Tailored vs. General COVID-19 prevention for adults with mental disabilities residing in group homes: a randomized controlled effectiveness-implementation trial.

BACKGROUND: People with serious mental illness (SMI) and people with intellectual disabilities/developmental disabilities (ID/DD) are at higher risk for COVID-19 and more severe outcomes. We compare a tailored versus general best practice COVID-19 prevention program in group homes (GHs) for people with SMI or ID/DD in Massachusetts (MA). METHODS: A hybrid effectiveness-implementation cluster randomized control trial compared a four-component implementation strategy (Tailored Best Practices: TBP) to dissemination of standard prevention guidelines (General Best-Practices: GBP) in GHs across six MA behavioral health agencies. GBP consisted of standard best practices for preventing COVID-19. TBP included GBP plus four components including: (1) trusted-messenger peer testimonials on benefits of vaccination; (2) motivational interviewing; (3) interactive education on preventive practices; and (4) fidelity feedback dashboards for GHs. Primary implementation outcomes were full COVID-19 vaccination rates (baseline: 1/1/2021-3/31/2021) and fidelity scores (baseline: 5/1/21-7/30/21), at 3-month intervals to 15-month follow-up until October 2022. The primary effectiveness outcome was COVID-19 infection (baseline: 1/1/2021-3/31/2021), measured every 3 months to 15-month follow-up. Cumulative incidence of vaccinations were estimated using Kaplan-Meier curves. Cox frailty models evaluate differences in vaccination uptake and secondary outcomes. Linear mixed models (LMMs) and Poisson generalized linear mixed models (GLMMs) were used to evaluate differences in fidelity scores and incidence of COVID-19 infections. RESULTS: GHs (n=415) were randomized to TBP (n=208) and GBP (n=207) including 3,836 residents (1,041 ID/DD; 2,795 SMI) and 5,538 staff. No differences were found in fidelity scores or COVID-19 incidence rates between TBP and GBP, however TBP had greater acceptability, appropriateness, and feasibility. No overall differences in vaccination rates were found between TBP and GBP. However, among unvaccinated group home residents with mental disabilities, non-White residents achieved full vaccination status at double the rate for TBP (28.6%) compared to GBP (14.4%) at 15 months. Additionally, the impact of TBP on vaccine uptake was over two-times greater for non-White residents compared to non-Hispanic White residents (ratio of HR for TBP between non-White and non-Hispanic White: 2.28, p = 0.03). CONCLUSION: Tailored COVID-19 prevention strategies are beneficial as a feasible and acceptable implementation strategy with the potential to reduce disparities in vaccine acceptance among the subgroup of non-White individuals with mental disabilities. TRIAL REGISTRATION: ClinicalTrials.gov, NCT04726371, 27/01/2021. https://clinicaltrials.gov/study/NCT04726371 .

Predictors of COVID-19 infection and hospitalization in group homes for individuals with intellectual and/or developmental disabilities.

BACKGROUND: More than seven million people with intellectual and/or developmental disabilities (ID/DD) live in the US and may face an elevated risk for COVID-19. OBJECTIVE: To identify correlates of COVID-19 and related hospitalizations among people with ID/DD in group homes in Massachusetts. METHODS: We collected data during March 1, 2020-June 30, 2020 (wave 1) and July 1, 2020-March 31, 2021 (wave 2) from the Massachusetts Department of Public Health and six organizations administering 206 group homes for 1035 residents with ID/DD. The main outcomes were COVID-19 infections and related hospitalizations. We fit multilevel Cox proportional hazards models to estimate associations with observed predictors and assess contextual home- and organizational-level effects. RESULTS: Compared with Massachusetts residents, group home residents had a higher age-adjusted rate of COVID-19 in wave 1 (incidence rate ratio [IRR], 12.06; 95 % confidence interval [CI], 10.51-13.84) and wave 2 (IRR, 2.47; 95 % CI, 2.12-2.88) and a higher age-adjusted rate of COVID-19 hospitalizations in wave 1 (IRR, 17.64; 95 % CI, 12.59-24.70) and wave 2 (IRR, 4.95; 95 % CI, 3.23-7.60). COVID-19 infections and hospitalizations were more likely among residents aged 65+ and in group homes with 6+ resident beds and recent infection among staff and residents. CONCLUSIONS: Aggressive efforts to decrease resident density, staff-to-resident ratios, and staff infections through efforts such as vaccination, in addition to ongoing access to personal protective equipment and COVID-19 testing, may reduce COVID-19 and related hospitalizations in people with ID/DD living in group homes.

