A framework for multiple imputation in cluster analysis

Data-driven subclassification of ANCA-associated vasculitis: model-based clustering of a federated international cohort

BACKGROUND

Antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis is a heterogenous autoimmune disease. While traditionally stratified into two conditions, granulomatosis with polyangiitis (GPA) and microscopic polyangiitis (MPA), the subclassification of ANCA-associated vasculitis is subject to continued debate. Here we aim to identify phenotypically distinct subgroups and develop a data-driven subclassification of ANCA-associated vasculitis, using a large real-world dataset.

METHODS

In the collaborative data reuse project FAIRVASC (Findable, Accessible, Interoperable, Reusable, Vasculitis), registry records of patients with ANCA-associated vasculitis were retrieved from six European vasculitis registries: the Czech Registry of ANCA-associated vasculitis (Czech Republic), the French Vasculitis Study Group Registry (FVSG; France), the Joint Vasculitis Registry in German-speaking Countries (GeVas; Germany), the Polish Vasculitis Registry (POLVAS; Poland), the Irish Rare Kidney Disease Registry (RKD; Ireland), and the Skåne Vasculitis Cohort (Sweden). We performed model-based clustering of 17 mixed-type clinical variables using a parsimonious mixture of two latent Gaussian variable models. Clinical validation of the optimal cluster solution was made through summary statistics of the clusters' demography, phenotypic and serological characteristics, and outcome. The predictive value of models featuring the cluster affiliations were compared with classifications based on clinical diagnosis and ANCA specificity. People with lived experience were involved throughout the FAIRVASVC project.

FINDINGS

A total of 3868 patients diagnosed with ANCA-associated vasculitis between Nov 1, 1966, and March 1, 2023, were included in the study across the six registries (Czech Registry n=371, FVSG n=1780, GeVas n=135, POLVAS n=792, RKD n=439, and Skåne Vasculitis Cohort n=351). There were 2434 (62·9%) patients with GPA and 1434 (37·1%) with MPA. Mean age at diagnosis was 57·2 years (SD 16·4); 2006 (51·9%) of 3867 patients were men and 1861 (48·1%) were women. We identified five clusters, with distinct phenotype, biochemical presentation, and disease outcome. Three clusters were characterised by kidney involvement: one severe kidney cluster (555 [14·3%] of 3868 patients) with high C-reactive protein (CRP) and serum creatinine concentrations, and variable ANCA specificity (SK cluster); one myeloperoxidase (MPO)-ANCA-positive kidney involvement cluster (782 [20·2%]) with limited extrarenal disease (MPO-K cluster); and one proteinase 3 (PR3)-ANCA-positive kidney involvement cluster (683 [17·7%]) with widespread extrarenal disease (PR3-K cluster). Two clusters were characterised by relative absence of kidney involvement: one was a predominantly PR3-ANCA-positive cluster (1202 [31·1%]) with inflammatory multisystem disease (IMS cluster), and one was a cluster (646 [16·7%]) with predominantly ear-nose-throat involvement and low CRP, with mainly younger patients (YR cluster). Compared with models fitted with clinical diagnosis or ANCA status, cluster-assigned models demonstrated improved predictive power with respect to both patient and kidney survival.

INTERPRETATION

Our study reinforces the view that ANCA-associated vasculitis is not merely a binary construct. Data-driven subclassification of ANCA-associated vasculitis exhibits higher predictive value than current approaches for key outcomes.

FUNDING

European Union's Horizon 2020 research and innovation programme under the European Joint Programme on Rare Diseases.

Cognitive heterogeneity in Attention Deficit Hyperactivity Disorder: Implications for maths

This study investigated whether cognitive function better predicted maths test performance than a clinical diagnosis of attention deficit hyperactivity disorder (ADHD). Forty-four drug naïve children (Mage = 101.34 months, SD = 19.39; 30% girls) were recruited from clinical ADHD referral waiting lists. Children underwent assessment of Executive Functions (EF), lower-level cognitive processes, and maths performance. Children were grouped using a categorical approach comprising (1) children with a clinical ADHD diagnosis and (2) children without a diagnosis (i.e., subthreshold ADHD). Secondly, hierarchical cluster analysis generated subgroups of children using EF scores. Children were compared on cognition, maths, and parent-rated symptoms of ADHD and co-occurring difficulties. Children's diagnostic outcomes did not differentiate maths performance. By contrast, EF subgroups generated meaningful cognitive clusters which differentiated maths test scores. This suggests that cognitive patterns of performance, rather than children's diagnostic outcomes, are more informative for identifying meaningful groups with variable maths performance which has implications for remedial support.

