A Principled Approach to Characterize and Analyze Partially Observed Confounder Data from Electronic Health Records.

Objective: Partially observed confounder data pose challenges to the statistical analysis of electronic health records (EHR) and systematic assessments of potentially underlying missingness mechanisms are lacking. We aimed to provide a principled approach to empirically characterize missing data processes and investigate performance of analytic methods. Methods: Three empirical sub-cohorts of diabetic SGLT2 or DPP4-inhibitor initiators with complete information on HbA1c, BMI and smoking as confounders of interest (COI) formed the basis of data simulation under a plasmode framework. A true null treatment effect, including the COI in the outcome generation model, and four missingness mechanisms for the COI were simulated: completely at random (MCAR), at random (MAR), and two not at random (MNAR) mechanisms, where missingness was dependent on an unmeasured confounder and on the value of the COI itself. We evaluated the ability of three groups of diagnostics to differentiate between mechanisms: 1)-differences in characteristics between patients with or without the observed COI (using averaged standardized mean differences [ASMD]), 2)-predictive ability of the missingness indicator based on observed covariates, and 3)-association of the missingness indicator with the outcome. We then compared analytic methods including "complete case", inverse probability weighting, single and multiple imputation in their ability to recover true treatment effects. Results: The diagnostics successfully identified characteristic patterns of simulated missingness mechanisms. For MAR, but not MCAR, the patient characteristics showed substantial differences (median ASMD 0.20 vs 0.05) and consequently, discrimination of the prediction models for missingness was also higher (0.59 vs 0.50). For MNAR, but not MAR or MCAR, missingness was significantly associated with the outcome even in models adjusting for other observed covariates. Comparing analytic methods, multiple imputation using a random forest algorithm resulted in the lowest root-mean-squared-error. Conclusion: Principled diagnostics provided reliable insights into missingness mechanisms. When assumptions allow, multiple imputation with nonparametric models could help reduce bias.

Trust in the Food and Drug Administration: A National Survey Study.

Building trust in public health agencies like the US Food and Drug Administration (FDA) has become a key government priority. Understanding the roots of FDA mistrust is important if the agency is to develop targeted messaging and reforms aimed at building confidence in the agency. We conducted a survey of 2,021 respondents in the US probing attitudes toward the FDA. The primary outcome was FDA trust, defined as the mean score that each respondent assigned to the FDA across four prespecified axes: (1) competence and effectiveness; (2) commitment to acting in the best interests of the American public; (3) abiding by the rules and regulations set forth by policy or law; and (4) expertise in health, science, and medicine. On multivariable ordinal logistic regression, FDA mistrust was associated with female gender (odds ratio [OR] = 0.74, 95% confidence interval [CI] 0.62-0.88), rural community (OR 0.85, 95% CI 0.75-0.96), conservative political views (OR 0.77, 95% CI 0.74-0.81), worse self-reported health (OR 0.89, 95% CI 0.80-0.98), lower satisfaction with health care received (OR 0.63, 95% CI 0.56-0.71), less attention to health and science news (OR 0.72, 95% CI 0.64-0.80), and not having children under the age of 18 (OR 0.72, 95% CI 0.60-0.86). These findings underscore the challenges faced by US political leaders in convincing a heterogeneous American public to trust the FDA. The FDA should develop and deploy targeted outreach strategies to populations with lower levels of trust and strengthen internal processes that minimize biases and ensure sound decision-making.

Changes in Use of Hepatitis C Direct-Acting Antivirals After Access Restrictions Were Eased by State Medicaid Programs.