COVID-19-Related Outcomes Among Group Home Residents with Serious Mental Illness in Massachusetts in the First Year of the Pandemic.

This study examined COVID-19 infection and hospitalizations among people with serious mental illness who resided in residential care group homes in Massachusetts during the first year of the COVID-19 pandemic. The authors analyzed data on 2261 group home residents and COVID-19 data from the Massachusetts Department of Public Health. Outcomes included positive COVID-19 tests and COVID-19 hospitalizations March 1, 2020-June 30, 2020 (wave 1) and July 1, 2020-March 31, 2021 (wave 2). Associations between hazard of outcomes and resident and group home characteristics were estimated using multi-level Cox frailty models including home- and city-level frailties. Between March 2020 and March 2021, 182 (8%) residents tested positive for COVID-19, and 51 (2%) had a COVID-19 hospitalization. Compared with the Massachusetts population, group home residents had age-adjusted rate ratios of 3.0 (4.86 vs. 1.60 per 100) for COVID infection and 13.5 (1.99 vs. 0.15 per 100) for COVID hospitalizations during wave 1; during wave 2, the rate ratios were 0.5 (4.55 vs. 8.48 per 100) and 1.7 (0.69 vs. 0.40 per 100). In Cox models, residents in homes with more beds, higher staff-to-resident ratios, recent infections among staff and other residents, and in cities with high community transmission risk had greater hazard of COVID-19 infection. Policies and interventions that target group home-specific risks are needed to mitigate adverse communicable disease outcomes in this population.Clinical Trial Registration Number This study provides baseline (i.e., pre-randomization) data from a clinical trial study NCT04726371.

Group Home Staff Experiences With Work and Health in the COVID-19 Pandemic in Massachusetts.

Importance: Direct reports of the experiences of staff working in group homes for people with serious mental illness (SMI) and/or intellectual or developmental disabilities (ID/DD) are rarely reported. Hearing from workers about their experiences in the COVID-19 pandemic may inform future workforce and public policy. Objective: To gather baseline data on worker experience with the perceived effects of COVID-19 on health and work in the pandemic prior to initiating an intervention to mitigate the spread of COVID-19 and to measure differences in worker experience by gender, race, ethnicity, education, and resident population served (persons with SMI and/or IDD/DD). Design, Setting, and Participants: This mixed-mode, cross-sectional survey study was conducted using online then paper-based self-administration from May to September 2021 at the end of the first year of the pandemic. Staff working in 415 group homes that provided care within 6 Massachusetts organizations serving adults aged 18 years or older with SMI and/or ID/DD were surveyed. The eligible survey population included a census of staff who were currently employed in participating group homes during the study period. A total of 1468 staff completed or partially completed surveys. The overall survey response rate was 44% (range by organization, 20% to 52%). Main Outcomes and Measures: Self-reported experiential outcomes were measured in work, health, and vaccine completion. Bivariate and multivariate analyses explore experiences by gender, race, ethnicity, education, trust in experts and employers, and population served. Results: The study population included 1468 group home staff (864 [58.9%] women; 818 [55.7%] non-Hispanic Black; 98 [6.7%] Hispanic or Latino). A total of 331 (22.5%) group home staff members reported very serious perceived effects on health; 438 (29.8%) reported very serious perceived effects on mental health; 471 (32.1%) reported very serious perceived effects on health of family and friends; and 414 reported very serious perceived effects (28.2%) on access to health services, with statistically significant differences observed by race and ethnicity. Vaccine acceptance was higher among persons with higher educational attainment and trust in scientific expertise and lower among persons who self-reported as Black or Hispanic/Latino. A total of 392 (26.7%) respondents reported needing support for health needs, and 290 (19.8%) respondents reported needing support for loneliness or isolation. Conclusions and Relevance: In this survey study, approximately one-third of group home workers reported serious personal health and access to health care barriers during the first year of the COVID-19 pandemic in Massachusetts. Addressing unmet health needs and access to health and mental health services, including inequities and disparities by race, ethnicity, and education, should benefit staff health and safety, as well as that of the individuals with disabilities who rely on them for support and care.

Best practices to reduce COVID-19 in group homes for individuals with serious mental illness and intellectual and developmental disabilities: Protocol for a hybrid type 1 effectiveness-implementation cluster randomized trial.

