Leveraging information from secondary endpoints to enhance dynamic borrowing across subpopulations.

Randomized trials seek efficient treatment effect estimation within target populations, yet scientific interest often also centers on subpopulations. Although there are typically too few subjects within each subpopulation to efficiently estimate these subpopulation treatment effects, one can gain precision by borrowing strength across subpopulations, as is the case in a basket trial. While dynamic borrowing has been proposed as an efficient approach to estimating subpopulation treatment effects on primary endpoints, additional efficiency could be gained by leveraging the information found in secondary endpoints. We propose a multisource exchangeability model (MEM) that incorporates secondary endpoints to more efficiently assess subpopulation exchangeability. Across simulation studies, our proposed model almost uniformly reduces the mean squared error when compared to the standard MEM that only considers data from the primary endpoint by gaining efficiency when subpopulations respond similarly to the treatment and reducing the magnitude of bias when the subpopulations are heterogeneous. We illustrate our model's feasibility using data from a recently completed trial of very low nicotine content cigarettes to estimate the effect on abstinence from smoking within three priority subpopulations. Our proposed model led to increases in the effective sample size two to four times greater than under the standard MEM.

Commentary on Chen et al. (2022): The need for continued methodological research on leveraging information in secondary endpoints for more efficient RCTs.

Chen et al. (2022) recently proposed a set of estimating equations that incorporate data from secondary endpoints to improve precision in parameter estimates related to a primary endpoint. We were motivated to translate their methodology to the context of randomized controlled trials to gain precision in treatment effect estimation using data from secondary endpoints. Our results suggest that this estimator cannot gain efficiency in this context because of random treatment assignment, especially when there is a treatment effect on secondary endpoints, and that further methodological work in this area is needed.

Racial and ethnic differences in unmet needs among older adults receiving publicly-funded home and community-based services.

BACKGROUND: Unmet need for home and community-based services (HCBS) may disparately impact older adults from racial and ethnic minoritized groups. We examined racial and ethnic differences in unmet need for HCBS among consumers ≥65 years using publicly funded HCBS. METHODS: We analyzed the National Core Indicators-Aging and Disability survey data (2015-2019) from 21,739 community-dwelling HCBS consumers aged ≥65 years in 23 participating states. Outcome measures included self-reported unmet need in six service types (i.e., personal care, homemaker/chore, delivered meals, adult day services, transportation, and caregiver support). Racial and ethnic groups included non-Hispanic Black, Asian, non-Hispanic White, Hispanic, and multiracial groups. Logistic regression models examined associations between race and ethnicity and unmet need, adjusting for sociodemographic, health, and HCBS program (i.e., Medicaid, Older Americans Act [OAA], Program for All-Inclusive Care for the Elderly [PACE]) characteristics, and use of specific service types. RESULTS: Among 21,739 respondents, 23.3% were Black, 3.4% were Asian, 10.8% were Hispanic, 58.8% were non-Hispanic White, and 3.7% were multiracial or identified with other races/ethnicities. Asian and Black consumers had higher odds of reporting unmet need in personal care than White consumers (adjusted odds ratio [aOR], 1.45, p value < 0.01; and aOR, 1.25, p < 0.001, respectively). Asian and Black consumers had significantly higher odds of unmet need in adult day services versus White consumers (aOR, 1.94, p < 0.001 and aOR, 1.39, p < 0.001, respectively). Black consumers had higher odds of unmet need versus non-Hispanic White consumers in meal delivery and caregiver support services (aOR, 1.29; p < 0.01; and aOR 1.26, p < 0.05, respectively). Race and ethnicity were not significantly associated with experiencing unmet need for homemaker/chore or transportation services. CONCLUSIONS: Future research should identify driving forces in disparities in unmet need to develop culturally appropriate solutions.

Reported Unmet Hearing Aid Need in Older People With Dementia: The US National Core Indicators Survey.

OBJECTIVES: Hearing aids have important health benefits for older adults with Alzheimer disease and related dementias (ADRD); however, hearing aid adoption in this group is low. This study aimed to determine where to target hearing aid interventions for American long-term care recipients with ADRD by examining the association of ADRD and residence type with respondent-reported unmet hearing aid need. DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study used data from the United States National Core Indicators-Aging and Disabilities survey (2015-2019) for long-term care recipients aged ≥65 years. METHODS: We used multivariable logistic regression to model the likelihood of reporting unmet hearing aid need conditional on ADRD status and residence type (own/family house or apartment, residential care, or nursing facility/home), adjusting for sociodemographic factors and response type (self vs proxy). RESULTS: Of the 25,492 respondents [median (IQR) age, 77 (71, 84) years; 7074 (27.8%) male], 5442 (21.4%) had ADRD and 3659 (14.4%) owned hearing aids. Residence types were 17,004 (66.8%) own/family house or apartment, 4966 (19.5%) residential care, and 3522 (13.8%) nursing home. Among non-hearing aid owners, ADRD [adjusted odds ratio (AOR) 0.90, 95% CI 0.80-1.0] and residence type were associated with respondent-reported unmet hearing aid need. Compared to the nursing home reference group, respondents in their own/family home (AOR 1.85, 95% CI 1.61-2.13) and residential care (AOR 1.30, 95% CI 1.10-1.53) were more likely to report unmet hearing aid need. This pattern was significantly more pronounced in people with ADRD than in those without, stemming from an interaction between ADRD and residence type. CONCLUSIONS AND IMPLICATIONS: American long-term care recipients with ADRD living in their own/family home are more likely to report unmet hearing aid need than those with ADRD in institutional and congregate settings. This information can inform the design and delivery of hearing interventions for older adults with ADRD.

The Minnesota Medical Operations Coordination Center: A COVID-19 Statewide Response to Ensure Access to Critical Care and Medical-Surgical Beds.

BACKGROUND: COVID-19 led to unprecedented inpatient capacity challenges, particularly in ICUs, which spurred development of statewide or regional placement centers for coordinating transfer (load-balancing) of adult patients needing intensive care to hospitals with remaining capacity. RESEARCH QUESTION: Do Medical Operations Coordination Centers (MOCC) augment patient placement during times of severe capacity challenges? STUDY DESIGN AND METHODS: The Minnesota MOCC was established with a focus on transfer of adult ICU and medical-surgical patients; trauma, cardiac, stroke, burn, and extracorporeal membrane oxygenation cases were excluded. The center operated within one health care system's bed management center, using a dedicated 24/7 telephone number. Major health care systems statewide and two tertiary centers in a neighboring state participated, sharing information on system status, challenges, and strategies. Patient volumes and transfer data were tracked; client satisfaction was evaluated through an anonymous survey. RESULTS: From August 1, 2020, through March 31, 2022, a total of 5,307 requests were made, 2,008 beds identified, 1,316 requests canceled, and 1,981 requests were unable to be fulfilled. A total of 1,715 patients had COVID-19 (32.3%), and 2,473 were negative or low risk for COVID-19 (46.6%). COVID-19 status was unknown in 1,119 (21.1%). Overall, 760 were patients on ventilators (49.1% COVID-19 positive). The Minnesota Critical Care Coordination Center placed most patients during the fall 2020 surge with the Minnesota Governor's stay-at-home order during the peak. However, during the fall 2021 surge, only 30% of ICU patients and 39% of medical-surgical patients were placed. Indicators characterizing severe surge include the number of Critical Care Coordination Center requests, decreasing placements, longer placement times, and time series analysis showing significant request-acceptance differences. INTERPRETATION: Implementation of a large-scale Minnesota MOCC program was effective at placing patients during the first COVID-19 pandemic fall 2020 surge and was well regarded by hospitals and health systems. However, under worsening duress of limited resources during the fall 2021 surge, placement of ICU and medical-surgical patients was greatly decreased.

Implementation and Outcomes of an Empiric Ivermectin Strongyloides Treatment Protocol for Patients Receiving High-Dose Corticosteroids for Severe COVID-19.

Strongyloides stercoralis is a parasitic roundworm that is present worldwide and can cause lifelong, often asymptomatic, infection. Immunosuppression, particularly by corticosteroids, is a risk factor for hyperinfection syndrome and disseminated strongyloidiasis-severe disease states that can lead to septic shock and death. Our institution implemented a strongyloidiasis screening and empiric ivermectin treatment protocol for inpatients receiving high-dose corticosteroids for severe COVID-19. Among 487 COVID-19 admissions treated with high-dose corticosteroids from June 10, 2020 to March 31, 2021, 61% of those with demographics at risk for Strongyloides exposure were screened for Strongyloides and treated empirically with ivermectin. Adherence to the protocol declined over time during the study period. The empiric ivermectin protocol appeared safe, but more research is needed to determine the effect on hyperinfection and/or disseminated strongyloidiasis risk and mortality rate, as well as to improve institutional adherence to the protocol.

Practical guidance on modeling choices for the virtual twins method.

Individuals can vary drastically in their response to the same treatment, and this heterogeneity has driven the push for more personalized medicine. Accurate and interpretable methods to identify subgroups that respond to the treatment differently from the population average are necessary to achieving this goal. The Virtual Twins (VT) method is a highly cited and implemented method for subgroup identification because of its intuitive framework. However, since its initial publication, many researchers still rely heavily on the authors' initial modeling suggestions without examining newer and more powerful alternatives. This leaves much of the potential of the method untapped. We comprehensively evaluate the performance of VT with different combinations of methods in each of its component steps, under a collection of linear and nonlinear problem settings. Our simulations show that the method choice for Step 1 of VT, in which dense models with high predictive performance are fit for the potential outcomes, is highly influential in the overall accuracy of the method, and Superlearner is a promising choice. We illustrate our findings by using VT to identify subgroups with heterogeneous treatment effects in a randomized, double-blind trial of very low nicotine content cigarettes.

