Dataset Information

Use of Latent Class Analysis and k-Means Clustering to Identify Complex Patient Profiles.

ABSTRACT:

Importance

Medically complex patients are a heterogeneous group that contribute to a substantial proportion of health care costs. Coordinated efforts to improve care and reduce costs for this patient population have had limited success to date.

Objective

To define distinct patient clinical profiles among the most medically complex patients through clinical interpretation of analytically derived patient clusters.

Design, setting, and participants

This cohort study analyzed the most medically complex patients within Kaiser Permanente Northern California, a large integrated health care delivery system, based on comorbidity score, prior emergency department admissions, and predicted likelihood of hospitalization, from July 18, 2018, to July 15, 2019. From a starting point of over 5000 clinical variables, we used both clinical judgment and analytic methods to reduce to the 97 most informative covariates. Patients were then grouped using 2 methods (latent class analysis, generalized low-rank models, with k-means clustering). Results were interpreted by a panel of clinical stakeholders to define clinically meaningful patient profiles.

Main outcomes and measures

Complex patient profiles, 1-year health care utilization, and mortality outcomes by profile.

Results

The analysis included 104?869 individuals representing 3.3% of the adult population (mean [SD] age, 70.7 [14.5] years; 52.4% women; 39% non-White race/ethnicity). Latent class analysis resulted in a 7-class solution. Stakeholders defined the following complex patient profiles (prevalence): high acuity (9.4%), older patients with cardiovascular complications (15.9%), frail elderly (12.5%), pain management (12.3%), psychiatric illness (12.0%), cancer treatment (7.6%), and less engaged (27%). Patients in these groups had significantly different 1-year mortality rates (ranging from 3.0% for psychiatric illness profile to 23.4% for frail elderly profile; risk ratio, 7.9 [95% CI, 7.1-8.8], P?Conclusions and relevanceThe findings suggest that highly medically complex patient populations may be categorized into distinct patient profiles that are amenable to varying strategies for resource allocation and coordinated care interventions.

SUBMITTER: Grant RW

PROVIDER: S-EPMC7733156 | biostudies-literature | 2020 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Use of Latent Class Analysis and k-Means Clustering to Identify Complex Patient Profiles.

Grant Richard W RW McCloskey Jodi J Hatfield Meghan M Uratsu Connie C Ralston James D JD Bayliss Elizabeth E Kennedy Chris J CJ

JAMA network open 20201201 12

<h4>Importance</h4>Medically complex patients are a heterogeneous group that contribute to a substantial proportion of health care costs. Coordinated efforts to improve care and reduce costs for this patient population have had limited success to date.<h4>Objective</h4>To define distinct patient clinical profiles among the most medically complex patients through clinical interpretation of analytically derived patient clusters.<h4>Design, setting, and participants</h4>This cohort study analyzed t ...[more]

PMID: 33306116

Similar Datasets

Project description:IntroductionOver the last two decades, the incidence of hepatitis C virus (HCV) co-infection among men who have sex with men (MSM) living with HIV began increasing in post-industrialized countries. Little is known about transmission of acute or recent HCV, in particular among MSM living with HIV co-infection, which creates uncertainty about potential for reinfection after HCV treatment. Using phylogenetic methods, clinical, epidemiological and molecular data can be combined to better understand transmission patterns. These insights may help identify strategies to reduce reinfection risk, enhancing effectiveness of HCV treatment as prevention strategies. The aim of this study was to identify multi-risk profiles and factors associated with phylogenetic pairs and clusters among people with recent HCV infection.MethodsData and specimens from five studies of recent HCV in Australia and New Zealand (2004 to 2015) were used. HCV Core-E2 sequences were used to infer maximum likelihood trees. Clusters were identified using 90% bootstrap and 5% genetic distance threshold. Multivariate logistic regression and latent class analyses were performed.ResultsAmong 237 participants with Core-E2 sequences, 47% were in a pair/cluster. Among HIV/HCV co-infected participants, 60% (74/123) were in a pair/cluster, compared to 30% (34/114) with HCV mono-infection (p < 0.001). HIV/HCV co-infection (vs. HCV mono-infection; adjusted odds ratio (AOR), 2.37, 95% confidence interval (CI), 1.45, 5.15) was independently associated with phylogenetic clustering. Latent class analysis identified three distinct risk profiles: (1) people who inject drugs, (2) HIV-positive gay and bisexual men (GBM) with low probability of injecting drug use (IDU) and (3) GBM with IDU & sexual risk behaviour. Class 2 (vs. Class 1, AOR 3.40; 95% CI, 1.52, 7.60), was independently associated with phylogenetic clustering. Many clusters displayed homogeneous characteristics, such as containing individuals exclusively from one city, individuals all with HIV/HCV co-infection or individuals sharing the same route of acquisition of HCV.ConclusionsClusters containing individuals with specific characteristics suggest that HCV transmission occurs through discrete networks, particularly among HIV/HCV co-infected individuals. The greater proportion of clustering found among HIV/HCV co-infected participants highlights the need to provide broad direct-acting antiviral access encouraging rapid uptake in this population and ongoing monitoring of the phylogeny.

Project description:BackgroundDigital technology and media use is integral to adolescents' lives and has been associated with both positive and negative health consequences. Previous studies have largely focused on understanding technology behaviors and outcomes within adolescent populations, which can promote assumptions about adolescent technology use as homogeneous. Furthermore, many studies on adolescent technology use have focused on risks and negative outcomes. To better understand adolescent digital technology use, we need new approaches that can assess distinct profiles within study populations and take a balanced approach to understanding the risks and benefits of digital technology use.ObjectiveThe purpose of this study was to identify profiles of adolescent technology use within a large study population focusing on four evidence-based constructs: technology ownership and use, parental involvement, health outcomes, and well-being indicators.MethodsAdolescent-parent dyads were recruited for a cross-sectional web-based survey using the Qualtrics (Qualtrics International, Inc) platform and panels. Technology use measures included ownership of devices, social media use frequency, and the Adolescents' Digital Technology Interactions and Importance scale. Parent involvement measures included household media rules, technology-related parenting practices, parent social media use frequency, and the parent-child relationship. Health outcome measures included physical activity, sleep, problematic internet use, and mental health assessments. Well-being indicators included mental wellness, communication, and empathy. We used latent class analysis (LCA) to identify distinct profile groups across the aforementioned 4 critical constructs.ResultsAmong the 3981 adolescent-parent dyads recruited, adolescent participants had a mean age of 15.0 (SD 1.43) years; a total of 46.3% (1842/3981) were female, 67.8% (2701/3981) were White, and 75% (2986/3981) lived in a household with an income above the poverty line. The LCA identified 2 discrete classes. Class 1 was made up of 62.8% (2501/3981) of the participants. Class 1 participants were more likely than Class 2 participants to report family-owned devices, have lower technology importance scores, have household technology rules often centered on content, have positive parent relationships and lower parent social media use, and report better health outcomes and well-being indicators.ConclusionsFindings from this national cross-sectional survey using LCA led to 2 distinct profile groups of adolescent media use and their association with technology use and parent involvement as well as health and well-being outcomes. The two classes included a larger Class 1 (Family-Engaged Adolescents) and a smaller Class 2 (At-Risk Adolescents). The findings of this study can inform interventions to reinforce positive technology use and family support.

Dataset Information

Use of Latent Class Analysis and k-Means Clustering to Identify Complex Patient Profiles.

Importance

Objective

Design, setting, and participants

Main outcomes and measures

Results

Publications

Use of Latent Class Analysis and k-Means Clustering to Identify Complex Patient Profiles.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets