Cluster analysis and its application to healthcare claims data: a study of end-stage renal disease patients who initiated hemodialysis

Exploring kidney dialysis costs in the United States: a scoping review

AIMS

The increasing prevalence of end-stage renal disease (ESRD) in the United States (US) represents a considerable economic burden due to the high cost of dialysis treatment. This review examines data from real-world studies to identify cost drivers and explore areas where dialysis costs could be reduced.

METHODS

We identified and synthesized evidence published from 2016-2023 reporting direct dialysis costs in adult US patients from a comprehensive literature search of MEDLINE, Embase, and grey literature sources (e.g. US Renal Data System reports).

RESULTS

Most identified data related to Medicare expenditures. Overall Medicare spending in 2020 was $29B for hemodialysis and $2.8B for peritoneal dialysis (PD). Dialysis costs accounted for almost 80% of total Medicare expenditures on ESRD beneficiaries. Private insurance payers consistently pay more for dialysis; for example, per person per month spending by private insurers on outpatient dialysis was estimated at $10,149 compared with Medicare spending of $3,364. Dialysis costs were higher in specific high-risk patient groups (e.g. type 2 diabetes, hepatitis C). Spending on hemodialysis was higher than on PD, but the gap in spending between PD and hemodialysis is closing. Vascular access costs accounted for a substantial proportion of dialysis costs.

LIMITATIONS

Insufficient detail in the identified studies, especially related to outpatient costs, limits opportunities to identify key drivers. Differences between the studies in methods of measuring dialysis costs make generalization of these results difficult.

CONCLUSIONS

These findings indicate that prevention of or delay in progression to ESRD could have considerable cost savings for Medicare and private payers, particularly in patients with high-risk conditions such as type 2 diabetes. More efficient use of resources is needed, including low-cost medication, to improve clinical outcomes and lower overall costs, especially in high-risk groups. Widening access to PD where it is safe and appropriate may help to reduce dialysis costs.

Subtyping irritable bowel syndrome using cluster analysis: a systematic review

BACKGROUND

Irritable bowel syndrome (IBS) is a common chronic functional gastrointestinal disorder associated with a wide range of clinical symptoms. Some researchers have used cluster analysis (CA), a group of non-supervised learning methods that identifies homogenous clusters within different entities based on their similarity.

OBJECTIVE AND METHODS

This literature review aims to identify published articles that apply CA to IBS patients. We searched relevant keywords in PubMed, Embase, Web of Science, and Scopus. We reviewed studies in terms of the selected variables, participants' characteristics, data collection, methodology, number of clusters, clusters' profiles, and results.

RESULTS

Among the 14 articles focused on the heterogeneity of IBS, eight of them utilized K-means Cluster Analysis (K-means CA), four employed Hierarchical Cluster Analysis, and only two studies utilized Latent Class Analysis. Seven studies focused on clinical symptoms, while four articles examined anocolorectal functions. Two studies were centered around immunological findings, and only one study explored microbial composition. The number of clusters obtained ranged from two to seven, showing variation across the studies. Males exhibited lower symptom severity and fewer psychological findings. The association between symptom severity and rectal perception suggests that altered rectal perception serves as a biological indicator of IBS. Ultra-slow waves observed in IBS patients are linked to increased activity of the anal sphincter, higher anal pressure, dystonia, and dyschezia.

CONCLUSION

IBS has different subgroups based on different factors. Most IBS patients have low clinical severity, good QoL, high rectal sensitivity, delayed left colon transit time, increased systemic cytokines, and changes in microbial composition, including increased Firmicutes-associated taxa and depleted Bacteroidetes-related taxa. However, the number of clusters is inconsistent across studies due to the methodological heterogeneity. CA, a valuable non-supervised learning method, is sensitive to hyperparameters like the number of clusters and random initialization of cluster centers. The random nature of these parameters leads to diverse outcomes even with the same algorithm. This has implications for future research and practical applications, necessitating further studies to improve our understanding of IBS and develop personalized treatments.

Identifying user profiles of healthcare, social and employment services in a working-age population: A cluster analysis with linked individual-level register data from Finland

A thorough understanding of the use of services in the population is important in order to comprehend the varying service needs of different groups. This explorative study aimed to find distinct user profiles in a working-age population based on individuals' annual use of healthcare, social and employment services and to explore socio-demographic and morbidity-related predictors of the user groups. Administrative register data on the use of various services and individual-level covariates from year 2018 were linked for all residents aged 18-64 of the municipality of Oulu, Finland (N = 119,740). K-means cluster analysis was used to group the study subjects into clusters, based on their frequency of using 22 distinct healthcare, social and employment services during 2018. Multinomial logistic regression models were utilized to assess the associations of cluster assignment with socio-demographic and health-related covariates (sex, age, marital status, education, occupational class, income, days in employment, chronic disease and receipt of different social benefits). Five distinct clusters were identified in terms of service use, labelled low to moderate users of healthcare (82.0%), regular employment services users with moderate use of healthcare (9.6%), supported employment services users with moderate use of healthcare with an emphasis on preventive care (2.9%), frequent users of healthcare, social and employment services (2.9%), and rehabilitation, disability services and specialized healthcare users (2.6%). Each cluster not only showed different patterns of service use but were also differently associated with demographic, socio-economic and morbidity-related covariates, creating distinct service user types. Knowledge on the different user profiles and their determinants may help predict future need and use of services in a population, plan timely, coordinated and integrated services, and design early interventions and prevention measures. This is important in order to save costs and improve the effectiveness of services for groups with different care needs.

Data Mining and Fusion Framework for In-Home Monitoring Applications

Sensor Data Fusion (SDT) algorithms and models have been widely used in diverse applications. One of the main challenges of SDT includes how to deal with heterogeneous and complex datasets with different formats. The present work utilised both homogenous and heterogeneous datasets to propose a novel SDT framework. It compares data mining-based fusion software packages such as RapidMiner Studio, Anaconda, Weka, and Orange, and proposes a data fusion framework suitable for in-home applications. A total of 574 privacy-friendly (binary) images and 1722 datasets gleaned from thermal and Radar sensing solutions, respectively, were fused using the software packages on instances of homogeneous and heterogeneous data aggregation. Experimental results indicated that the proposed fusion framework achieved an average Classification Accuracy of 84.7% and 95.7% on homogeneous and heterogeneous datasets, respectively, with the help of data mining and machine learning models such as Naïve Bayes, Decision Tree, Neural Network, Random Forest, Stochastic Gradient Descent, Support Vector Machine, and CN2 Induction. Further evaluation of the Sensor Data Fusion framework based on cross-validation of features indicated average values of 94.4% for Classification Accuracy, 95.7% for Precision, and 96.4% for Recall. The novelty of the proposed framework includes cost and timesaving advantages for data labelling and preparation, and feature extraction.