Importance: Direct-acting antivirals (DAAs) are safe and highly effective for curing hepatitis C virus (HCV) infection, but their high cost led certain state Medicaid programs to impose coverage restrictions. Since 2015, many of these restrictions have been lifted voluntarily in response to advocacy or because of litigation. Objective: To estimate how the prescribing of DAAs to Medicaid patients changed after states eased access restrictions. Design, Setting, and Participants: This modified difference-in-differences analysis of 39 state Medicaid programs included Medicaid beneficiaries who were prescribed a DAA from January 1, 2015, to December 31, 2019. DAA coverage restrictions were measured based on a series of cross-sectional assessments performed from 2014 through 2022 by the US National Viral Hepatitis Roundtable and the Center for Health Law and Policy Innovation. Exposure: Calendar quarter when states eased or eliminated 3 types of DAA coverage restrictions: limiting treatment to patients with severe liver disease, restricting use among patients with active substance use, and requiring prescriptions to be written by or in consultation with specialists. States with none of these restrictions at baseline were excluded. Main Outcomes and Measures: Quarterly number of HCV DAA treatment courses per 100 000 Medicaid beneficiaries. Results: Of 39 states, 7 (18%) eliminated coverage restrictions, 25 (64%) eased restrictions, and 7 (18%) maintained the same restrictions from 2015 to 2019. During this period, the average quarterly use of DAAs increased from 669 to 3601 treatment courses per 100 000 Medicaid beneficiaries. After states eased or eliminated restrictions, the use of DAAs increased by 966 (95% CI, 409-1523) treatment courses per 100 000 Medicaid beneficiaries each quarter compared with states that did not ease or eliminate restrictions. Conclusions and Relevance: The results of this study suggest that there was greater use of DAAs after states relaxed coverage restrictions related to liver disease severity, sobriety, or prescriber specialty. Further reductions or elimination of these rules may improve access to a highly effective public health intervention for patients with HCV.

Patient Out-of-Pocket Costs for Biologic Drugs After Biosimilar Competition.

Importance: Biologic drugs account for a growing share of US pharmaceutical spending. Competition from follow-on biosimilar products (subsequent versions that have no clinically meaningful differences from the original biologic) has led to modest reductions in US health care spending, but these savings may not translate to lower out-of-pocket (OOP) costs for patients. Objective: To investigate whether biosimilar competition is associated with lower OOP spending for patients using biologics. Design, Setting, and Participants: This cohort study used a national commercial claims database (Optum Clinformatics Data Mart) to identify outpatient claims for 1 of 7 clinician-administered biologics (filgrastim, infliximab, pegfilgrastim, epoetin alfa, bevacizumab, rituximab, and trastuzumab) from January 2009 through March 2022. Claims by commercially insured patients younger than 65 years were included. Exposure: Year relative to first biosimilar availability and use of original or biosimilar version. Main Outcomes and Measures: Patients' annual OOP spending on biologics for each calendar year was determined, and OOP spending per claim between reference biologic and biosimilar versions was compared. Two-part regression models assessed for differences in OOP spending, adjusting for patient and clinical characteristics (age, sex, US Census region, health plan type, diagnosis, and place of service) and year relative to initial biosimilar entry. Results: Over 1.7 million claims from 190 364 individuals (median [IQR] age, 53 [42-59] years; 58.3% females) who used at least 1 of the 7 biologics between 2009 and 2022 were included in the analysis. Over 251 566 patient-years of observation, annual OOP costs increased before and after biosimilar availability. Two years after the start of biosimilar competition, the adjusted odds ratio of nonzero annual OOP spending was 1.08 (95% CI, 1.04-1.12; P < .001) and average nonzero annual spending was 12% higher (95% CI, 10%-14%; P < .001) compared with the year before biosimilar competition. After biosimilars became available, claims for biosimilars were more likely than reference biologics to have nonzero OOP costs (adjusted odds ratio, 1.13 [95% CI, 1.11-1.16]; P < .001) but had 8% lower mean nonzero OOP costs (adjusted mean ratio, 0.92 [95% CI, 0.90-0.93; P < .001). Findings varied by drug. Conclusions and Relevance: Findings of this cohort study suggest that biosimilar competition was not consistently associated with lower OOP costs for commercially insured outpatients, highlighting the need for targeted policy interventions to ensure that the savings generated from biosimilar competition translate into increased affordability for patients who need biologics.