BACKGROUND: People with serious mental illness (SMI) and intellectual disabilities and/or developmental disabilities (ID/DD) living in group homes (GHs) and residential staff are at higher risk for COVID-19 infection, hospitalization, and death compared with the general population. METHODS: We describe a hybrid type 1 effectiveness-implementation cluster randomized trial to assess evidence-based infection prevention practices to prevent COVID-19 for residents with SMI or ID/DD and the staff in GHs. The trial will use a cluster randomized design in 400 state-funded GHs in Massachusetts for adults with SMI or ID/DD to compare effectiveness and implementation of "Tailored Best Practices" (TBP) consisting of evidence-based COVID-19 infection prevention practices adapted for residents with SMI and ID/DD and GH staff; to "General Best Practices" (GBP), consisting of required standard of care reflecting state and federal standard general guidelines for COVID-19 prevention in GHs. External (i.e., community-based research staff) and internal (i.e., GH staff leadership) personnel will facilitate implementation of TBP. The primary effectiveness outcome is incident SARS-CoV-2 infection and secondary effectiveness outcomes include COVID-19-related hospitalizations and mortality in GHs. The primary implementation outcomes are fidelity to TBP and rates of COVID-19 vaccination. Secondary implementation outcomes are adoption, adaptation, reach, and maintenance. Outcomes will be assessed at baseline, 3-, 6-, 9-, 12-, and 15-months post-randomization. CONCLUSIONS: This study will advance knowledge on comparative effectiveness and implementation of two different strategies to prevent COVID-19-related infection, morbidity, and mortality and promote fidelity and adoption of these interventions in high-risk GHs for residents with SMI or ID/DD and staff. CLINICAL TRIAL REGISTRATION NUMBER: NCT04726371.

Simulation-assisted machine learning.

MOTIVATION: In a predictive modeling setting, if sufficient details of the system behavior are known, one can build and use a simulation for making predictions. When sufficient system details are not known, one typically turns to machine learning, which builds a black-box model of the system using a large dataset of input sample features and outputs. We consider a setting which is between these two extremes: some details of the system mechanics are known but not enough for creating simulations that can be used to make high quality predictions. In this context we propose using approximate simulations to build a kernel for use in kernelized machine learning methods, such as support vector machines. The results of multiple simulations (under various uncertainty scenarios) are used to compute similarity measures between every pair of samples: sample pairs are given a high similarity score if they behave similarly under a wide range of simulation parameters. These similarity values, rather than the original high dimensional feature data, are used to build the kernel. RESULTS: We demonstrate and explore the simulation-based kernel (SimKern) concept using four synthetic complex systems-three biologically inspired models and one network flow optimization model. We show that, when the number of training samples is small compared to the number of features, the SimKern approach dominates over no-prior-knowledge methods. This approach should be applicable in all disciplines where predictive models are sought and informative yet approximate simulations are available. AVAILABILITY AND IMPLEMENTATION: The Python SimKern software, the demonstration models (in MATLAB, R), and the datasets are available at https://github.com/davidcraft/SimKern. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The value of prior knowledge in machine learning of complex network systems.

MOTIVATION: Our overall goal is to develop machine-learning approaches based on genomics and other relevant accessible information for use in predicting how a patient will respond to a given proposed drug or treatment. Given the complexity of this problem, we begin by developing, testing and analyzing learning methods using data from simulated systems, which allows us access to a known ground truth. We examine the benefits of using prior system knowledge and investigate how learning accuracy depends on various system parameters as well as the amount of training data available. RESULTS: The simulations are based on Boolean networks-directed graphs with 0/1 node states and logical node update rules-which are the simplest computational systems that can mimic the dynamic behavior of cellular systems. Boolean networks can be generated and simulated at scale, have complex yet cyclical dynamics and as such provide a useful framework for developing machine-learning algorithms for modular and hierarchical networks such as biological systems in general and cancer in particular. We demonstrate that utilizing prior knowledge (in the form of network connectivity information), without detailed state equations, greatly increases the power of machine-learning algorithms to predict network steady-state node values ('phenotypes') and perturbation responses ('drug effects'). AVAILABILITY AND IMPLEMENTATION: Links to codes and datasets here: https://gray.mgh.harvard.edu/people-directory/71-david-craft-phd. CONTACT: dcraft@broadinstitute.org. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.