Machine learning-based phenogroups and prediction model in patients with functional gastrointestinal disorders to reveal distinct disease subsets associated with gas production

Background and Objectives

Symptom-based subtyping for functional gastrointestinal disorders (FGIDs) has limited value in identifying underlying mechanisms and guiding therapeutic strategies. Small intestinal dysbiosis is implicated in the development of FGIDs. We tested if machine learning (ML) algorithms utilizing both gastrointestinal (GI) symptom characteristics and lactulose breath tests could provide distinct clusters.

Materials and Methods

This was a prospective cohort study. We performed lactulose hydrogen methane breath tests and hydrogen sulfide breath tests in 508 patients with GI symptoms. An unsupervised ML algorithm was used to categorize subjects by integrating GI symptoms and breath gas characteristics. Generalized Estimating Equation (GEE) models were used to examine the longitudinal associations between cluster patterns and breath gas time profiles. An ML-based prediction model for identifying excessive gas production in FGIDs patients was developed and internal validation was performed.

Results

FGIDs were confirmed in 300 patients. K-means clustering identified 4 distinct clusters. Cluster 2, 3, and 4 showed enrichments for abdominal distention and diarrhea with a high proportion of excessive gas production, whereas Cluster 1 was characterized by moderate lower abdominal discomforts with the most psychological complaints and the lowest proportion of excessive gas production. GEE models showed that breath gas concentrations varied among different clusters over time. We further sought to develop an ML-based prediction model to determine excessive gas production. The model exhibited good predictive capabilities.

Conclusion

ML-based phenogroups and prediction model approaches could provide distinct FGIDs subsets and efficiently determine FGIDs subsets with greater gas production, thereby facilitating clinical decision-making and guiding treatment.

Blood biomarker profiles and exceptional longevity: comparison of centenarians and non-centenarians in a 35-year follow-up of the Swedish AMORIS cohort

Comparing biomarker profiles measured at similar ages, but earlier in life, among exceptionally long-lived individuals and their shorter-lived peers can improve our understanding of aging processes. This study aimed to (i) describe and compare biomarker profiles at similar ages between 64 and 99 among individuals eventually becoming centenarians and their shorter-lived peers, (ii) investigate the association between specific biomarker values and the chance of reaching age 100, and (iii) examine to what extent centenarians have homogenous biomarker profiles earlier in life. Participants in the population-based AMORIS cohort with information on blood-based biomarkers measured during 1985-1996 were followed in Swedish register data for up to 35 years. We examined biomarkers of metabolism, inflammation, liver, renal, anemia, and nutritional status using descriptive statistics, logistic regression, and cluster analysis. In total, 1224 participants (84.6% females) lived to their 100th birthday. Higher levels of total cholesterol and iron and lower levels of glucose, creatinine, uric acid, aspartate aminotransferase, gamma-glutamyl transferase, alkaline phosphatase, lactate dehydrogenase, and total iron-binding capacity were associated with reaching 100 years. Centenarians overall displayed rather homogenous biomarker profiles. Already from age 65 and onwards, centenarians displayed more favorable biomarker values in commonly available biomarkers than individuals dying before age 100. The differences in biomarker values between centenarians and non-centenarians more than one decade prior death suggest that genetic and/or possibly modifiable lifestyle factors reflected in these biomarker levels may play an important role for exceptional longevity.

Structural determinants and cardiometabolic typologies related to frailty in community-dwelling older adults

Frailty is a geriatric syndrome linked to adverse outcomes. Co-occurring cardiometabolic factors increase frailty risk; however, their distinct combinations (typologies) associated with frailty are unclear. We aimed to identify subgroups of older adults with distinct cardiometabolic typologies and characterize their relationship with structural determinants and frailty to inform tailored approaches to prevent and delay frailty. This study was cross-sectional design and included 7984 community-dwelling older adults (65+ years) enrolled in the Health and Retirement Study (2006 and 2008). Latent class analysis was performed using seven cardiometabolic indicators (abdominal obesity, obesity, low high-density lipoprotein; and elevated blood pressure, blood sugar, total cholesterol, C-reactive protein). Frailty was indicated by ≥3 features (weakness, slowness, fatigue, low physical activity, unintentional weight loss). Logistic regression was used to examine the relationship between structural determinants (gender, race/ethnicity, and education), cardiometabolic typologies, and frailty. Three cardiometabolic subgroups were identified: insulin-resistant (n = 3547), hypertensive dyslipidemia (n = 1246), and hypertensive (n = 3191). Insulin-resistant subgroup members were more likely to be female, non-Hispanic Black, and college non-graduates; hypertensive dyslipidemia subgroup members were more likely to be non-Hispanic Others and report high school education; and hypertensive subgroup members were more likely to be male and college educated (p≤.05). Frailty risk was higher for females, Hispanic or Non-Hispanic Black older adults, and those with lower education (p≤.001). Frailty risk was greater in the insulin-resistant compared to the other subgroups (both aOR=2.0, both p<.001). Findings highlight a need to design tailored interventions targeting cardiometabolic typologies to prevent and delay frailty.