Targeted Learning with an Undersmoothed Lasso Propensity Score Model for Large-Scale Covariate Adjustment in Healthcare Database Studies.

Lasso regression is widely used for large-scale propensity score (PS) estimation in healthcare database studies. In these settings, previous work has shown that undersmoothing (overfitting) Lasso PS models can improve confounding control, but it can also cause problems of non-overlap in covariate distributions. It remains unclear how to select the degree of undersmoothing when fitting large-scale Lasso PS models to improve confounding control while avoiding issues that can result from reduced covariate overlap. Here, we used simulations to evaluate the performance of using collaborative-controlled targeted learning to data-adaptively select the degree of undersmoothing when fitting large-scale PS models within both singly and doubly robust frameworks to reduce bias in causal estimators. Simulations showed that collaborative learning can data-adaptively select the degree of undersmoothing to reduce bias in estimated treatment effects. Results further showed that when fitting undersmoothed Lasso PS-models, the use of cross-fitting was important for avoiding non-overlap in covariate distributions and reducing bias in causal estimates.

An open-source implementation of tree-based scan statistics.

PURPOSE: We develop an open-source R package to implement tree-based scan statistics (TBSS) analyses. METHODS: TBSS are data mining methods used by the United States Food and Drug Administration and the Centers for Disease Control. They simultaneously screen thousands of hierarchically aggregated outcomes to identify unsuspected adverse effects of drugs or vaccines, accounting for multiple comparisons. The general structure of TBSS is highly adaptable, with four essential components: (1) a hierarchical outcome structure, (2) a test statistic to be computed for each element of the hierarchy, (3) an algorithm to generate data replicates under a null distribution, and (4) observed outcomes at the lower level of the hierarchy. We encode the general TBSS framework in a convenient R package that offers user-friendly functions for the most used TBSS methods. To illustrate the performance of our software, we evaluated two examples of archetypical TBSS analyses previously analyzed using proprietary, closed-source TreeScan™ software. The first considers the risk of congenital malformations associated with first-trimester exposure to valproate, and the second compares exposure to newly prescribed canagliflozin with a dipeptidyl peptidase 4 inhibitor in adults affected by type 2 diabetes. RESULTS: The results of the original studies are replicated. CONCLUSIONS: The diffusion of an open-source implementation of TBSS can enhance innovation of TBSS methods and foster collaborations. We offer an intuitive R package implementing standard TBSS methods with accompanying tutorials. Our unified object-oriented implementation allows expert users to extend the framework, introduce new features, or enhance existing ones.

How Do Risk Evaluation and Mitigation Strategies Impact Clinical Practice? A National Survey of Physicians.

The US Food and Drug Administration can require risk evaluation and mitigation strategy (REMS) programs for prescription drugs to ensure the benefits of use outweigh the risks. We conducted a national survey of physicians' experiences prescribing eight REMS-covered drugs: (1) ambrisentan; (2) bosentan; (3) clozapine; (4) isotretinoin; (5-7) the multiple myeloma (MM) drugs lenalidomide, pomalidomide, thalidomide; and (8) sodium oxybate. Between May 2022 and January 2023, we surveyed 5,331 physician prescribers of these drugs, and 1,295 (24%) returned surveys (range: 149 for bosentan to 226 for MM drugs). Although 765 (68%) respondents thought the certification process provided useful drug information, 757 (67%) wanted materials to include benefit data and 944 (84%) non-REMS-related risk data. A majority (704, 63%) thought the safe use requirements facilitated discussion with patients, but a similar number (637, 57%) attributed delayed medication access to these requirements. In multivariable modeling, MM drug and isotretinoin respondents were less likely than sodium oxybate respondents to agree that the certification process provided useful drug information (MM drug: odds ratio (OR) = 0.37, 95% confidence interval (CI) = 0.25-0.55; isotretinoin: OR = 0.39, 95% CI = 0.25-0.61), and isotretinoin, clozapine, and bosetan respondents were more likely than sodium oxybate respondents to agree that the safe use requirements often delayed medication access (isotretinoin: OR = 5.83, 95% CI = 3.70-9.19; clozapine: OR = 1.65, 95% CI = 1.08-2.54; bosentan: OR = 1.78, 95% CI = 1.12-2.85). Most physicians believe REMS programs convey useful drug safety information and facilitate discussion with patients but also seek information on benefits and non-REMS-related risks and better integration of REMS processes into clinical workflows.

