Search

BackgroundAdverse maternal outcomes and multimorbidity affect a growing proportion of pregnant individuals in the United States and with racial and ethnic disparities in the rates of multimorbidity. There is limited research on multimorbidity, measured as the co-occurrence of two or more chronic conditions, and how it impacts adverse maternal outcomes.ObjectivesWe sought to: (1) determine how pre-pregnancy conditions and their combined effects contribute to racial disparities in adverse maternal outcomes; and (2) incorporate conditions and their combined effects into predictive modeling algorithms to improve risk prediction using information available in the first trimester, specifically for racial minority subgroups.Study DesignWe used data from the “Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-be (nuMoM2b)” observational cohort study. We defined multimorbidity as the co-occurrence of two or more pre-pregnancy conditions. The primary outcomes of interest were severe preeclampsia, postpartum readmission, and blood transfusion during pregnancy or up to 14 days postpartum. We used weighted Poisson regression with robust variance to estimate adjusted risk ratios and 95% confidence intervals, and we used mediation analysis to evaluate the contribution of the combined effects of pre-pregnancy condition to racial/ethnic disparities. We also evaluated the performance of our models by racial subgroup using the area under the receiver operating characteristic curve (AUC) metric.ResultsIn the nuMoM2b cohort (n=8729), 3.6% (n=318) experienced severe preeclampsia, 1.8% (n=157) experienced blood transfusion, and 1.8% (n=154) experienced postpartum readmission. 22.8% of all participants experienced two or more co-occurring pre-pregnancy conditions. Cardiovascular conditions were associated with an increased relative risk for severe preeclampsia (aRR, 1.71; CI, 1.59-1.84), and this risk was additionally exacerbated when hematologic conditions were co-occurring with cardiovascular conditions (aRR, 1.34; CI, 1.12-1.60). The mediation analysis results were not statistically significant; however, cardiovascular conditions explained 36.6% of non-Hispanic Black individuals’ association with increased risk for severe preeclampsia (p=0.07). The addition of pre-pregnancy conditions increased AUC for the prediction of blood transfusion and severe preeclampsia, and moreover, the multiracial subgroup experienced the greatest improvement in predictive performance of blood transfusion due to this addition (an increase in AUC 0.51 to 0.69). The best subgroup performance was in predicting severe preeclampsia in multiracial individuals using confounders and condition types, with an AUC of 0.74.ConclusionThough there is a greater burden of some pre-pregnancy conditions and co-occurring conditions among minority individuals, these conditions alone do not fully explain the higher rates of adverse maternal outcomes. However, some conditions and their combined effects are associated with adverse maternal outcomes. The incorporation of pre-pregnancy multimorbidity improves the performance of some risk prediction models, especially for correctly classifying individuals of minority race.

ObjectivesWe sought to identify changes in non-profit hospitals’ community benefit spending from 2014 to 2019. Secondly, we wanted to find novel predictors of spending in the most recent available data.MethodsFor our longitudinal analysis, we used tax filing data for 1072 hospitals from 2014-2019 and time-based ANOVA. We gathered information about hospital characteristics, the social determinants of health, and hospital partnerships with local communities for 1192 hospitals for the year 2019. We employed multivariate regression to identify significant factors.ResultsTotal community benefit spending rose from 8.1% in 2014 to 9.1% of operating expenses in 2019, driven by increased spending on patient care. There was no such increase in spending on activities targeted at improving community health. The presence and strength of partnerships between hospitals and their communities were associated with higher community benefit spending.ConclusionsWe found no evidence of dramatic shifts in community benefit spending from 2014 to 2019. Further, we identified partnerships for population health improvement as an effective, novel predictor of community benefit spending. Supporting partnerships between hospitals and communities may help facilitate strategic investments in community health improvement.