Associations between combined urban and lifestyle factors and respiratory health in European children

INTRODUCTION

Previous studies identified some environmental and lifestyle factors independently associated with children respiratory health, but few focused on exposure mixture effects. This study aimed at identifying, in pregnancy and in childhood, combined urban and lifestyle environment profiles associated with respiratory health in children.

METHODS

This study is based on the European Human Early-Life Exposome (HELIX) project, combining six birth cohorts. Associations between profiles of pregnancy (38 exposures) and childhood (84 exposures) urban and lifestyle factors, identified by clustering analysis, and respiratory health were estimated by regression models adjusted for confounders.

RESULTS

Among the 1033 included children (mean ± standard-deviation (SD) age: 8.2 ± 1.6 years old, 47% girls) the mean ± SD forced expiratory volume in 1s (FEV1) and forced vital capacity (FVC) were 99 ± 13% and 101 ± 14%, respectively, and 12%, 12% and 24% reported ever-asthma, wheezing and rhinitis, respectively. Four profiles of pregnancy exposures and four profiles of childhood exposures were identified. Compared to the reference childhood exposure profile (low exposures), two exposure profiles were associated with lower levels of FEV1. One profile was characterized by few natural spaces in the surroundings and high exposure to the built environment and road traffic. The second profile was characterized by high exposure to meteorological factors and low levels of all other exposures and was also associated with an increased risk of ever-asthma and wheezing. A pregnancy exposure profile characterized by high exposure levels to all risk factors, but a healthy maternal lifestyle, was associated with a lower risk of wheezing and rhinitis in children, compared to the reference pregnancy profile (low exposures).

CONCLUSION

This comprehensive approach revealed pregnancy and childhood profiles of urban and lifestyle exposures associated with lung function and/or respiratory conditions in children. Our findings highlight the need to pursue the study of combined exposures to improve prevention strategies for multifactorial diseases such as asthma.

Identification of obstructive sleep apnea in children with obesity: A cluster analysis approach

BACKGROUND

Obstructive sleep apnea (OSA) is a heterogeneous disorder with a prevalence of 25%-60% in children with obesity. There is a lack of diagnostic tools to identify those at high risk for OSA.

METHOD

Children with obesity, aged 8-19 years old, were enrolled into an ongoing multicenter, prospective cohort study related to OSA. We performed k-means cluster analysis to identify clinical variables which could help identify obesity related OSA.

RESULTS

In this study, 118 participants were included in the analysis; 40.7% were diagnosed with OSA, 46.6% were female and the mean (SD) body mass index (BMI) and age were 39.7 (9.6) Kg/m², and 14.4 (2.6) years, respectively. The mean (SD) obstructive apnea-hypopnea index (OAHI) was 11.0 (21.1) events/h. We identified two distinct clusters based on three clustering variables (age, BMI z-score, and neck-height ratio [NHR]). The prevalence of OSA in clusters 1 and 2, were 22.4% and 58.3% (p = 0.001), respectively. Children in cluster 2, in comparison to cluster 1, had higher BMI z-score (4.7 (1.1) versus 3.2 (0.7), p < 0.001), higher NHR (0.3 (0.02) versus 0.2 (0.01), p < 0.001) and were older (15.0 (2.2) versus 13.7 (2.9) years, p = 0.09), respectively. However, there were no significant differences in sex and OSA symptoms between the clusters. The results from hierarchical clustering were similar to k-means analysis suggesting that the resulting OSA clusters were stable to different analysis approaches.

INTERPRETATION

BMI, NHR, and age are easily obtained in a clinical setting and can be utilized to identify children at high risk for OSA.

Identification of eczema clusters and their association with filaggrin and atopic comorbidities: analysis of five birth cohorts

BACKGROUND

Longitudinal modelling of the presence/absence of current eczema through childhood has identified similar phenotypes, but their characteristics often differ between studies.

OBJECTIVES

To demonstrate that a more comprehensive description of longitudinal pattern of symptoms may better describe trajectories than binary information on eczema presence.

METHODS

We derived six multidimensional variables of eczema spells from birth to 18 years of age (including duration, temporal sequencing and the extent of persistence/recurrence). Spells were defined as consecutive observations of eczema separated by no eczema across 5 epochs in five birth cohorts: infancy (first year); early childhood (age 2-3 years); preschool/early school age (4-5 years); middle childhood (8-10 years); adolescence (14-18 years). We applied Partitioning Around Medoids clustering on these variables to derive clusters of the temporal patterns of eczema. We then investigated the stability of the clusters, within-cluster homogeneity and associated risk factors, including FLG mutations.