Testing interventions to reduce clinical inertia in the treatment of hypertension: rationale and design of a pragmatic randomized controlled trial.

BACKGROUND: Clinical inertia, or failure to intensify treatment when indicated, leads to suboptimal blood pressure control. Interventions to overcome inertia and increase antihypertensive prescribing have been modestly successful in part because their effectiveness varies based on characteristics of the provider, the patient, or the provider-patient interaction. Understanding for whom each intervention is most effective could help target interventions and thus increase their impact. METHODS: This three-arm, randomized trial tests the effectiveness of 2 interventions to reduce clinical inertia in hypertension prescribing compared to usual care. Forty five primary care providers (PCPs) caring for patients with hypertension in need of treatment intensification completed baseline surveys that assessed behavioral traits and were randomized to one of three arms: 1) Pharmacist e-consult, in which a clinical pharmacist provided patient-specific recommendations for hypertension medication management to PCPs in advance of upcoming visits, 2) Social norming dashboards that displayed PCP's hypertension control rates compared to those of their peers, or 3) Usual care (no intervention). The primary outcome was the rate of intensification of hypertension treatment. We will compare this outcome between study arms and then evaluate the association between characteristics of providers, patients, their clinical interactions, and intervention responsiveness. RESULTS: Forty-five primary care providers were enrolled and randomized: 16 providers and 173 patients in the social norming dashboards arm, 15 providers and 143 patients in the pharmacist e-consult arm, and 14 providers and 150 patients in the usual care arm. On average, the mean patient age was 64 years, 47% were female, and 73% were white. Baseline demographic and clinical characteristics of patients were similar across arms, with the exception of more Hispanic patients in the usual care arm and fewest in the pharmacist e-consult arm. CONCLUSIONS: This study can help identify interventions to reduce inertia in hypertension care and potentially identify the characteristics of patients, providers, or patient-provider interactions to understand for whom each intervention would be most beneficial. TRIAL REGISTRATION: Clinicaltrials.gov (NCT, Registered: NCT04603560).

Semiparametric Allocation of Subjects to Cohort Strata.

BACKGROUND: We illustrate a method for stratum assignment in small cohort studies that avoids modeling assumptions. METHODS: Off-the-shelf software ( rgenoud ) made stratum assignments to minimize a loss function built on within-stratum and population-adjusted Euclidean distances. RESULTS: In 100 trials using simulated data of 300 records with a binary treatment and four dissimilar covariate treatment predictors, minimizing a loss based on Euclidean distance reduced covariate imbalance by a median of 99%. Stratification by propensity score and weighting records by the inverse of their probability of treatment reduced imbalance by 76%-89% and 83%-94%, respectively. Loss minimization applied to a cohort of 361 children undergoing immunotherapy achieved nearly complete elimination of covariate differences for important treatment predictors. CONCLUSION: With the availability of semiparametric stratum-assignment algorithms, analysts can tailor loss functions to meet design goals. Here, a loss function that emphasized covariate balance performed well under limited testing.

Comparative Effectiveness and Safety of Generic Versus Brand-Name Fluticasone-Salmeterol to Treat Chronic Obstructive Pulmonary Disease.