Post-9/11 U.S. veterans and servicemembers are at increased risk for suicide, indicating an important need to identify and mitigate suicidal ideation and behaviors in this population.Using data modeling techniques, we examined correlates of suicidal ideation and behavior at intake in 261 Post-9/11 veterans and servicemembers seeking mental health treatment.Our sample endorsed high rates of suicidal ideation and behavior. Approximately 40% of our sample scored in a range on the Suicide Behaviors Questionnaire-Revised (SBQ-R), indicating high clinical risk for suicide. Results from multivariate analyses indicate that greater state and/or trait depression severity, greater anger and anger expression, less impulse control, and lower rank were consistently associated with suicidal ideation and behavior across our models. Negative posttraumatic thoughts about the self, gender, and military branch of service were also significantly associated with suicidal ideation and behavior.Suicidal ideation and behaviors are common in veterans seeking mental health treatment. State and/or trait depression, anger and impulse control were predictors of increased risk for suicidal ideation and behavior across models. Consistencies and differences across models as well as limitations and practical implications for the findings are discussed.

Objective Noroviruses are the leading cause of acute gastroenteritis in the United States and outbreaks frequently occur in daycare settings. Results of norovirus vaccine trials have been promising, however there are open questions as to whether vaccination of daycare children would be cost-effective. We investigated the incremental cost-effectiveness of a hypothetical norovirus vaccination for children in daycare settings compared to no vaccination. Methods We conducted a model-based cost-effectiveness analysis using a disease transmission model of children attending daycare. Vaccination with a 90% coverage rate in addition to the observed standard of care (exclusion of symptomatic children from daycare) was compared to the observed standard of care. The main outcomes measures were infections and deaths averted, quality-adjusted life years (QALYs), costs, and incremental cost-effectiveness ratio (ICER). Cost-effectiveness was analyzed from a societal perspective, including medical costs to children as well as productivity losses of parents, over a two-year time horizon. Data sources included outbreak surveillance data and published literature. Results A 50% efficacious norovirus vaccine averts 571.83 norovirus cases and 0.003 norovirus-related deaths per 10,000 children compared to the observed standard of care. A $200 norovirus vaccine that is 50% efficacious has a net cost increase of $178.10 per child and 0.025 more QALYs, resulting in an ICER of $7,028/QALY. Based on the probabilistic sensitivity analysis, we estimated that a $200 vaccination with 50% efficacy was 94.0% likely to be cost-effective at a willingness-to-pay of $100,000/QALY threshold and 95.3% likely at a $150,000/QALY threshold. Conclusion Due to the large disease burden associated with norovirus, it is likely that vaccinating children in daycares could be cost-effective, even with modest vaccine efficacy and a high per-child cost of vaccination. Norovirus vaccination of children in daycare has a cost-effectiveness ratio similar to other commonly recommended childhood vaccines.

We consider the problem of decision-making in Markov decision processes (MDPs) when the reward or transition probability parameters are not known with certainty. We study an approach in which the d...

Infectious diseases, like COVID-19, pose serious challenges to university campuses, which typically adopt closure as a non-pharmaceutical intervention to control spread and ensure a gradual return to normalcy. Intervention policies, such as remote instruction (RI) where large classes are offered online, reduce potential contact but also have broad side-effects on campus by hampering the local economy, students’ learning outcomes, and community wellbeing. In this paper, we demonstrate that university policymakers can mitigate these tradeoffs by leveraging anonymized data from their WiFi infrastructure to learn community mobility —- a methodology we refer to as WiFi mobility models (WiMob). This approach enables policymakers to explore more granular policies like localized closures (LC). WiMob can construct contact networks that capture behavior in various spaces, highlighting new potential transmission pathways and temporal variation in contact behavior. Additionally, WiMob enables us to design LC policies that close super-spreader locations on campus. By simulating disease spread with contact networks from WiMob, we find that LC maintains the same reduction in cumulative infections as RI while showing greater reduction in peak infections and internal transmission. Moreover, LC reduces campus burden by closing fewer locations, forcing fewer students into completely online schedules, and requiring no additional isolation. WiMob can empower universities to conceive and assess a variety of closure policies to prevent future outbreaks.