RESULTS

Analysis of 7464 participants with complete data identified five clusters: (i) no eczema (51.0%); (ii) early transient eczema (21.6%); (iii) late-onset eczema (LOE; 8.1%); (iv) intermittent eczema (INT; 7.5%); and (v) persistent eczema (PE; 11.8%). There was very-high agreement between the assignment of individual children into clusters when using complete or imputed (n = 15 848) data (adjusted Rand index = 0.99; i.e. the clusters were very stable). Within-individual symptom patterns across clusters confirmed within-cluster homogeneity, with consistent patterns of symptoms among participants within each cluster and no overlap between the clusters. Clusters were characterized by differences in associations with risk factors (e.g. parental eczema was associated with all clusters apart from LOE; sensitization to inhalant allergens was associated with all clusters, with the highest risk in the PE cluster). All clusters apart from LOE were associated with FLG mutations. Of note, the strongest association was for PE [relative risk ratio (RRR) 2.70, 95% confidence interval (CI) 2.24-3.26; P < 0.001] followed by INT (RRR 2.29, 95% CI 1.82-2.88; P < 0.001).

CONCLUSIONS

Clustering of multidimensional variables identified stable clusters with different genetic architectures. Using multidimensional variables may capture eczema development and derive stable and internally homogeneous clusters. However, deriving homogeneous symptom clusters does not necessarily mean that these are underpinned by completely unique mechanisms.

An overview of clustering methods with guidelines for application in mental health research

Cluster analyzes have been widely used in mental health research to decompose inter-individual heterogeneity by identifying more homogeneous subgroups of individuals. However, despite advances in new algorithms and increasing popularity, there is little guidance on model choice, analytical framework and reporting requirements. In this paper, we aimed to address this gap by introducing the philosophy, design, advantages/disadvantages and implementation of major algorithms that are particularly relevant in mental health research. Extensions of basic models, such as kernel methods, deep learning, semi-supervised clustering, and clustering ensembles are subsequently introduced. How to choose algorithms to address common issues as well as methods for pre-clustering data processing, clustering evaluation and validation are then discussed. Importantly, we also provide general guidance on clustering workflow and reporting requirements. To facilitate the implementation of different algorithms, we provide information on R functions and libraries.

Phenotyping for percutaneous coronary intervention and long-term recurrent weighted outcomes

INTRODUCTION

Percutaneous coronary interventions (PCI) are often performed in multimorbid patients with heterogeneous characteristics and variable clinical outcomes. We aimed to identify distinct clinical phenotypes utilizing machine learning and explore their relationship with long-term recurrent and weighted outcomes.

METHODS

This prospective observational cohort study enrolled all-comer PCI patients in 2020-2021. Multiple imputation k-means clustering was utilized to detect specific phenotypes. The study endpoints were patient-oriented and device oriented composite endpoints (POCE, DOCE), its individual components, and major bleeding. We applied semiparametric regression models for recurrent and weighted endpoints.

RESULTS

The study included a total of 643 patients. We unveiled three phenotype clusters: 1) inflammatory (n = 44, with high white blood cell counts, high values of C-reactive protein (CRP) and neutrophil-to-lymphocyte ratio), 2) high erythrocyte sedimentation rate (ESR) (n = 204), and 3) non-inflammatory (n = 395). For ACS-only population, we four distinct phenotypes (high-CRP, high-ESR, high aspartate-aminotransferase, and normal). For all-comer PCI patients, identified phenotypes had a higher risk of POCE (mean ratio (MR) 1.42 (95% confidence interval (CI) 1.11-1.81) and MR 2.01 (95% CI 1.58-2.56), respectively), DOCE (MR 1.61 (95% CI 1.20-2.16), MR 2.60 (95%CI 1.94-3.48), respectively), and stroke (hazard ratio (HR) 2.86 (95% CI 1.10-7.4), 6.83 (95% CI 2.01-23.2)). Similarly, high-ESR and high-CRP phenotypes of ACS patients were significantly associated with the development of clinical composite outcomes.

CONCLUSION

Machine learning unveiled three distinct phenotype clusters in patients after PCI that were linked with the risk of recurrent and weighted clinical endpoints. German Clinical Trial Registry number: DRKS00020892.

Heterogeneity of asthma with nasal polyposis phenotypes: A cluster analysis

BACKGROUND

Chronic rhinosinusitis with nasal polyposis (CRSwNP) affects a significant number of asthmatic patients and is notably associated with a more difficult-to-control asthma and marked inflammation. We need more studies on this specific asthma phenotype and its possible subphenotypes, in order to better individualize treatments.

AIM

The aim of this study is to identify and characterize subphenotypes of asthma patients with CRSwNP using clinical, physiological and inflammatory variables.

METHODS

K-means cluster analysis was performed on 17 clinical, physiological, and inflammatory variables from 1263 patients of all asthma severity and on a subpopulation of patients with asthma and CRSwNP. Study was registered on ClinicalTrials.gov (NCT03694847).