BACKGROUND: In 2019, the U.S. Food and Drug Administration (FDA) approved the first generic maintenance inhaler for asthma and chronic obstructive pulmonary disease (COPD). The inhaler, Wixela Inhub (fluticasone-salmeterol; Viatris), is a substitutable version of the dry powder inhaler Advair Diskus (fluticasone-salmeterol; GlaxoSmithKline). When approving complex generic products like inhalers, the FDA applies a special "weight-of-evidence" approach. In this case, manufacturers were required to perform a randomized controlled trial in patients with asthma but not COPD, although the product received approval for both indications. OBJECTIVE: To compare the effectiveness and safety of generic (Wixela Inhub) and brand-name (Advair Diskus) fluticasone-salmeterol among patients with COPD treated in routine care. DESIGN: A 1:1 propensity score-matched cohort study. SETTING: A large, longitudinal health care database. PATIENTS: Adults older than 40 years with a diagnosis of COPD. MEASUREMENTS: Incidence of first moderate or severe COPD exacerbation (effectiveness outcome) and incidence of first pneumonia hospitalization (safety outcome) in the 365 days after cohort entry. RESULTS: Among 45 369 patients (27 305 Advair Diskus users and 18 064 Wixela Inhub users), 10 012 matched pairs were identified for the primary analysis. Compared with Advair Diskus use, Wixela Inhub use was associated with a nearly identical incidence of first moderate or severe COPD exacerbation (hazard ratio [HR], 0.97 [95% CI, 0.90 to 1.04]) and first pneumonia hospitalization (HR, 0.99 [CI, 0.86 to 1.15]). LIMITATIONS: Follow-up times were short, reflecting real-world clinical practice. The possibility of residual confounding cannot be completely excluded. CONCLUSION: Use of generic and brand-name fluticasone-salmeterol was associated with similar outcomes among patients with COPD treated in routine practice. PRIMARY FUNDING SOURCE: National Heart, Lung, and Blood Institute.

Association of Advisory Committee Votes With US Food and Drug Administration Decision-Making on Prescription Drugs, 2010-2021.

Importance: The US Food and Drug Administration (FDA) often relies on independent advisory committees when making decisions about the approval of prescription drugs or their withdrawal from the market. These committees provide the FDA with valuable insight and an opportunity to build public trust through transparent deliberation, but recent controversies have raised questions about the optimal use of FDA advisory committees. Objective: To assess the frequency, purposes, and voting outcomes of human drug advisory committees convened from 2010 to 2021 and the FDA's corresponding actions. Design, Setting, and Participants: This qualitative study used a manual review of meeting summaries prepared by FDA staff for the 18 human drug advisory committees active at any time from January 1, 2010, to December 31, 2021, as well as FDA announcements and press releases, drug labels and approval data, industry publications, and company press releases. Main Outcomes and Measures: Outcomes of votes on regulatory questions were recorded using meeting minutes. Alignment of FDA action with advisory votes for new drugs and indications was judged as of 1 year after the vote was held and as of November 30, 2022. Results: The FDA held 409 human drug advisory committee meetings from 2010 to 2021. Committees were convened less frequently over time, from a high of 50 in 2012 to a low of 18 in 2020 and 2021. Much of this decrease occurred at committee meetings involving votes on initial approvals, which declined from a high of 26 in 2012 to a low of 8 in 2021. Overall, FDA regulatory actions aligned with 262 of 298 advisory committee votes on initial approvals, supplemental approvals, withdrawals of approval, and safety actions (88%). Approval followed 142 of 147 positive votes for initial approvals (97%) and 33 of 36 positive votes for supplemental indications (92%), while nonapproval followed 40 of 60 negative votes on initial approvals (67%) and 18 of 21 negative votes on supplemental indications (86%). Conclusions and Relevance: In this qualitative study, there was consistent alignment between advisory votes and FDA action across years and subject areas, but the number of meetings decreased over time. Discordance between FDA actions and advisory committee votes was most frequently an approval after a negative vote. This study demonstrated that these committees have played a key role in the FDA's decision-making process but that the FDA sought independent expert advice less frequently over time even as it continued to follow it. The role of advisory committees in the current regulatory landscape should be more clearly and publicly defined.

Establishing a natural history of X-linked dystonia parkinsonism.

