Finite Mixture Modeling for Program Evaluation: Resampling and Pre-processing Approaches

Background Finite mixture models cluster individuals into latent subgroups based on observed traits. However, inaccurate enumeration of clusters can have lasting implications on policy decisions and allocations of resources. Applied and methodological researchers accept no obvious best model fit statistic, and different measures could suggest different numbers of latent clusters. Objectives The purpose of this article is to evaluate and compare different cluster enumeration techniques. Research Design Study I demonstrates how recently proposed resampling methods result in no precise number of clusters on which all fit statistics agree. We recommend the pre-processing method in Study II as an alternative. Both studies used nationally representative data on working memory, cognitive flexibility, and inhibitory control. Conclusions The data plus priors method shows promise to address inconsistencies among fit measures and help applied researchers using finite mixture models in the future.

Impact of Users' Attitudes Toward Anonymous Internet Interventions for Cannabis vs. Alcohol Use: A Secondary Analysis of Data From Two Clinical Trials

Background: Numerous trials have demonstrated the efficacy of internet interventions targeting alcohol or cannabis use, yet a substantial proportion of users do not benefit from the format, warranting further research to identify moderators of treatment effects. Users' initial attitudes toward treatment is a potential moderator, yet no previous study has investigated users' attitudes in the context of internet interventions for addictive disorders.Method: In this secondary analysis on two internet-based trials targeting harmful alcohol use (n = 1,169) and regular cannabis use (n = 303), respectively, we compared user groups' attitudes at the item level; explored within-group heterogeneity by submitting attitude scores to a k-means cluster analysis; and investigated whether latent subgroups in each user group moderated the treatment effects. Outcome models were run using generalized linear models with 10,000 bias-corrected bootstraps accounting for subject-level clustering.Results: While substance groups and latent subgroups converged in enjoying the anonymity provided by the format, their interest toward treatment differed. Outcome analyses revealed a significant and negative time by subgroup effect on grams of cannabis consumed and screening test score (CAST), favoring the subgroup with positive treatment attitudes. There were not any significant effects of subgroup on alcohol consumption. Despite initial treatment reluctance, participants in the neutral subgroup decreased their cannabis use (gram) significantly when receiving the intervention vs. control.Conclusions: This first, exploratory study revealed key differences between substance groups' attitudes, but more importantly that within-group heterogeneity appear to affect cannabis outcomes. Assessing attitudes could be key in patient-treatment matching, yet more research is needed.

Mixture models and networks: The stochastic blockmodel

Mixture models are probabilistic models aimed at uncovering and representing latent subgroups within a population. In the realm of network data analysis, the latent subgroups of nodes are typically identified by their connectivity behaviour, with nodes behaving similarly belonging to the same community. In this context, mixture modelling is pursued through stochastic blockmodelling. We consider stochastic blockmodels and some of their variants and extensions from a mixture modelling perspective. We also explore some of the main classes of estimation methods available and propose an alternative approach based on the reformulation of the blockmodel as a graphon. In addition to the discussion of inferential properties and estimating procedures, we focus on the application of the models to several real-world network datasets, showcasing the advantages and pitfalls of different approaches.

Multiple Chronic Conditions among Seriously Ill Adults Receiving Palliative Care

The objective of this study was to characterize multiple chronic conditions (MCCs) among seriously ill adults receiving palliative care at the end of life. A latent class analysis was conducted to identify latent subgroups of seriously ill older adults based on a baseline Charlson comorbidity index (CCI) measurement, a measure of comorbidity burden, and mortality risk. The three latent subgroups were: (1) low to moderate CCI with MCC, (2) high CCI with MCC, and (3) high CCI and metastatic cancer. The “low to moderate CCI and MCC” subgroup included older adults with chronic obstructive pulmonary disease (COPD), cardiovascular disease, congestive heart failure, myocardial infarction, dementia, diabetes, and lymphoma. A “high CCI and MCC” subgroup included individuals with severe illness including liver or renal disease among other MCCs. A “high CCI and metastatic cancer” included all participants with metastatic cancer. This study sheds light on the MCC profile of seriously ill adults receiving palliative care.

Association of fluid balance trajectories with clinical outcomes in patients with septic shock: a prospective multicenter cohort study

Background Septic shock has a high incidence and mortality rate in Intensive Care Units (ICUs). Earlier intravenous fluid resuscitation can significantly improve outcomes in septic patients but easily leads to fluid overload (FO), which is associated with poor clinical outcomes. A single point value of fluid cannot provide enough fluid information. The aim of this study was to investigate the impact of fluid balance (FB) latent trajectories on clinical outcomes in septic patients. Methods Patients were diagnosed with septic shock during the first 48 h, and sequential fluid data for the first 3 days of ICU admission were included. A group-based trajectory model (GBTM) which is designed to identify groups of individuals following similar developmental trajectories was used to identify latent subgroups of individuals following a similar progression of FB. The primary outcomes were hospital mortality, organ dysfunction, major adverse kidney events (MAKE) and severe respiratory adverse events (SRAE). We used multivariable Cox or logistic regression analysis to assess the association between FB trajectories and clinical outcomes. Results Nine hundred eighty-six patients met the inclusion criteria and were assigned to GBTM analysis, and three latent FB trajectories were detected. 64 (6.5%), 841 (85.3%), and 81 (8.2%) patients were identified to have decreased, low, and high FB, respectively. Compared with low FB, high FB was associated with increased hospital mortality [hazard ratio (HR) 1.63, 95% confidence interval (CI) 1.22–2.17], organ dysfunction [odds ratio (OR) 2.18, 95% CI 1.22–3.42], MAKE (OR 1.80, 95% CI 1.04–2.63) and SRAE (OR 2.33, 95% CI 1.46–3.71), and decreasing FB was significantly associated with decreased MAKE (OR 0.46, 95% CI 0.29–0.79) after adjustment for potential covariates. Conclusion Latent subgroups of septic patients followed a similar FB progression. These latent fluid trajectories were associated with clinical outcomes. The decreasing FB trajectory was associated with a decreased risk of hospital mortality and MAKE.

Clustering of COVID-19 Symptoms Among Iranian Patients: The Role of Preexisting Comorbidity on Latent Class Membership

This study aimed to identify subgroups of coronavirus disease 2019 (COVID-19) symptoms and assess the role that preexisting comorbidity on membership of specific subgroup. This cross-sectional study took place in Ardabil, northwest of Iran. All patients (16 183) who were admitted to the hospitals of Ardabil province were recruited. Six indicator variables were selected to identify latent subgroups of patients using the result of polymerase chain reaction (PCR) test as a grouping variable. Data analysis was performed using χ2, independent t test, and latent class analysis. This study found that among PCR-positive patients, there were 3 latent classes: (1) mild disease (16.1%), (2) semi-severe disease (62.5%), and (3) severe disease (21.3%). This study showed that having preexisting comorbidity increase the odds of membership in semi-severe disease (odds ratio = 2.30) and severe disease (odds ratio = 1.60) classes compared with mild disease class. Focusing on patients who experience co-occurrence of more symptoms may be helpful in control of COVID-19.