A data-driven approach to categorize patients with traumatic spinal cord injury: cluster analysis of a multicentre database

Background

Conducting clinical trials for traumatic spinal cord injury (tSCI) presents challenges due to patient heterogeneity. Identifying clinically similar subgroups using patient demographics and baseline injury characteristics could lead to better patient-centered care and integrated care delivery.

Purpose

We sought to (1) apply an unsupervised machine learning approach of cluster analysis to identify subgroups of tSCI patients using patient demographics and injury characteristics at baseline, (2) to find clinical similarity within subgroups using etiological variables and outcome variables, and (3) to create multi-dimensional labels for categorizing patients.

Study design

Retrospective analysis using prospectively collected data from a large national multicenter SCI registry.

Methods

A method of spectral clustering was used to identify patient subgroups based on the following baseline variables collected since admission until rehabilitation: location of the injury, severity of the injury, Functional Independence Measure (FIM) motor, and demographic data (age, and body mass index). The FIM motor score, the FIM motor score change, and the total length of stay were assessed on the subgroups as outcome variables at discharge to establish the clinical similarity of the patients within derived subgroups. Furthermore, we discussed the relevance of the identified subgroups based on the etiological variables (energy and mechanism of injury) and compared them with the literature. Our study also employed a qualitative approach to systematically describe the identified subgroups, crafting multi-dimensional labels to highlight distinguishing factors and patient-focused insights.

Results

Data on 334 tSCI patients from the Rick Hansen Spinal Cord Injury Registry was analyzed. Five significantly different subgroups were identified (p-value ≤0.05) based on baseline variables. Outcome variables at discharge superimposed on these subgroups had statistically different values between them (p-value ≤0.05) and supported the notion of clinical similarity of patients within each subgroup.

Conclusion

Utilizing cluster analysis, we identified five clinically similar subgroups of tSCI patients at baseline, yielding statistically significant inter-group differences in clinical outcomes. These subgroups offer a novel, data-driven categorization of tSCI patients which aligns with their demographics and injury characteristics. As it also correlates with traditional tSCI classifications, this categorization could lead to improved personalized patient-centered care.

Predicting outcomes at the individual patient level: what is the best method?

OBJECTIVE

When developing prediction models, researchers commonly employ a single model which uses all the available data (end-to-end approach). Alternatively, a similarity-based approach has been previously proposed, in which patients with similar clinical characteristics are first grouped into clusters, then prediction models are developed within each cluster. The potential advantage of the similarity-based approach is that it may better address heterogeneity in patient characteristics. However, it remains unclear whether it improves the overall predictive performance. We illustrate the similarity-based approach using data from people with depression and empirically compare its performance with the end-to-end approach.

METHODS

We used primary care data collected in general practices in the UK. Using 31 predefined baseline variables, we aimed to predict the severity of depressive symptoms, measured by Patient Health Questionnaire-9, 60 days after initiation of antidepressant treatment. Following the similarity-based approach, we used k-means to cluster patients based on their baseline characteristics. We derived the optimal number of clusters using the Silhouette coefficient. We used ridge regression to build prediction models in both approaches. To compare the models' performance, we calculated the mean absolute error (MAE) and the coefficient of determination (R2) using bootstrapping.

RESULTS

We analysed data from 16 384 patients. The end-to-end approach resulted in an MAE of 4.64 and R2 of 0.20. The best-performing similarity-based model was for four clusters, with MAE of 4.65 and R2 of 0.19.

CONCLUSIONS

The end-to-end and the similarity-based model yielded comparable performance. Due to its simplicity, the end-to-end approach can be favoured when using demographic and clinical data to build prediction models on pharmacological treatments for depression.

K-means clustering of outpatient prescription claims for health insureds in Iran

OBJECTIVE

The segmentation of consumers based on their behavior and needs is the most crucial action of the health insurance organization. This study's objective is to cluster Iranian health insureds according to their demographics and data on outpatient prescriptions.

SETTING

The population in this study corresponded to the research sample. The Health Insurance Organization's outpatient claims were registered consecutively in 2016, 2017, 2018, and 2019 were clustered.

DESIGN

The k-means clustering algorithm was used to cross-sectionally and retrospectively analyze secondary data from outpatient prescription claims for secondary care using Python 3.10.

PARTICIPANTS

The current analysis transformed 21 776 350 outpatient prescription claims from health insured into 193 552 insureds.

RESULTS

Insureds using IQR were split into three classes: low, middle, and high risk. Based on the silhouette coefficient, the insureds of all classes were divided into three clusters. In all data for a period of four years, the first through third clusters, there were 21 799, 7170, and 19 419 insureds in the low-risk class. Middle-risk class had 48 348,23 321, 25 107 insureds, and 14 037, 28 504, 5847 insured in the high-risk class were included. For the first cluster of low-risk insureds: the total average cost of prescriptions paid by the insurance for the insureds was $211, the average age was 26 years, the average franchise was 88.5US$, the average number of medications and prescriptions were 409 and 62, the total average costs of prescriptions Outpatient was 302.5 US$, the total average number of medications for acute and chronic disease was 178 and 215, respectively. The majority of insureds were men, and those who were part of the householder's family.

CONCLUSIONS

By segmenting insurance customers, insurers can set insurance premium rates, controlling the risk of loss, which improves their capacity to compete in the insurance market.

Can we design the next generation of digital health communication programs by leveraging the power of artificial intelligence to segment target audiences, bolster impact and deliver differentiated services? A machine learning analysis of survey data from rural India

OBJECTIVES

Direct to beneficiary (D2B) mobile health communication programmes have been used to provide reproductive, maternal, neonatal and child health information to women and their families in a number of countries globally. Programmes to date have provided the same content, at the same frequency, using the same channel to large beneficiary populations. This manuscript presents a proof of concept approach that uses machine learning to segment populations of women with access to phones and their husbands into distinct clusters to support differential digital programme design and delivery.

SETTING

Data used in this study were drawn from cross-sectional survey conducted in four districts of Madhya Pradesh, India.

PARTICIPANTS

Study participant included pregnant women with access to a phone (n=5095) and their husbands (n=3842) RESULTS: We used an iterative process involving K-Means clustering and Lasso regression to segment couples into three distinct clusters. Cluster 1 (n=1408) tended to be poorer, less educated men and women, with low levels of digital access and skills. Cluster 2 (n=666) had a mid-level of digital access and skills among men but not women. Cluster 3 (n=1410) had high digital access and skill among men and moderate access and skills among women. Exposure to the D2B programme 'Kilkari' showed the greatest difference in Cluster 2, including an 8% difference in use of reversible modern contraceptives, 7% in child immunisation at 10 weeks, 3% in child immunisation at 9 months and 4% in the timeliness of immunisation at 10 weeks and 9 months.

CONCLUSIONS

Findings suggest that segmenting populations into distinct clusters for differentiated programme design and delivery may serve to improve reach and impact.

TRIAL REGISTRATION NUMBER

NCT03576157.

Pharyngeal biorhythms during oral milk challenge in high-risk infants: Do they predict chronic tube feeding?