X-linked dystonia parkinsonism is a neurodegenerative movement disorder that affects men whose mothers originate from the island of Panay, Philippines. Current evidence indicates that the most likely cause is an expansion in the TAF1 gene that may be amenable to treatment. To prepare for clinical trials of therapeutic candidates for X-linked dystonia parkinsonism, we focused on the identification of quantitative phenotypic measures that are most strongly associated with disease progression. Our main objective is to establish a comprehensive, quantitative assessment of movement dysfunction and bulbar motor impairments that are sensitive and specific to disease progression in persons with X-linked dystonia parkinsonism. These measures will set the stage for future treatment trials. We enrolled patients with X-linked dystonia parkinsonism and performed a comprehensive oromotor, speech and neurological assessment. Measurements included patient-reported questionnaires regarding daily living activities and both neurologist-rated movement scales and objective quantitative measures of bulbar function and nutritional status. Patients were followed for 18 months from the date of enrollment and evaluated every 6 months during that period. We analysed a total of 87 men: 29 were gene-positive and had symptoms at enrollment, seven were gene-positive and had no symptoms at enrollment and 51 were gene-negative. We identified measures that displayed a significant change over the study. We used principal variables analysis to identify a minimal battery of 21 measures that explains 67.3% of the variance over the course of the study. These measures included patient-reported, clinician-rated and objective quantitative outcomes that may serve as endpoints in future clinical trials.

Inference in response-adaptive clinical trials when the enrolled population varies over time.

A common assumption of data analysis in clinical trials is that the patient population, as well as treatment effects, do not vary during the course of the study. However, when trials enroll patients over several years, this hypothesis may be violated. Ignoring variations of the outcome distributions over time, under the control and experimental treatments, can lead to biased treatment effect estimates and poor control of false positive results. We propose and compare two procedures that account for possible variations of the outcome distributions over time, to correct treatment effect estimates, and to control type-I error rates. The first procedure models trends of patient outcomes with splines. The second leverages conditional inference principles, which have been introduced to analyze randomized trials when patient prognostic profiles are unbalanced across arms. These two procedures are applicable in response-adaptive clinical trials. We illustrate the consequences of trends in the outcome distributions in response-adaptive designs and in platform trials, and investigate the proposed methods in the analysis of a glioblastoma study.

Patient Out-of-Pocket Costs Following the Availability of Biosimilar Versions of Infliximab.

After market exclusivity ends for biologic drugs, biosimilars-follow-on versions made by other manufacturers-can compete with lower prices. Biosimilars have modestly reduced prescription drug spending for US payers, but it is unclear whether patients have directly experienced any savings. In this study we assessed whether availability of biosimilar infliximab was associated with lower out-of-pocket (OOP) costs, using claims from a national data set of commercially insured patients from 2014 to 2018. We used two-part models, adjusting for patient demographics, clinical characteristics, insurance plan type, and calendar month. Compared with the reference biologic, there was no difference in the percentage of biosimilar claims with OOP costs (30.1% vs. 30.8%; adjusted odds ratio (aOR) 0.98, 95% confidence interval (CI), 0.84-1.15, P = 0.84) or the average nonzero OOP cost (median $378 vs. $538, adjusted mean ratio (aMR) 0.97, 95% CI, 0.80-1.18, P = 0.77). The percentage of claims with OOP costs was lower after biosimilar competition (30.7% vs. 35.0%, aOR 0.96, 95% CI, 0.94-0.99, P = 0.003), but average nonzero costs increased (median $534 vs. $520, aMR 1.04, 95% CI, 1.01-1.07, P = 0.004). Thus, early biosimilar infliximab competition did not improve affordability for patients. Policymakers need to better assure that competition in the biosimilar market translates to lower costs for patients using these medications.

Designing and conducting adaptive trials to evaluate interventions in health services and implementation research: practical considerations.

