chronic disease care. DESIGN, SETTING, AND PARTICIPANTS: We compared quality before and after a mandatory transition of disabled Medicaid enrollees older than 21 years from fee-for-service (FFS) or primary care case management (PCCM) to STAR+PLUS in 28 counties, relative to enrollees in counties remaining in the FFS or PCCM models. MEASURES AND ANALYSIS: Person-level claims and encounter data for 2006-2010 were used to compute adherence to 6 quality measures. With county as the independent sampling unit, we employed a longitudinal linear mixed-model analysis accounting for administrative clustering and geographic and individual factors. RESULTS: Although quality was similar among programs at baseline, STAR+PLUS enrollees experienced large and sustained improvements in use of ß-blockers after discharge for heart attack (49% vs. 81% adherence posttransition; P<0.01) and appropriate use of systemic corticosteroids and bronchodilators after a chronic obstructive pulmonary disease event (39% vs. 68% adherence posttransition; P<0.0001) compared with FFS/PCCM enrollees. No statistically significant effects were identified for quality measures for asthma, diabetes, or cardiovascular disease. CONCLUSION: In 1 large Medicaid-managed care HCBS program, the quality of chronic disease care linked to acute events improved while that provided during routine encounters appeared unaffected.

Recommendations for choosing an analysis method that controls Type I error for unbalanced cluster sample designs with Gaussian outcomes.

We used theoretical and simulation-based approaches to study Type I error rates for one-stage and two-stage analytic methods for cluster-randomized designs. The one-stage approach uses the observed data as outcomes and accounts for within-cluster correlation using a general linear mixed model. The two-stage model uses the cluster specific means as the outcomes in a general linear univariate model. We demonstrate analytically that both one-stage and two-stage models achieve exact Type I error rates when cluster sizes are equal. With unbalanced data, an exact size α test does not exist, and Type I error inflation may occur. Via simulation, we compare the Type I error rates for four one-stage and six two-stage hypothesis testing approaches for unbalanced data. With unbalanced data, the two-stage model, weighted by the inverse of the estimated theoretical variance of the cluster means, and with variance constrained to be positive, provided the best Type I error control for studies having at least six clusters per arm. The one-stage model with Kenward-Roger degrees of freedom and unconstrained variance performed well for studies having at least 14 clusters per arm. The popular analytic method of using a one-stage model with denominator degrees of freedom appropriate for balanced data performed poorly for small sample sizes and low intracluster correlation. Because small sample sizes and low intracluster correlation are common features of cluster-randomized trials, the Kenward-Roger method is the preferred one-stage approach.

Determinants of First-Time Cancer Examinations in a Rural Community: A Mechanism for Behavior Change.

OBJECTIVES: After conducting a media campaign focusing on the importance of oral and pharyngeal cancer (OPC) examinations, we assessed mechanisms of behavior change among individuals receiving an OPC examination for the first time. METHODS: We used data from 2 waves of telephone surveys of individuals residing in 36 rural census tracts in northern Florida (n = 806). The second survey occurred after our media intervention. We developed media messages and modes of message delivery with community members via focus groups and intercept interviews. We performed a mediation analysis to examine behavior change mechanisms. RESULTS: Greater exposure to media messages corresponded with heightened concern about OPC. Heightened concern, in turn, predicted receipt of a first-time OPC examination, but only among men. CONCLUSIONS: We extended earlier studies by measuring an outcome behavior (receipt of an OPC examination) and demonstrating that the putative mechanism of action (concern about the disease) explained the link between a media intervention and engaging in the target behavior. Improving the quality of media campaigns by engaging community stakeholders in selecting messages and delivery methods is an effective strategy in building public health interventions aimed at changing behaviors.

Adaptive designs for comparative effectiveness research trials.

CONTEXT: Medical and health policy decision makers require improved design and analysis methods for comparative effectiveness research (CER) trials. In CER trials, there may be limited information to guide initial design choices. In general settings, adaptive designs (ADs) have effectively overcome limits on initial information. However, CER trials have fundamental differences from standard clinical trials including population heterogeneity and a vaguer concept of a "minimum clinically meaningful difference". OBJECTIVE: To explore the use of a particular form of ADs for comparing treatments within the CER trial context. METHODS: We review the current state of clinical CER, identify areas of CER as particularly strong candidates for application of novel ADs, and illustrate potential usefulness of the designs and methods for two group comparisons. RESULTS: ADs can stabilize power. The designs ensure adequate power for true effects are at least at clinically significant preplanned effect size, or when variability is larger than expected. The designs allow for sample size savings when the true effect is larger or when variability is smaller than planned. CONCLUSION: ADs in CER have great potential to allow trials to successfully and efficiently make important comparisons.