BACKGROUND

Eating difficulties are common in high-risk neonatal intensive care unit (NICU) infants; mechanisms remain unclear. Crib-side pharyngo-esophageal motility testing is utilized to assess contiguous swallowing physiology, and cross-system interplay with cardio-respiratory rhythms. Aims were to: (1) identify whether distinct pharyngeal rhythms exist during oral milk challenge (OMC), and (2) develop a chronic tube feeding risk prediction model in high-risk infants.

METHODS

Symptomatic NICU infants (N = 56, 29.7 ± 3.7 weeks birth gestation) underwent pharyngo-esophageal manometry with OMC at 40.9 ± 2.5 weeks postmenstrual age (PMA). Exploratory cluster data analysis (partitioning around k-medoids) was performed to identify patient groups using pharyngeal contractile rhythm data (solitary swallows and swallows within bursts). Subsequently, (a) pharyngeal-esophageal, cardio-respiratory, and eating method characteristics were compared among patient groups using linear mixed models, and (b) chronic tube feeding prediction model was created using linear regression.

RESULTS

Three distinct patient groups were identified with validity score of 0.6, and termed sparse (high frequency of solitary swallows), intermediate, or robust (high swallow rate within bursts). Robust group infants had: lesser pharyngeal and esophageal variability, greater deglutition apnea, pharyngeal activity, and esophageal activity (all p < 0.05), but less frequent heart rate decreases (p < 0.05) with improved clinical outcomes (milk transfer rate, p < 0.001, and independent oral feeding at discharge, p < 0.03). Chronic tube feeding risk = -11.37 + (0.22 × PMA) + (-0.73 × bronchopulmonary dysplasia) + (1.46 × intermediate group) + (2.57 × sparse group).

CONCLUSIONS

Robust pharyngeal rhythm may be an ideal neurosensorimotor biomarker of independent oral feeding. Differential maturation of cranial nerve-mediated excitatory and inhibitory components involving foregut, airway, and cardiac rhythms distinguishes the physiologic and pathophysiologic basis of swallowing and cardio-respiratory adaptation.

Shared decision-making practices and patient values in pharmacist outpatient care for rheumatic disease: A multiple correspondence analysis

Purpose: To investigate the value-to-value relationships, relationship between values and patient background, continuation rate of treatment after shared decision-making (SDM), and disease status in order to clarify the values involved in drug therapy decisions for patients with rheumatic disease. Methods: We investigated patient values (efficacy of drug therapy [effectiveness], safety, economics, daily life, and other) and the continuance rate and disease status of treatment after 6 months in 94 patients with rheumatic disease aged ≥18 years who made decisions with pharmacists and physicians in the pharmacy outpatient clinic between September 2019 and April 2021. Multiple correspondence and K-means cluster analyses were performed to show the relationship between values and basic patient information. Results: Among the selected patients, 87% and 47% selected effectiveness for multiple selections and single selection, respectively. Effectiveness was at the center of the graph; three clusters containing other values were placed around it. History of allergy or side effects caused by biologics or Janus kinase inhibitors were in the safety cluster. The non-usage history of biologics or Janus kinase inhibitors was in the economic cluster. Conclusion: Effectiveness was the most important factor for patients with rheumatic disease; the values that patients consider important may shift from effectiveness to other values based on each patient's subjective experience with the treatment and/or the stage of life in which they were treated. It is important to positively link patient values and information about the treatment plan in shared decision-making while establishing rapport with the patient.

Use of Multiple Correspondence Analysis and K-means to Explore Associations Between Risk Factors and Likelihood of Colorectal Cancer: Cross-sectional Study

BACKGROUND

Previous works have shown that risk factors are associated with an increased likelihood of colorectal cancer.

OBJECTIVE

The purpose of this study was to detect these associations in the region of Lleida (Catalonia) by using multiple correspondence analysis (MCA) and k-means.

METHODS

This cross-sectional study was made up of 1083 colorectal cancer episodes between 2012 and 2015, extracted from the population-based cancer registry for the province of Lleida (Spain), the Primary Care Centers database, and the Catalan Health Service Register. The data set included risk factors such as smoking and BMI as well as sociodemographic information and tumor details. The relations between the risk factors and patient characteristics were identified using MCA and k-means.

RESULTS

The combination of these techniques helps to detect clusters of patients with similar risk factors. Risk of death is associated with being elderly and obesity or being overweight. Stage III cancer is associated with people aged ≥65 years and rural/semiurban populations, while younger people were associated with stage 0.

CONCLUSIONS

MCA and k-means were significantly useful for detecting associations between risk factors and patient characteristics. These techniques have proven to be effective tools for analyzing the incidence of some factors in colorectal cancer. The outcomes obtained help corroborate suspected trends and stimulate the use of these techniques for finding the association of risk factors with the incidence of other cancers.

Analysis of the mandibular canal course using unsupervised machine learning algorithm

Objectives

Anatomical structure classification is necessary task in medical field, but the inevitable variability of interpretation among experts makes reliable classification difficult. This study aims to introduce cluster analysis, unsupervised machine learning method, for classification of three-dimensional (3D) mandibular canal (MC) courses, and to visualize standard MC courses derived from cluster analysis in the Korean population.

Materials and methods

A total of 429 cone-beam computed tomography images were used. Four sites in the mandible were selected for the measurement of the MC course and four parameters, two vertical and two horizontal parameters were measured per site. Cluster analysis was carried out as follows: parameter measurement, parameter normalization, cluster tendency evaluation, optimal number of clusters determination, and k-means cluster analysis. The 3D MC courses were classified into three types with statistically significant mean differences by cluster analysis.

Results

Cluster 1 showed a smooth line running towards the lingual side in the axial view and a steep slope in the sagittal view. Cluster 2 ran in an almost straight line closest to the lingual and inferior border of mandible. Cluster 3 showed the pathway with a bent buccally in the axial view and an increasing slope in the sagittal view in the posterior area. Cluster 2 showed the highest distribution (42.1%), and males were more widely distributed (57.1%) than the females (42.9%). Cluster 3 comprised similar ratio of male and female cases and accounted for 31.9% of the total distribution. Cluster 1 had the least distribution (26.0%) Distributions of the right and left sides did not show a statistically significant difference.

Conclusion

The MC courses were automatically classified as three types through cluster analysis. Cluster analysis enables the unbiased classification of the anatomical structures by reducing observer variability and can present representative standard information for each classified group.