Randomized controlled clinical trials are widely considered the gold standard for evaluating the efficacy or effectiveness of interventions in health care. Adaptive trials incorporate changes as the study proceeds, such as modifying allocation probabilities or eliminating treatment arms that are likely to be ineffective. These designs have been widely used in drug discovery studies but can also be useful in health services and implementation research and have been minimally used. As motivating examples, we use an ongoing adaptive trial and two completed parallel group studies and highlight potential advantages, disadvantages, and important considerations when using adaptive trial designs in health services and implementation research. In addition, we investigate the impact on power and the study duration if the two completed parallel-group trials had instead been conducted using adaptive principles. Compared with traditional trial designs, adaptive designs can often allow one to evaluate more interventions, adjust participant allocation probabilities (e.g., to achieve covariate balance), and identify participants who are likely to agree to enroll. These features could reduce resources needed to conduct a trial. However, adaptive trials have potential disadvantages and practical aspects that need to be considered, most notably outcomes that can be rapidly measured and extracted (e.g., long-term outcomes that take significant time to measure from data sources can be challenging), minimal missing data, and time trends. In conclusion, adaptive designs are a promising approach to help identify how best to implement evidence-based interventions into real-world practice in health services and implementation research.

Dynamic modeling of the Italians' attitude towards Covid-19.

We analyze repeated cross-sectional survey data collected by the Institute of Global Health Innovation, to characterize the perception and behavior of the Italian population during the Covid-19 pandemic, focusing on the period that spans from April 2020 to July 2021. To accomplish this goal, we propose a Bayesian dynamic latent-class regression model, that accounts for the effect of sampling bias including survey weights into the likelihood function. According to the proposed approach, attitudes towards covid-19 are described via ideal behaviors that are fixed over time, corresponding to different degrees of compliance with spread-preventive measures. The overall tendency toward a specific profile dynamically changes across survey waves via a latent Gaussian process regression, that adjusts for subject-specific covariates. We illustrate the evolution of Italians' behaviors during the pandemic, providing insights on how the proportion of ideal behaviors has varied during the phases of the lockdown, while measuring the effect of age, sex, region and employment of the respondents on the attitude toward covid-19.

MULTIVARIATE MIXED MEMBERSHIP MODELING: INFERRING DOMAIN-SPECIFIC RISK PROFILES.

Characterizing the shared memberships of individuals in a classification scheme poses severe interpretability issues, even when using a moderate number of classes (say 4). Mixed membership models quantify this phenomenon, but they typically focus on goodness-of-fit more than on interpretable inference. To achieve a good numerical fit, these models may in fact require many extreme profiles, making the results difficult to interpret. We introduce a new class of multivariate mixed membership models that, when variables can be partitioned into subject-matter based domains, can provide a good fit to the data using fewer profiles than standard formulations. The proposed model explicitly accounts for the blocks of variables corresponding to the distinct domains along with a cross-domain correlation structure, which provides new information about shared membership of individuals in a complex classification scheme. We specify a multivariate logistic normal distribution for the membership vectors, which allows easy introduction of auxiliary information leveraging a latent multivariate logistic regression. A Bayesian approach to inference, relying on Pólya gamma data augmentation, facilitates efficient posterior computation via Markov Chain Monte Carlo. We apply this methodology to a spatially explicit study of malaria risk over time on the Brazilian Amazon frontier.

Learning in Medicine: The Importance of Statistical Thinking.

In many fields, including medicine and biology, there has been in the last years an increasing diffusion and availability of complex data from different sources. Examples include biological experiments or data from health care providers. These data encompass information that can potentially enhance therapeutic advancement and constitute the core of modern system medicine. When analyzing these complex data, it is important to appropriately quantify uncertainty, avoiding using only algorithmic and automated approaches, which are not always appropriate. Improper application of algorithmic approaches, which ignore domain knowledge, could result in filling the literature with imprecise and/or misleading conclusions. In this chapter, we highlight the importance of statistical thinking when leveraging complex data and models to enhance science progress. In particular, we discuss the reproducibility and replicability issues, the importance of uncertainty quantification, and some common pitfalls in the analysis of big data.