ImportanceWhen an emerging infectious disease outbreak occurs, such as COVID-19, institutions of higher education (IHEs) must weigh decisions about how to operate their campuses. These decisions entail whether campuses should remain open, how courses should be delivered (in-person, online, or a mixture of the two), and what safety plans should be enacted for those on campus. These issues have weighed heavily on campus administrators during the on-going COVID-19 pandemic. However, there is still limited knowledge about how such decisions affect students' enrollment decisions and campus safety in practice when considering compliance.ObjectivesTo assess 1) students' willingness to comply with health protocols and contrast their perception of their classmates' compliance, 2) whether students prefer in-person or online learning during a pandemic, and 3) the importance weights of different aspects of campus operations (i.e., modes of course delivery and safety plans) for students when they decide to enroll or defer.Design setting and participantsAn internet-based survey of college students took place from June 25, 2020 to July 10, 2020. Participants included 398 industrial engineering students at the Georgia Institute of Technology, a medium-size public university in Atlanta, Georgia. The survey included a discrete choice experiment with questions that asked students to choose whether to enroll or defer when presented with hypothetical scenarios related to Fall 2020 modes of course delivery and aspects of campus safety. The survey also asked students about expected compliance with health protocols, whether they preferred in-person or online courses, and sociodemographic information.Main outcomes and measuresWe examine students' willingness to comply with potential health protocols. We estimated logistic regression models to infer significant factors that lead to a student's choice between in-person and online learning. Additionally, we estimated discrete choice models to infer the importance of different modes of course delivery and safety measures to students when deciding to enroll or defer.ResultsThe survey response rate was 20.8%. A latent class model showed three classes of students: those who were "low-concern" (comprising a 29% expected share of the sample), those who were "moderate-concern" (54%) and those who were "high-concern" (17%). We found that scenarios that offered an on-campus experience with large classes delivered online and small classes delivered in-person, strict safety protocols in terms of mask-wearing, testing, and residence halls, and lenient safety protocols in terms of social gatherings were broadly the scenarios with the highest expected enrollment probabilities. The decision to enroll or defer for all students was largely determined by the mode of delivery for courses and the safety measures on campus around COVID-19 testing and mask-wearing. A logistic regression model showed that a higher perceived risk of infection of COVID-19, a more suitable home environment, being older, and being less risk-seeking were significant factors for a person to choose online learning. Students stated for themselves and their classmates that they would comply with some but not all health protocols against COVID-19, especially those limiting social gatherings.Conclusions and relevanceThe majority of students indicated a preference to enroll during the COVID-19 pandemic so long as sufficient safety measures were put in place and all classes were not entirely in-person. As IHEs consider different options for campus operations during pandemics, they should consider the heterogeneous preferences among their students. Offering flexibility in course modes may be a way to appeal to many students who vary in terms of their concern about the pandemic. At the same time, since students overall preferred some safety measures placed around mask-wearing and COVID-19 testing on campus, IHEs may want to recommend or require wearing masks and doing some surveillance tests for all students, faculty, and staff. Students were expecting themselves and their fellow classmates to comply with some but not all health protocols, which may help IHEs identify protocols that need more education and awareness, like limits on social gatherings and the practice of social distancing at social gatherings.

Markov decision processes (MDPs) are models for sequential decision-making that inform decision making in many fields, including healthcare, manufacturing, and others. However, the optimal policy for an MDP may be sensitive to the reward and transition parameters which are often uncertain because parameters are typically estimated from data or rely on expert opinion. To address parameter uncertainty in MDPs, it has been proposed that multiple models of the parameters be incorporated into the solution process, but solving these problems can be computationally challenging. In this article, we propose a policy-based branch-and-bound approach that leverages the structure of these problems and numerically compare several important algorithmic designs. We demonstrate that our approach outperforms existing methods on test cases from the literature including randomly generated MDPs, a machine maintenance MDP, and an MDP for medical decision making.