Leishmania infantum strains from cats are similar in biological properties to canine and human strains

Zoonotic visceral leishmaniosis is a worldwide severe disease caused by Leishmania infantum, a protozoan that has phlebotomine sand flies as vectors and dogs as primary reservoir hosts. Over the last few decades, cats have been regarded as an indisputable piece within the ecological system in which L. infantum is maintained indefinitely. However, little is known about feline strains, including their phenotypic plasticity and infectivity. In this study, the phenotypic behaviour of seven L. infantum feline strains was compared to those of well-characterised counterparts isolated from two dogs and two humans in terms of growth profile, adaptive capacity under several stress conditions, susceptibility to antileishmanial drugs, and infectivity to host cells. Feline strains displayed a similar growth profile, survival capacity, and ability to infect feline, canine, and human monocyte-derived primary macrophages. Furthermore, multivariate cluster analysis suggested that most strains studied did not display distinctive phenotypic features. To our knowledge, this is the first study to analyse the phenotypic behaviour of feline L. infantum strains. This study brings new insights into the hypothetical role of cats as reservoir hosts of L. infantum since the parasites found in them are phenotypically identical to those of dogs and humans. However, further studies on the transmission dynamics should be encouraged to fully establish the status of cats in the maintenance of L. infantum foci.

Evaluating the relationship between clinical and demographic characteristics of insulin-using people with diabetes and their health outcomes: a cluster analysis application

Background

The aim of this study was to determine how clusters or subgroups of insulin-treated people with diabetes, based upon healthcare resource utilization, select social demographic and clinical characteristics, and diabetes management parameters, are related to health outcomes including acute care visits and hospital admissions.

Methods

This was a non-experimental, retrospective cluster analysis. We utilized Aetna administrative claims data to identify insulin-using people with diabetes with service dates from 01 January 2015 to 30 June 2018. The study included adults over the age of 18 years who had a diagnosis of type 1 (T1DM) or type 2 diabetes mellitus (T2DM) on insulin therapy and had Aetna medical and pharmacy coverage for at least 18 months (6 months prior and 12 months after their index date, defined as either their first insulin prescription fill date or their earliest date allowing for 6 months’ prior coverage). We used K-means clustering methods to identify relevant subgroups of people with diabetes based on 13 primary outcome variables.

Results

A total of 100,650 insulin-using people with diabetes were identified in the Aetna administrative claims database and met study criteria, including 11,826 (11.7%) with T1DM and 88,824 (88.3%) with T2DM. Of these 79,053 (78.5%) people were existing insulin users. Seven distinct clusters were identified with different characteristics and potential risks of diabetes complications. Overall, clusters were significantly associated with differences in healthcare utilization (emergency room visits, inpatient admissions, and total inpatient days) after multivariable adjustment.

Conclusions

This analysis of healthcare claims data using clustering methodologies identified meaningful subgroups of patients with diabetes using insulin. The subgroups differed in comorbidity burden, healthcare utilization, and demographic factors which could be used to identify higher risk patients and/or guide the management and treatment of diabetes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-021-06603-0.

A novel hybrid DEMATEL-K-means clustering algorithm for modeling the barriers of green computing adoption in the Philippines

Purpose

This paper aims to propose a novel hybrid-decision-making trial and evaluation laboratory-K means clustering algorithm as a decision-making framework for analyzing the barriers of green computing adoption.

Design/methodology/approach

A literature review is conducted to extract relevant green computing barriers. An expert elicitation process is performed to finalize the barriers and to establish their corresponding interrelationships.

Findings

The proposed approach offers a comprehensive framework for modeling the barriers of green computing adoption.

Research limitations/implications

The results of this paper provide insights on how the barriers of green computing adoption facilitate the adoption of stakeholders. Moreover, the paper provides a framework for analyzing the structural relationships that exist between factors in a tractable manner.

Originality/value

The paper is one of the very first attempts to analyze the barriers of green computing adoption. Furthermore, it is the first to offer lenses in a Philippine perspective. The paper offers a novel algorithm that can be useful in modeling the barriers of innovation, particularly, in green computing adoption.

Cluster Analysis of Care Pathways in Adults with Major Depressive Disorder with Acute Suicidal Ideation or Behavior in the USA

BACKGROUND AND OBJECTIVE

Suicidal ideation or behavior are core symptoms of major depressive disorder (MDD). This study aimed to understand heterogeneity among patients with MDD and acute suicidal ideation or behavior.

METHODS

Adults with a diagnosis of MDD on the same day or 6 months before a claim for suicidal ideation or behavior (index date) were identified in the MarketScan^® Databases (10/01/2014-04/30/2019). A mathematical algorithm was used to cluster patients on characteristics of care measured pre-index. Patient care pathways were described by cluster during the 12-month pre-index period and up to 12 months post-index.

RESULTS

Among 38,876 patients with MDD and acute suicidal ideation or behavior, three clusters were identified. Across clusters, pre-index exposure to mental healthcare was revealed as a key differentiator: Cluster 1 (N = 16,025) was least exposed, Cluster 2 (N = 5640) moderately exposed, and Cluster 3 (N = 17,211) most exposed. Patients whose MDD diagnosis was first observed during their index event comprised 86.0% and 72.8% of Clusters 1 and 2, respectively; in Cluster 3, all patients had an MDD diagnosis pre-index. Within 30 days post-index, in Clusters 1, 2, and 3, respectively, 79.3%, 85.2%, and 88.2% used mental health services, including outpatient visits for MDD. Within 12 months post-index, 61.5%, 91.5%, and 84.6% had one or more antidepressant claim, respectively. Per-patient index event costs averaged $5614, $6645, and $5853, respectively.

CONCLUSIONS

Patients with MDD and acute suicidal ideation or behavior least exposed to the healthcare system pre-index similarly received the least care post-index. An opportunity exists to optimize treatment and follow-up with mental health services.

Pattern Recognition of the COVID-19 Pandemic in the United States: Implications for Disease Mitigation

The novel coronavirus (COVID-19) pandemic presents a severe threat to human health worldwide. The United States (US) has the highest number of reported COVID-19 cases, and over 16 million people were infected up to the 12 December 2020. To better understand and mitigate the spread of the disease, it is necessary to recognize the pattern of the outbreak. In this study, we explored the patterns of COVID-19 cases in the US from 1 March to 12 December 2020. The county-level cases and rates of the disease were mapped using a geographic information system (GIS). The overall trend of the disease in the US, as well as in each of its 50 individual states, were analyzed by the seasonal-trend decomposition. The disease curve in each state was further examined using K-means clustering and principal component analysis (PCA). The results showed that three clusters were observed in the early phase (1 March–31 May). New York has a unique pattern of the disease curve and was assigned one cluster alone. Two clusters were observed in the middle phase (1 June–30 September). California, Texas and Florida were assigned in the same cluster, which has the pattern different from the remaining states. In the late phase (1 October–12 December), California has a unique pattern of the disease curve and was assigned a cluster alone. In the whole period, three clusters were observed. California, Texas and Florida still have similar patterns and were assigned in the same cluster. The trend analysis consolidated the patterns identified from the cluster analysis. The results from this study provide insight in making disease control and mitigation strategies.

Multilevel clustering approach driven by continuous glucose monitoring data for further classification of type 2 diabetes

INTRODUCTION

Mining knowledge from continuous glucose monitoring (CGM) data to classify highly heterogeneous patients with type 2 diabetes according to their characteristics remains unaddressed. A refined clustering method that retrieves hidden information from CGM data could provide a viable method to identify patients with different degrees of dysglycemia and clinical phenotypes.