Multimarker omnibus tests by leveraging individual marker summary statistics from large biobanks.

As biobanks become increasingly popular, access to genotypic and phenotypic data continues to increase in the form of precomputed summary statistics (PCSS). Widespread accessibility of PCSS alleviates many issues related to biobank data, including that of data privacy and confidentiality, as well as high computational costs. However, questions remain about how to maximally leverage PCSS for downstream statistical analyses. Here we present a novel method for testing the association of an arbitrary number of single nucleotide variants (SNVs) on a linear combination of phenotypes after adjusting for covariates for common multimarker tests (e.g., SKAT, SKAT-O) without access to individual patient-level data (IPD). We validate exact formulas for each method, and demonstrate their accuracy through simulation studies and an application to fatty acid phenotypic data from the Framingham Heart Study.

A permutation procedure to detect heterogeneous treatment effects in randomized clinical trials while controlling the type I error rate.

BACKGROUND/AIMS: Secondary analyses of randomized clinical trials often seek to identify subgroups with differential treatment effects. These discoveries can help guide individual treatment decisions based on patient characteristics and identify populations for which additional treatments are needed. Traditional analyses require researchers to pre-specify potential subgroups to reduce the risk of reporting spurious results. There is a need for methods that can detect such subgroups without a priori specification while allowing researchers to control the probability of falsely detecting heterogeneous subgroups when treatment effects are uniform across the study population. METHODS: We propose a permutation procedure for tuning parameter selection that allows for type I error control when testing for heterogeneous treatment effects framed within the Virtual Twins procedure for subgroup identification. We verify that the type I error rate can be controlled at the nominal rate and investigate the power for detecting heterogeneous effects when present through extensive simulation studies. We apply our method to a secondary analysis of data from a randomized trial of very low nicotine content cigarettes. RESULTS: In the absence of type I error control, the observed type I error rate for Virtual Twins was between 99% and 100%. In contrast, models tuned via the proposed permutation were able to control the type I error rate and detect heterogeneous effects when present. An application of our approach to a recently completed trial of very low nicotine content cigarettes identified several variables with potentially heterogeneous treatment effects. CONCLUSIONS: The proposed permutation procedure allows researchers to engage in secondary analyses of clinical trials for treatment effect heterogeneity while maintaining the type I error rate without pre-specifying subgroups.

Using Summary Statistics to Model Multiplicative Combinations of Initially Analyzed Phenotypes With a Flexible Choice of Covariates.

While the promise of electronic medical record and biobank data is large, major questions remain about patient privacy, computational hurdles, and data access. One promising area of recent development is pre-computing non-individually identifiable summary statistics to be made publicly available for exploration and downstream analysis. In this manuscript we demonstrate how to utilize pre-computed linear association statistics between individual genetic variants and phenotypes to infer genetic relationships between products of phenotypes (e.g., ratios; logical combinations of binary phenotypes using "and" and "or") with customized covariate choices. We propose a method to approximate covariate adjusted linear models for products and logical combinations of phenotypes using only pre-computed summary statistics. We evaluate our method's accuracy through several simulation studies and an application modeling ratios of fatty acids using data from the Framingham Heart Study. These studies show consistent ability to recapitulate analysis results performed on individual level data including maintenance of the Type I error rate, power, and effect size estimates. An implementation of this proposed method is available in the publicly available R package pcsstools.

Computationally efficient, exact, covariate-adjusted genetic principal component analysis by leveraging individual marker summary statistics from large biobanks.

The popularization of biobanks provides an unprecedented amount of genetic and phenotypic information that can be used to research the relationship between genetics and human health. Despite the opportunities these datasets provide, they also pose many problems associated with computational time and costs, data size and transfer, and privacy and security. The publishing of summary statistics from these biobanks, and the use of them in a variety of downstream statistical analyses, alleviates many of these logistical problems. However, major questions remain about how to use summary statistics in all but the simplest downstream applications. Here, we present a novel approach to utilize basic summary statistics (estimates from single marker regressions on single phenotypes) to evaluate more complex phenotypes using multivariate methods. In particular, we present a covariate-adjusted method for conducting principal component analysis (PCA) utilizing only biobank summary statistics. We validate exact formulas for this method, as well as provide a framework of estimation when specific summary statistics are not available, through simulation. We apply our method to a real data set of fatty acid and genomic data.