RESULTS

On the overall population, three groups were identified. Cluster T1 (n = 708) are young, have a short asthma duration and a low prevalence of CRSwNP. Cluster T2 (n = 263) have the longest asthma duration and Cluster T3 (n = 292) are older with the shortest asthma duration. Patients in Clusters T2 and T3 have similar prevalences of CRSwNP. On the subpopulation of asthma with CRSwNP, three clusters were also identified. Cluster S1 (n = 83) have mild-to-moderate asthma with normal lung function. Clusters S2 (N = 53) and S3 (N = 42) include patients with severe asthma and decreased lung function, but those in Cluster S2 have a longer asthma duration, whereas those Cluster S3 have late-onset asthma.

CONCLUSIONS

Despite coexistence of asthma and CRSwNP, not all patients have the same evolution of their asthma. Different phenotypes of asthma with CRSwNP can be identified and exploration of the characteristics of these subgroups could lead to a better individualized, targeted management.

Multiobjective semisupervised learning with a right-censored endpoint adapted to the multiple imputation framework

Semisupervised learning aims to use additional knowledge in the search for data structure. In clinical applications, including predictive information in the construction of a data-driven classification is of major importance. This work was motivated by a study that aimed to identify different patterns of immune parameters that would be associated with relapse-free survival in a cohort of breast cancer patients. Supervised and unsupervised objectives can be concomitantly optimized using multiobjective optimization. We propose such a procedure that addresses two challenges in the semisupervised approach, that is, missing data and additional knowledge based on survival time. The former was handled by using multiple imputation and consensus clustering. Survival information was incorporated in the supervised objective through the estimation of a cross-validation error of a Cox regression. A simulation study was performed to assess the performance of the proposed procedure. On complete datasets, the performances were compared to those of an existing modified multiobjective semisupervised learning method. The added value of including the survival data in the learning process was assessed by comparing the procedure to unsupervised learning. The proposed procedure showed better performance than the existing method, notably in the selection of the number of clusters. On incomplete datasets, the procedure showed little sensitivity to most of its parameters, even though a high number of imputations and partition initialization seeds improved the performance. The performance was degraded with a high proportion of missing data (40%) and with more ambiguous data structures. Simulation results and application on real data support the conclusion that our procedure enables the construction of a classification associated with a right-censored endpoint on a possibly incomplete dataset.

Machine learning of COVID-19 clinical data identifies population structures with therapeutic potential

Clinical outcomes for patients with COVID-19 are heterogeneous and there is interest in defining subgroups for prognostic modeling and development of treatment algorithms. We obtained 28 demographic and laboratory variables in patients admitted to hospital with COVID-19. These comprised a training cohort (n = 6099) and two validation cohorts during the first and second waves of the pandemic (n = 996; n = 1011). Uniform manifold approximation and projection (UMAP) dimension reduction and Gaussian mixture model (GMM) analysis was used to define patient clusters. 29 clusters were defined in the training cohort and associated with markedly different mortality rates, which were predictive within confirmation datasets. Deconvolution of clinical features within clusters identified unexpected relationships between variables. Integration of large datasets using UMAP-assisted clustering can therefore identify patient subgroups with prognostic information and uncovers unexpected interactions between clinical variables. This application of machine learning represents a powerful approach for delineating disease pathogenesis and potential therapeutic interventions.

Multi-sided platforms in competitive B2B networks with varying governmental influence - a taxonomy of Port and Cargo Community System business models

Our knowledge on differences in business model characteristics of thriving and failing Multi-Sided Platforms in competitive B2B networks (B2B-MSP) and potential influences of increasing governmental involvement remains fragmented. This study develops a taxonomy to classify special B2B-MSP with varying governmental influence in the supply chain and transportation context, viz. Port and Cargo Community Systems (CS). Based on the classification of 44 international CS, we identify four archetypes using cluster analysis. The taxonomy provides practitioners with a differentiated view on the configuration options of CS business models including the involvement of governmental institutions, while the presented archetypes contribute an aggregated view of CS business models. The statistical analysis of our results provides initial explanatory approaches on CS business model dimension interdependencies, thereby laying the basis for a deeper understanding of sectoral and geographic differences of B2B-MSP and their diffusion dynamics as well as facilitating a higher contextualization of future research.

Identifying clinical subtypes in sepsis-survivors with different one-year outcomes: a secondary latent class analysis of the FROG-ICU cohort

BACKGROUND

Late mortality risk in sepsis-survivors persists for years with high readmission rates and low quality of life. The present study seeks to link the clinical sepsis-survivors heterogeneity with distinct biological profiles at ICU discharge and late adverse events using an unsupervised analysis.

METHODS

In the original FROG-ICU prospective, observational, multicenter study, intensive care unit (ICU) patients with sepsis on admission (Sepsis-3) were identified (N = 655). Among them, 467 were discharged alive from the ICU and included in the current study. Latent class analysis was applied to identify distinct sepsis-survivors clinical classes using readily available data at ICU discharge. The primary endpoint was one-year mortality after ICU discharge.

RESULTS

At ICU discharge, two distinct subtypes were identified (A and B) using 15 readily available clinical and biological variables. Patients assigned to subtype B (48% of the studied population) had more impaired cardiovascular and kidney functions, hematological disorders and inflammation at ICU discharge than subtype A. Sepsis-survivors in subtype B had significantly higher one-year mortality compared to subtype A (respectively, 34% vs 16%, p < 0.001). When adjusted for standard long-term risk factors (e.g., age, comorbidities, severity of illness, renal function and duration of ICU stay), subtype B was independently associated with increased one-year mortality (adjusted hazard ratio (HR) = 1.74 (95% CI 1.16-2.60); p = 0.006).

CONCLUSIONS

A subtype with sustained organ failure and inflammation at ICU discharge can be identified from routine clinical and laboratory data and is independently associated with poor long-term outcome in sepsis-survivors. Trial registration NCT01367093; https://clinicaltrials.gov/ct2/show/NCT01367093 .

Modeling Wheezing Spells Identifies Phenotypes with Different Outcomes and Genetic Associates

Rationale: Longitudinal modeling of current wheezing identified similar phenotypes, but their characteristics often differ between studies. Objectives: We propose that a more comprehensive description of wheeze may better describe trajectories than binary information on the presence/absence of wheezing. Methods: We derived six multidimensional variables of wheezing spells from birth to adolescence (including duration, temporal sequencing, and the extent of persistence/recurrence). We applied partition-around-medoids clustering on these variables to derive phenotypes in five birth cohorts. We investigated within- and between-phenotype differences compared with binary latent class analysis models and ascertained associations of these phenotypes with asthma and lung function and with polymorphisms in asthma loci 17q12-21 and CDHR3 (cadherin-related family member 3). Measurements and Main Results: Analysis among 7,719 participants with complete data identified five spell-based wheeze phenotypes with a high degree of certainty: never (54.1%), early-transient (ETW) (23.7%), late-onset (LOW) (6.9%), persistent (PEW) (8.3%), and a novel phenotype, intermittent wheeze (INT) (6.9%). FEV₁/FVC was lower in PEW and INT compared with ETW and LOW and declined from age 8 years to adulthood in INT. 17q12-21 and CDHR3 polymorphisms were associated with higher odds of PEW and INT, but not ETW or LOW. Latent class analysis- and spell-based phenotypes appeared similar, but within-phenotype individual trajectories and phenotype allocation differed substantially. The spell-based approach was much more robust in dealing with missing data, and the derived clusters were more stable and internally homogeneous. Conclusions: Modeling of spell variables identified a novel intermittent wheeze phenotype associated with lung function decline to early adulthood. Using multidimensional spell variables may better capture wheeze development and provide a more robust input for phenotype derivation.

Social Vulnerability to Irrigation Water Loss: Assessing the Effects of Water Policy Change on Farmers in Idaho, USA

In the Eastern Snake Plain of Idaho, increasing rates of groundwater extraction for irrigation have corresponded with the adoption of more efficient irrigation technologies; higher use and lower incidental recharge have led to increasing groundwater scarcity. This paper assesses farmer vulnerability to a water resource policy that responds to that scarcity by reducing availability of groundwater for irrigation by 4-20%. Using results from a household survey of impacted farmers, we examine vulnerability in two stages, contributing to theorization of farmer vulnerability in a changing climate as well as producing important regional policy insights. The first stage, multimodel selection and inference, analyzes the primary predictors of two forms of vulnerability to groundwater scarcity among this population of farmers. The second stage, a segmentation analysis, highlights policy-relevant variation in adaptive capacity and in vulnerability predictors across the population. Individual-level results indicate that key indicators of vulnerability include several dimensions of adaptive capacity and sensitivity. At the population level, we find that reductions in sensitivity may play an important role in reducing farmer vulnerability. Accelerating global environmental change will require agriculture in arid and semi-arid regions to adapt to shifts in water availability. As water resources shift, institutional contexts and policy landscapes will shift in parallel, as seen with the reduction in groundwater availability in our case study. These institutional shifts may change the face of adaptation and farmer vulnerability in unexpected ways. Our results indicate that such institutional shifts could lean on efforts to enhance farmer adaptive capacity or reduce farmer sensitivity as mechanisms for reducing farmer vulnerability to adaptation policy changes.

Complex post-traumatic stress disorder and post-migration living difficulties in traumatised refugees and asylum seekers: the role of language acquisition and barriers

Background

Numerous traumatic experiences and post-migration living difficulties (PMLDs) increase the risk of developing symptoms of complex post-traumatic stress disorder (CPTSD) among Afghan refugees and asylum seekers, living in Austria. Research has repeatedly associated higher levels of CPTSD with higher levels of PMLDs. Summarizing PMLDs into empirically derived factors might facilitate a further understanding of their interaction with symptom presentation within distinct clusters of CPTSD.

Objective

The current study aimed to investigate homogeneous subgroups of ICD-11 CPTSD and their association with demographic variables, traumatic experiences, and empirically derived factors of PMLDs.

Method

Within a randomized controlled trail (RCT) CPTSD, PMLDs, and traumatic experiences were assessed in a sample of 93 treatment-seeking Afghan refugees and asylum seekers through a fully structured face-to-face and interpreter-assisted interview using the ITQ, the PMLDC, and a trauma checklist. Underlying clusters of CPTSD, superior factors of PMLDs, and their associations were investigated.

Results

In total, 19.4% of the sample met the diagnostic criteria for PTSD and 49.5% for CPTSD. We identified a 2-cluster solution consisting of two distinct subgroups as best fitting: (1) a CPTSD cluster and (2) a PTSD cluster. The multitude of PMLDs was summarized into four superior factors. CPTSD cluster membership was associated with childhood potentially traumatic experience types, and one of four PMLD factors, namely 'language acquisition & barriers'.

Conclusions

The results suggest that not PMLDs in general, but rather specific types of PMLDs, are associated with CPTSD. An assumed bidirectional relationship between these PMLD factors and CPTSD symptoms might lead to a downward spiral of increasing distress, and could be considered in treatment strategies.

Recurrent Severe Preschool Wheeze: From Pre-Specified Diagnostic Labels to Underlying Endotypes

Rationale: Preschool wheezing is heterogeneous, but the underlying mechanisms are poorly understood.

Objectives: To investigate lower airway inflammation and infection in preschool children with different clinical diagnoses undergoing elective bronchoscopy and BAL.

Methods: We recruited 136 children aged 1–5 years (105 with recurrent severe wheeze [RSW]; 31 with nonwheezing respiratory disease [NWRD]). Children with RSW were assigned as having episodic viral wheeze (EVW) or multiple-trigger wheeze (MTW). We compared lower airway inflammation and infection in different clinical diagnoses and undertook data-driven analyses to determine clusters of pathophysiological features, and we investigated their relationships with prespecified diagnostic labels.

Measurements and Main Results: Blood eosinophil counts and percentages and allergic sensitization were significantly higher in children with RSW than in children with a NWRD. Blood neutrophil counts and percentages, BAL eosinophil and neutrophil percentages, and positive bacterial culture and virus detection rates were similar between groups. However, pathogen distribution differed significantly, with higher detection of rhinovirus in children with RSW and higher detection of Moraxella in sensitized children with RSW. Children with EVW and children with MTW did not differ in terms of blood or BAL-sample inflammation, or bacteria or virus detection. The Partition around Medoids algorithm revealed four clusters of pathophysiological features: 1) atopic (17.9%), 2) nonatopic with a low infection rate and high use of inhaled corticosteroids (31.3%), 3) nonatopic with a high infection rate (23.1%), and 4) nonatopic with a low infection rate and no use of inhaled corticosteroids (27.6%). Cluster allocation differed significantly between the RSW and NWRD groups (RSW was evenly distributed across clusters, and 60% of the NWRD group was assigned to cluster 4; P < 0.001). There was no difference in cluster membership between the EVW and MTW groups. Cluster 1 was dominated by Moraxella detection (P = 0.04), and cluster 3 was dominated by Haemophilus or Staphylococcus or Streptococcus detection (P = 0.02).

Conclusions: We identified four clusters of severe preschool wheeze, which were distinguished by using sensitization, peripheral eosinophilia, lower airway neutrophilia, and bacteriology.

Long-term outcomes of patients who rate symptoms of rheumatoid arthritis as 'satisfactory'

OBJECTIVES

To describe outcomes of patients with early RA in a patient acceptable symptom state (PASS) at treatment initiation and to identify clusters of symptoms associated with poor outcomes.

METHODS

Data came from the Rheumatoid Arthritis Medication Study, a UK multicentre cohort study of RA patients starting MTX. The HAQ, DAS28 and other patient-reported outcome measures (PROMs) were collected at baseline, and at 6 and 12 months. Patients answering yes to the question 'Is your current condition satisfactory, when you take your general functioning and your current pain into consideration?' were defined as PASS; patients answering no were defined as N-PASS. Symptom clusters in the baseline PASS group were identified using K-medians cluster analysis. Outcomes of baseline PASS vs N-PASS patients and each cluster are compared using random effects models.