RESEARCH DESIGN AND METHODS

From Shanghai Jiao Tong University Affiliated Sixth People's Hospital, we selected 908 patients with type 2 diabetes (18-83 years) who wore blinded CGM sensors (iPro2, Medtronic, California, USA). Participants were clustered based on CGM data during a 24-hour period by our method. The first level extracted the knowledge-based and statistics-based features to describe CGM signals from multiple perspectives. The Fisher score and variables cluster analysis were applied to fuse features into low dimensions at the second level. The third level divided subjects into subgroups with different clinical phenotypes. The four subgroups of patients were determined by clinical phenotypes.

RESULTS

Four subgroups of patients with type 2 diabetes with significantly different statistical features and clinical phenotypes were identified by our method. In particular, individuals in cluster 1 were characterized by the lowest glucose level factor and glucose fluctuation factor, and the highest negative glucose factor and C peptide index. By contrast, cluster 2 had the highest glucose level factor and the lowest C peptide index. Cluster 4 was characterized by the greatest degree of glucose fluctuation factor, was the most insulin-sensitive, and had the lowest insulin resistance. Cluster 3 ranked in the middle concerning the CGM-derived metrics and clinical phenotypes compared with those of the other three groups.

CONCLUSION

A novel multilevel clustering approach for knowledge mining from CGM data in type 2 diabetes is presented. The results demonstrate that subgroups are adequately distinguished with notable statistical and clinical differences.

Comparison of out-of-pocket expenditure and catastrophic health expenditure for severe disease by the health security system: based on end-stage renal disease in South Korea

Background

Korea’s health security system named the National Health Insurance and Medical Aid has revolutionized the nation’s mandatory health insurance and continues to reduce excessive copayments. However, few studies have examined healthcare utilization and expenditure by the health security system for severe diseases. This study looked at reverse discrimination regarding end-stage renal disease by the National Health Insurance and Medical Aid.

Methods

A total of 305 subjects were diagnosed with end-stage renal disease in the Korea Health Panel from 2008 to 2013. Chi-square, t-test, and ANCOVA were conducted to identify the healthcare utilization rate, out-of-pocket expenditure, and the prevalence of catastrophic expenditure. Mixed effect panel analysis was used to evaluate total out-of-pocket expenditure by the National Health Insurance and Medical Aid over a 6-year period.

Results

There were no significant differences in the healthcare utilization rate for emergency room visits, admissions, or outpatient department visits between the National Health Insurance and Medical Aid because these healthcare services were essential for individuals with serious diseases, such as end-stage renal disease. Meanwhile, each out-of-pocket expenditure for an admission and the outpatient department by the National Health Insurance was 2.6 and 3.1 times higher than that of Medical Aid (P < 0.05). The total out-of-pocket expenditure, including that for emergency room visits, admission, outpatient department visits, and prescribed drugs, was 2.9 times higher for the National Health Insurance than Medical Aid (P < 0.001). Over a 6-year period, in terms of total of out-of-pocket expenditure, subjects with the National Health Insurance spent more than those with Medical Aid (P < 0.01). If the total household income decile was less than the median and subjects were covered by the National Health Insurance, the catastrophic health expenditure rate was 92.2%, but it was only 58.8% for Medical Aid (P < 0.001).

Conclusion

Individuals with serious diseases, such as end-stage renal disease, can be faced with reverse discrimination depending on the type of insurance that is provided by the health security system. It is necessary to consider individuals who have National Health Insurance but are still poor.

Trajectory Modelling Techniques Useful to Epidemiological Research: A Comparative Narrative Review of Approaches

Trajectory modelling techniques have been developed to determine subgroups within a given population and are increasingly used to better understand intra- and inter-individual variability in health outcome patterns over time. The objectives of this narrative review are to explore various trajectory modelling approaches useful to epidemiological research and give an overview of their applications and differences. Guidance for reporting on the results of trajectory modelling is also covered. Trajectory modelling techniques reviewed include latent class modelling approaches, ie, growth mixture modelling (GMM), group-based trajectory modelling (GBTM), latent class analysis (LCA), and latent transition analysis (LTA). A parallel is drawn to other individual-centered statistical approaches such as cluster analysis (CA) and sequence analysis (SA). Depending on the research question and type of data, a number of approaches can be used for trajectory modelling of health outcomes measured in longitudinal studies. However, the various terms to designate latent class modelling approaches (GMM, GBTM, LTA, LCA) are used inconsistently and often interchangeably in the available scientific literature. Improved consistency in the terminology and reporting guidelines have the potential to increase researchers' efficiency when it comes to choosing the most appropriate technique that best suits their research questions.

The number and composition of work hours for attending physicians in Taiwan

Long work hours among physicians is a worldwide issue in the healthcare arena. Previous studies have largely focused on the work hours of resident physicians rather than those of attending physicians. The purpose of this study was to investigate total work hours and the composition of those work hours for attending physicians across different hospital settings and across different medical specialties through a nationwide survey. This included examining differences in physician workload and its composition with respect to different hospital characteristics, and grouping medical specialties according to the work similarities. A cross-sectional self-reported nationwide survey was conducted from June to September of 2018, and the two questionnaires were distributed to all accredited hospitals in Taiwan. The number of physician work hours in different types of duty shifts were answered by medical specialty in each surveyed hospital. Each medical specialty in a hospital filled only one response for its attending physicians. The findings reveal that the average total work hours per week of an attending physician is around 69.1 h, but the total work hours and their composition of different duty shifts varied among hospital accreditation levels, geographic locations, emergency care responsibilities, and medical specialties. Because of the variance in the number and composition of attending physicians' work hours, adjusting physician work hours to a reasonable level will be a major challenge for health authority and hospital managers.

Hierarchical Clustering Analysis for Predicting 1-Year Mortality After Starting Hemodialysis

Introduction

For patients with end-stage renal disease (ESRD), due to the heterogeneity of the population, appropriate risk assessment approaches and strategies for further follow-up remain scarce. We aimed to conduct a pilot study for better risk stratification, applying machine learning–based classification to patients with ESRD who newly started maintenance hemodialysis.

Methods

We prospectively studied 101 patients with ESRD, who were new to maintenance hemodialysis therapy, between August 2016 and March 2018. Baseline values of variables such as blood and urine tests were obtained before the initiation of hemodialysis. Agglomerative hierarchical clustering was conducted with the collected continuous data. The resulting clusters were followed up for the primary outcome of 1-year mortality, as analyzed by the Kaplan-Meier survival curve with log-rank test and the Cox proportional hazard model.