Background Intensive blood pressure (BP) treatment can avert cardiovascular disease (CVD) events but can cause some serious adverse events. We sought to develop and validate risk models for predicting absolute risk difference (increased risk or decreased risk) for CVD events and serious adverse events from intensive BP therapy. A secondary aim was to test if the statistical method of elastic net regularization would improve the estimation of risk models for predicting absolute risk difference, as compared to a traditional backwards variable selection approach. Methods and findings Cox models were derived from SPRINT trial data and validated on ACCORD-BP trial data to estimate risk of CVD events and serious adverse events; the models included terms for intensive BP treatment and heterogeneous response to intensive treatment. The Cox models were then used to estimate the absolute reduction in probability of CVD events (benefit) and absolute increase in probability of serious adverse events (harm) for each individual from intensive treatment. We compared the method of elastic net regularization, which uses repeated internal cross-validation to select variables and estimate coefficients in the presence of collinearity, to a traditional backwards variable selection approach. Data from 9,069 SPRINT participants with complete data on covariates were utilized for model development, and data from 4,498 ACCORD-BP participants with complete data were utilized for model validation. Participants were exposed to intensive (goal systolic pressure < 120 mm Hg) versus standard (, Using data from two large clinical trials that showed heterogeneity in blood pressure treatment effects, Sanjay Basu and colleagues investigate how risks of treatment benefit and harm vary across individuals., Author summary Why was this study done? It is known that elevated blood pressure is a major risk factor for cardiovascular and related diseases. Intensive treatment of elevated blood pressure (aimed at keeping systolic blood pressures less than or equal to 120 mm Hg) may avert cardiovascular disease events, but may also pose the risk of some serious adverse events. We sought to create risk calculators to estimate individual patients’ chances of benefit and harm from intensive treatment. We additionally sought to test whether the statistical method known as elastic net regularization, which aims to reduce overfitting and improve external validity, would improve the estimation of risk models for absolute risk reduction or increase. What did the researchers do and find? We developed statistical models of cardiovascular events and serious adverse events from individual participant data from the SPRINT trial of intensive blood pressure treatment (N = 9,069 with complete covariate data) and validated them on individual participant data from the ACCORD-BP trial of intensive blood pressure treatment (N = 4,498 with complete covariate data). We used the models to calculate the absolute reduction in probability of CVD events (benefit) and absolute increase in probability of serious adverse events (harm) for individuals from intensive BP treatment. We found that the models could identify groups with high and with low absolute risk reduction in cardiovascular events and, similarly, identify groups with high and with low absolute risk increase in serious adverse events. Some participants in both the SPRINT and ACCORD studies were in groups with high predicted absolute risk reduction and low predicted absolute risk increase, and vice versa. We additionally found that using the statistical method of elastic net regularization improved the ability to identify groups with high versus low absolute risk increase in serious adverse events, compared to traditional backwards variable selection. We made an online risk calculator available, along with statistical code to apply the method to other trial datasets. What do these findings mean? The models derived in this study helped identify subgroups of participants in both SPRINT and ACCORD-BP who had lower versus higher absolute risk decreases in CVD events, and participants who had lower versus higher absolute risk increases in serious adverse events. In the future, as individual participant data become increasingly available from randomized controlled trials, benefit and harm risk calculators for personalizing therapy may become more common. The study revealed that such risk calculations for serious adverse events were improved by using an elastic net regularization approach that involves rigorous cross-validation and improves model stability when risk factors for an outcome are correlated, as with cardiovascular disease risk factors. The limitations of the study include having a limited sample size in the intensive blood pressure treatment trial that enrolled people with type 2 diabetes (ACCORD-BP) and being a secondary data analysis that cannot provide mechanistic explanations for the observed heterogeneities in treatment effect.

In recent years, Markov decision processes (MDPs) and partially observable Markov decision processes (POMDPs) have found important applications to medical decision making in the context of prevention, screening, and treatment of diseases. In this chapter, we provide a review of state-of-the-art models and methods that have been applied to chronic diseases. We provide a tutorial about how to formulate and solve these important problems emphasizing some of the challenges specific to chronic diseases such as diabetes, heart disease, and cancer. Then, we illustrate important considerations for model formulation and solution methods through two examples. The first example is an MDP model for optimal control of drug treatment decisions for managing the risk of heart disease and stroke in patients with type 2 diabetes. The second example is a POMDP model for optimal design of biomarker-based screening policies in the context of prostate cancer. We end the chapter with a discussion of the challenges of using MDPs and POMDPs for medical contexts and describe some important future directions for research.