Results

The participants were divided into 3 clusters (cluster 1, n = 62; cluster 2, n = 15; cluster 3, n = 24) by hierarchical clustering, using 46 clinical variables. Patients in cluster 3 showed lower systolic blood pressures, and lower serum creatinine and urinary liver-type fatty acid-binding protein levels, before the initiation of hemodialysis. Consequently, cluster 3 was associated with the highest 1-year mortality in the study cohort (P < 0.001), and the difference was significant after adjustment for age and sex (hazard ratio: 10.2; 95% confidence interval: 2.94–46.8, cluster 1 as reference).

Conclusion

In this proof-of-concept study, hierarchical clustering discovered a subgroup with a higher 1-year mortality at the initiation of hemodialysis. Applying machine learning–derived classification to patients with ESRD may contribute to better risk stratification.

Visual Analytics for Dimension Reduction and Cluster Analysis of High Dimensional Electronic Health Records

Recent advancement in EHR-based (Electronic Health Record) systems has resulted in producing data at an unprecedented rate. The complex, growing, and high-dimensional data available in EHRs creates great opportunities for machine learning techniques such as clustering. Cluster analysis often requires dimension reduction to achieve efficient processing time and mitigate the curse of dimensionality. Given a wide range of techniques for dimension reduction and cluster analysis, it is not straightforward to identify which combination of techniques from both families leads to the desired result. The ability to derive useful and precise insights from EHRs requires a deeper understanding of the data, intermediary results, configuration parameters, and analysis processes. Although these tasks are often tackled separately in existing studies, we present a visual analytics (VA) system, called Visual Analytics for Cluster Analysis and Dimension Reduction of High Dimensional Electronic Health Records (VALENCIA), to address the challenges of high-dimensional EHRs in a single system. VALENCIA brings a wide range of cluster analysis and dimension reduction techniques, integrate them seamlessly, and make them accessible to users through interactive visualizations. It offers a balanced distribution of processing load between users and the system to facilitate the performance of high-level cognitive tasks in such a way that would be difficult without the aid of a VA system. Through a real case study, we have demonstrated how VALENCIA can be used to analyze the healthcare administrative dataset stored at ICES. This research also highlights what needs to be considered in the future when developing VA systems that are designed to derive deep and novel insights into EHRs.

Lifetime Costs of Surviving Cancer-A Queensland Study (COS-Q): Protocol of a Large Healthcare Data Linkage Study

Australia-wide, there are currently more than one million cancer survivors. There are over 32 million world-wide. A trend of increasing cancer incidence, medical innovations and extended survival places growing pressure on healthcare systems to manage the ongoing and late effects of cancer treatment. There are no published studies of the long-term health service use and cost of cancer survivorship on a population basis in Australia. All residents of the state of Queensland, Australia, diagnosed with a first primary malignancy from 1997–2015 formed the cohort of interest. State and national healthcare databases are linked with cancer registry records to capture all health service utilization and healthcare costs for 20 years (or death, if this occurs first), starting from the date of cancer diagnosis, including hospital admissions, emergency presentations, healthcare costing data, Medicare services and pharmaceuticals. Data analyses include regression and economic modeling. We capture the whole journey of health service contact and estimate long-term costs of all cancer patients diagnosed and treated in Queensland by linking routinely collected state and national healthcare data. Our results may improve the understanding of lifetime health effects faced by cancer survivors and estimate related healthcare costs. Research outcomes may inform policy and facilitate future planning for the allocation of healthcare resources according to the burden of disease.

The impact of chronic kidney disease on developed countries from a health economics perspective: A systematic scoping review

Chronic kidney disease (CKD) affects over 10% of the global population and poses significant challenges for societies and health care systems worldwide. To illustrate these challenges and inform cost-effectiveness analyses, we undertook a comprehensive systematic scoping review that explored costs, health-related quality of life (HRQoL) and life expectancy (LE) amongst individuals with CKD. Costs were examined from a health system and societal perspective, and HRQoL was assessed from a societal and patient perspective. Papers published in English from 2015 onward found through a systematic search strategy formed the basis of the review. All costs were adjusted for inflation and expressed in US$ after correcting for purchasing power parity. From the health system perspective, progression from CKD stages 1-2 to CKD stages 3a-3b was associated with a 1.1-1.7 fold increase in per patient mean annual health care cost. The progression from CKD stage 3 to CKD stages 4-5 was associated with a 1.3-4.2 fold increase in costs, with the highest costs associated with end-stage renal disease at $20,110 to $100,593 per patient. Mean EuroQol-5D index scores ranged from 0.80 to 0.86 for CKD stages 1-3, and decreased to 0.73-0.79 for CKD stages 4-5. For treatment with renal replacement therapy, transplant recipients incurred lower costs and demonstrated higher HRQoL scores with longer LE compared to dialysis patients. The study has provided a comprehensive updated overview of the burden associated with different CKD stages and renal replacement therapy modalities across developed countries. These data will be useful for the assessment of new renal services/therapies in terms of cost-effectiveness.

Curcumin Enhances Radiosensitization of Nasopharyngeal Carcinoma via Mediating Regulation of Tumor Stem-like Cells by a CircRNA Network

Circular RNAs (circRNAs) are involved in cancer development via inhibition of miRNAs, which are associated with differentiation, proliferation, migration, and carcinogenicity. Curcumin has antioxidant and anti-cancer properties, and it has also been used as a radiosensitizer. In this study, we explored the potential relationships among curcumin, circRNAs, and nasopharyngeal carcinoma (NPC). We compared the differences in circRNA levels in NPC cell lines after radiotherapy and after treatment with curcumin, using a high-throughput microarray. Further, a circRNA-miRNA-mRNA interaction network between radiation resistance NPC cell lines and tumor stem cells was constructed by applying bioinformatics. Finally, it was demonstrated by reverse transcription-quantitative polymerase chain reaction assay and wound healing assay that curcumin could enhance radiosensitization of NPC cell lines via mediating regulation of tumor stem-like cells by the "hsa_circRNA_102115"-"hsa-miR-335-3p"-"MAPK1" interaction network.

Preoperative Behavioral Health, Opioid, and Antidepressant Utilization and 2-year Costs After Spinal Fusion-Revelations From Cluster Analysis

STUDY DESIGN

Retrospective administrative claims database analysis.

OBJECTIVE

Identify distinct presurgery health care resource utilization (HCRU) patterns among posterior lumbar spinal fusion patients and quantify their association with postsurgery costs.

SUMMARY OF BACKGROUND DATA

Presurgical HCRU may be predictive of postsurgical economic outcomes and help health care providers to identify patients who may benefit from innovation in care pathways and/or surgical approach.

METHODS

Privately insured patients who received one- to two-level posterior lumbar spinal fusion between 2007 and 2016 were identified from a claims database. Agglomerative hierarchical clustering (HC), an unsupervised machine learning technique, was used to cluster patients by presurgery HCRU across 90 resource categories. A generalized linear model was used to compare 2-year postoperative costs across clusters controlling for age, levels fused, spinal diagnosis, posterolateral/interbody approach, and Elixhauser Comorbidity Index.

RESULTS

Among 18,770 patients, 56.1% were female, mean age was 51.3, 79.4% had one-level fusion, and 89.6% had inpatient surgery. Three patient clusters were identified: Clust1 (n = 13,987 [74.5%]), Clust2 (n = 4270 [22.7%]), Clust3 (n = 513 [2.7%]). The largest between-cluster differences were found in mean days supplied for antidepressants (Clust1: 97.1 days, Clust2: 175.2 days, Clust3: 287.1 days), opioids (Clust1: 76.7 days, Clust2: 166.9 days, Clust3: 129.7 days), and anticonvulsants (Clust1: 35.1 days, Clust2: 67.8 days, Clust3: 98.7 days). For mean medical visits, the largest between-cluster differences were for behavioral health (Clust1: 0.14, Clust2: 0.88, Clust3: 16.3) and nonthoracolumbar office visits (Clust1: 7.8, Clust2: 13.4, Clust3: 13.8). Mean (95% confidence interval) adjusted 2-year all-cause postoperative costs were lower for Clust1 ($34,048 [$33,265-$34,84]) versus both Clust2 ($52,505 [$50,306-$54,800]) and Clust3 ($48,452 [$43,007-$54,790]), P < 0.0001.

CONCLUSION

Distinct presurgery HCRU clusters were characterized by greater utilization of antidepressants, opioids, and behavioral health services and these clusters were associated with significantly higher 2-year postsurgical costs.

LEVEL OF EVIDENCE

3.

Four Subgroups of Blood Stasis Syndrome Are Identified by Manifestation Cluster Analysis in Males

Blood stasis syndrome (BSS) is an important pathological condition in traditional East Asian medicine and is associated with ischemic heart disease, cerebral vascular accident, diabetes mellitus, chronic renal failure, severe traumatic injury, and dysmenorrhea. However, previous studies have been unable to reveal the clinical and biological characteristics or biological markers of BSS. We hypothesized that the heterogeneity among the manifestations of BSS or non-BSS could interfere with an analysis to describe the characteristics of BSS. In this study, male participants based on the severity of BSS-associated symptoms and signs were clustered and classified into four subgroups: BSS subgroups (1), (2), (3), and (4). Non-BSS core subgroup was redefined using manifestation cluster analysis. Biological characteristics of subgroups BSS(1) and BSS(2) belong to the range of the non-BSS core subgroup (1), whereas that of subgroups BSS(3) and BSS(4) are characterized by different biological parameters such as systemic inflammatory conditions and elevated D-dimer level. Our results suggested that patients in subgroups of BSS(3) and BSS(4) are more likely to be exposed in an inflammatory state than other BSS subgroups. We found the heterogeneity among the manifestations which could mask the characteristics of BSS and identified the clinical and biological profiles of the four BSS subgroups through comparisons of the redefined non-BSS and BSS subgroups. This finding could provide accurate diagnostic criteria and new approaches for BSS treatments in different subgroups.

A fuzzy approach for analysing equitable and sustainable well-being in Italian regions

OBJECTIVES

Recently, the Italian Institute of Statistics (ISTAT) and the National Council for Economy and Labor (CNEL) have proposed a measure for the equitable and sustainable well-being called the BES ("Benessere Equo e Sostenibile"). This paper aims to propose an original application of the fuzzy k-means approach to providing an analysis of the Italian regions according to their BES.

METHODS

The fuzzy k-means algorithm was used for clustering the Italian regions according to BES data 2015. Afterwards, a principal component analysis was conducted to show and interpret the results.

RESULTS

There is a clear difference between the regions of the North and the South. The only exceptions are represented by Lazio and Abruzzo, which belong to both groups with almost equal degrees of truth. Moreover, Trentino-Alto Adige and Valle d'Aosta exhibit the best condition, whilst Molise is the worst region.

CONCLUSIONS

This study reveals that some Italian regions are in a state of backwardness regarding health, environment, minimum economic conditions, subjective well-being, education, employment conditions, social relationships, and working conditions. Therefore, institutions should consider local policies to address these issues.

The unique composition of Indian gut microbiome, gene catalogue, and associated fecal metabolome deciphered using multi-omics approaches

BACKGROUND

Metagenomic studies carried out in the past decade have led to an enhanced understanding of the gut microbiome in human health; however, the Indian gut microbiome has not been well explored. We analyzed the gut microbiome of 110 healthy individuals from two distinct locations (North-Central and Southern) in India using multi-omics approaches, including 16S rRNA gene amplicon sequencing, whole-genome shotgun metagenomic sequencing, and metabolomic profiling of fecal and serum samples.

RESULTS

The gene catalogue established in this study emphasizes the uniqueness of the Indian gut microbiome in comparison to other populations. The gut microbiome of the cohort from North-Central India, which was primarily consuming a plant-based diet, was found to be associated with Prevotella and also showed an enrichment of branched chain amino acid (BCAA) and lipopolysaccharide biosynthesis pathways. In contrast, the gut microbiome of the cohort from Southern India, which was consuming an omnivorous diet, showed associations with Bacteroides, Ruminococcus, and Faecalibacterium and had an enrichment of short chain fatty acid biosynthesis pathway and BCAA transporters. This corroborated well with the metabolomics results, which showed higher concentration of BCAAs in the serum metabolome of the North-Central cohort and an association with Prevotella. In contrast, the concentration of BCAAs was found to be higher in the fecal metabolome of the Southern-India cohort and showed a positive correlation with the higher abundance of BCAA transporters.

CONCLUSIONS

The study reveals the unique composition of the Indian gut microbiome, establishes the Indian gut microbial gene catalogue, and compares it with the gut microbiome of other populations. The functional associations revealed using metagenomic and metabolomic approaches provide novel insights on the gut-microbe-metabolic axis, which will be useful for future epidemiological and translational researches.

Morphological classification of melanoma metastasis with reflectance confocal microscopy

BACKGROUND

Cutaneous malignant melanoma metastases differential diagnosis is challenging, as clinical and dermoscopic features can simulate primary melanoma or other benign or malignant skin neoplasms, and in-vivo reflectance confocal microscopy could assist. Our aim was to identify specific reflectance confocal microscopy features for cutaneous malignant melanoma metastases, and epidermal and dermal involvement.

METHODS

A retrospective, multicentre observational study of lesions with proven cutaneous malignant melanoma metastases diagnosis between January 2005 and December 2016. Lesions were retrospectively assessed according to morphological features observed at reflectance confocal microscopy. Potential homogeneous subgroups of epidermal or dermal involvement were investigated with cluster analysis.

RESULTS

Cutaneous malignant melanoma metastases (51 lesions in 29 patients) exhibited different frequencies of features according to metastasis dermoscopy patterns. Lesions classified at dermoscopy with nevus-like globular and non-globular patterns were more likely to be epidermotropic, showing characteristics of epidermal and dermal involvement at reflectance confocal microscopy. Other dermoscopy pattern classifications were more likely to be dermotropic, showing characteristics od dermal involvement at reflectance confocal microscopy. Distinguishing features at reflectance confocal microscopy included irregular (78%) and altered (63%) epidermis, pagetoid infiltration (51%), disarranged junctional architecture (63%), non-edged papillae (76%), dense and sparse, and cerebriform nests in the upper dermis (74%), and vascularity (51%). Cluster analysis identified three groups, which were retrospectively correlated with histopathological diagnoses of dermotropic and epidermotropic diagnoses (P < 0.001). The third cluster represents lesions with deep dermis morphological changes, which were too deep for evaluation with reflectance confocal microscopy.

CONCLUSIONS

Specific reflectance confocal microscopy features of cutaneous malignant melanoma metastases for correct diagnosis, and subtype diagnosis, seem achievable in most cases where morphological alterations are located above the deep dermis.

Multimorbidity patterns with K-means nonhierarchical cluster analysis

BACKGROUND

The purpose of this study was to ascertain multimorbidity patterns using a non-hierarchical cluster analysis in adult primary patients with multimorbidity attended in primary care centers in Catalonia.

METHODS

Cross-sectional study using electronic health records from 523,656 patients, aged 45-64 years in 274 primary health care teams in 2010 in Catalonia, Spain. Data were provided by the Information System for the Development of Research in Primary Care (SIDIAP), a population database. Diagnoses were extracted using 241 blocks of diseases (International Classification of Diseases, version 10). Multimorbidity patterns were identified using two steps: 1) multiple correspondence analysis and 2) k-means clustering. Analysis was stratified by sex.

RESULTS

The 408,994 patients who met multimorbidity criteria were included in the analysis (mean age, 54.2 years [Standard deviation, SD: 5.8], 53.3% women). Six multimorbidity patterns were obtained for each sex; the three most prevalent included 68% of the women and 66% of the men, respectively. The top cluster included coincident diseases in both men and women: Metabolic disorders, Hypertensive diseases, Mental and behavioural disorders due to psychoactive substance use, Other dorsopathies, and Other soft tissue disorders.

CONCLUSION

Non-hierarchical cluster analysis identified multimorbidity patterns consistent with clinical practice, identifying phenotypic subgroups of patients.

Identifying patterns of general practitioner service utilisation and their relationship with potentially preventable hospitalisations in people with diabetes: The utility of a cluster analysis approach

AIMS

We aimed to characterise use of general practitioners (GP) simultaneously across multiple attributes in people with diabetes and examine its impact on diabetes related potentially preventable hospitalisations (PPHs).

METHODS

Five-years of panel data from 40,625 adults with diabetes were sourced from Western Australian administrative health records. Cluster analysis (CA) was used to group individuals with similar patterns of GP utilisation characterised by frequency and recency of services. The relationship between GP utilisation cluster and the risk of PPHs was examined using multivariable random-effects negative binomial regression.

RESULTS

CA categorised GP utilisation into three clusters: moderate; high and very high usage, having distinct patient characteristics. After adjusting for potential confounders, the rate of PPHs was significantly lower across all GP usage clusters compared with those with no GP usage; IRR = 0.67 (95%CI: 0.62-0.71) among the moderate, IRR = 0.70 (95%CI 0.66-0.73) high and IRR = 0.76 (95%CI 0.72-0.80) very high GP usage clusters.

CONCLUSIONS

Combination of temporal factors with measures of frequency of use of GP services revealed patterns of primary health care utilisation associated with different underlying patient characteristics. Incorporation of multiple attributes, that go beyond frequency-based approaches may better characterise the complex relationship between use of GP services and diabetes-related hospitalisation.

Multimorbidity patterns in the elderly: a prospective cohort study with cluster analysis

BACKGROUND

Multimorbidity is the coexistence of more than two chronic diseases in the same individual; however, there is no consensus about the best definition. In addition, few studies have described the variability of multimorbidity patterns over time. The aim of this study was to identify multimorbidity patterns and their variability over a 6-year period in patients older than 65 years attended in primary health care.

METHODS

A cohort study with yearly cross-sectional analysis of electronic health records from 50 primary health care centres in Barcelona. Selected patients had multimorbidity and were 65 years of age or older in 2009. Diagnoses (International Classification of Primary Care, second edition) were extracted using O'Halloran criteria for chronic diseases. Multimorbidity patterns were identified using two steps: 1) multiple correspondence analysis and 2) k-means clustering. Analysis was stratified by sex and age group (65-79 and ≥80 years) at the beginning of the study period.

RESULTS

Analysis of 2009 electronic health records from 190,108 patients with multimorbidity (59.8% women) found a mean age of 71.8 for the 65-79 age group and 84.16 years for those over 80 (Standard Deviation [SD] 4.35 and 3.46, respectively); the median number of chronic diseases was seven (Interquartil range [IQR] 5-10). We obtained 6 clusters of multimorbidity patterns (1 nonspecific and 5 specifics) in each group, being the specific ones: Musculoskeletal, Endocrine-metabolic, Digestive/Digestive-respiratory, Neurological, and Cardiovascular patterns. A minimum of 42.5% of the sample remained in the same pattern at the end of the study, reflecting the stability of these patterns.

CONCLUSIONS

This study identified six multimorbidity patterns per each group, one nonnspecific pattern and five of them with a specific pattern related to an organic system. The multimorbidity patterns obtained had similar characteristics throughout the study period. These data are useful to improve clinical management of each specific subgroup of patients showing a particular multimorbidity pattern.

Assessing the relationship between toxicity and economic cost of oncological target agents: A systematic review of clinical trials

Target agents are peculiar oncological drugs which differ from the traditional therapies in their ability of recognizing specific molecules expressed by tumor cells and microenvironment. Thus, their toxicity is generally lower than that associated to chemotherapy, and they represent nowadays a new standard of care in a number of tumors. This paper deals with the relationship between economic costs and toxicity of target agents. At this aim, a cluster analysis-based exploration of the main features of a large collection of them is carried out, with a specific focus on the variables leading to the identification of their toxicity and related costs. The analysis of the toxicity is based on the Severe Adverse Events (SAE) and Discontinuation (D) rates of each target agent considering data published on PubMed from 1965 to 2016 in the phase II and III studies that have led to the approval of these drugs for cancer patients by US Food and Drug Administration. The construction of the dataset represents a key step of the research, and is grounded on the critical analysis of a wide set of clinical studies. In order to capture different evaluation strategies of the toxicity, clustering is performed according to three different criteria (including Voronoi tessellation). Our procedure allows us to identify 5 different groups of target agents pooled by similar SAE and D rates and, at the same time, 3 groups based on target agents' costs for 1 month and for the median whole duration of therapy. Results highlight several specific regularities for toxicity and costs. This study present several limitations, being realized starting from clinical trials and not from individual patients' data. However, a macroscopic perspective suggests that costs are rather heterogeneous, and they do not clearly follow the clustering based on SAE and D rates.