RESULTS

Of 1127 patients, 572 (50.8%) reported being in PASS at baseline. Over one year, baseline PASS patients had lower DAS28 (mean difference = -0.71, 95% CI -0.83, -0.59) and HAQ scores (mean difference = -0.48, 95% CI -0.56, -0.41) compared with N-PASS patients. Within the baseline PASS group, we identified six symptom clusters. Clusters characterized by high disease activity and high PROMs, or moderate disease activity and high PROMs, had the worst outcomes compared with the other clusters.

CONCLUSION

Despite reporting their condition as 'satisfactory', early RA patients with high PROM scores are less likely to respond to therapy. This group may require increased vigilance to optimize outcomes.

Role of Private Long-Term Care Insurance in Financial Sustainability for an Aging Society

This work analyzes and quantifies the significance of private long-term care insurance for the elderly in protecting families from the increased expenses derived from dependency. We propose an economic and financial model for consumption and income deficit evolution. Survival/dependency are modeled by a Markov process with stochastic simulation techniques to obtain random variable distributions. Based on the Spanish survey of household finances data, Spanish families are classified using a cluster analysis for the wealth decumulation period. The conclusion is that, for a generic family, hiring long-term care insurance causes a significant reduction in the probability of lack of liquidity, the mean first time of lack of liquidity (if it occurs), and the mean present value of overall liquidity needs. It is also observed that there are important differences between these impacts on different groups of families. These results show that hiring long-term care insurance would considerably lower financial problems in the decumulation period.

Comprehensive Cluster Analysis for COPD Including Systemic and Airway Inflammatory Markers

Chronic obstructive pulmonary disease (COPD) is a complex, multidimensional and heterogeneous disease. The main purpose of the present study was to identify clinical phenotypes through cluster analysis in adults suffering from COPD. A retrospective study was conducted on 178 COPD patients in stable state recruited from ambulatory care at University hospital of Liege. All patients were above 40 years, had a smoking history of more than 20 pack years, post bronchodilator FEV1/FVC <70% and denied any history of asthma before 40 years. In this study, the patients were described by a total of 84 mixed sets of variables with some missing values. Hierarchical clustering on principal components (HCPC) was applied on multiple imputation. In the final step, patients were classified into homogeneous distinct groups by consensus clustering. Three different clusters, which shared similar smoking history were found. Cluster 1 included men with moderate airway obstruction (n = 67) while cluster 2 comprised men who were exacerbation-prone, with severe airflow limitation and intense granulocytic airway and neutrophilic systemic inflammation (n = 56). Cluster 3 essentially included women with moderate airway obstruction (n = 55). All clusters had a low rate of bacterial colonization (5%), a low median FeNO value (<20 ppb) and a very low sensitization rate toward common aeroallergens (0-5%). CAT score did not differ between clusters. Including markers of systemic airway inflammation and atopy and applying a comprehensive cluster analysis we provide here evidence for 3 clusters markedly shaped by sex, airway obstruction and neutrophilic inflammation but not by symptoms and T2 biomarkers.

Selflessness, Depression, and Neuroticism: An Interactionist Perspective on the Effects of Self-Transcendence, Perspective-Taking, and Materialism

Dominant theories of depression position self-concept as a central determinant of psychological functioning, but the relationship between the structure of self-concept and depression has not been extensively explored. The present study investigates the relationship between the structure of the self and psychopathological outcomes (depressive symptoms and neuroticism) with two methodological approaches. Using an established framework that draws insight from Buddhist psychology, the structure of the self is conceptualized in terms of selflessness and self-centeredness. Specifically, selflessness is construed as a multidimensional concept characterized by interdependence, outsider phenomenology, and impermanence. The three dimensions of the self were assessed at age 26 with inventories of self-transcendence, perspective-taking, and materialism, respectively (N = 814). First, a variable-centered approach was used to investigate potential interactions between the dimensions of selflessness. Self-transcendence negatively predicted depressive symptoms and neuroticism, whereas perspective-taking and materialism were positively associated with the outcomes. Self-transcendence moderated the relationship between perspective-taking and depressive symptoms. Perspective-taking was not statistically related to depressive symptoms for participants who exhibited higher levels of self-transcendence. The results clarify ambiguous associations between perspective-taking and depression found in previous research. Second, person-centered analyses were used to identify five profiles of self-structure: (1) Selfless, (2) Selfless Materialist, (3) Interdependent Insider, (4) Self-centered Non-materialist, and (5) Self-centered. As hypothesized, the Selfless cluster was associated with low levels of depressive symptoms and neuroticism, whereas the Self-centered cluster was associated with high levels. The profiles demonstrate the manifestation of several combinations of features of the self, which contributes to overall understanding of selflessness by complicating the traditional dichotomy between selflessness and self-centeredness.

Tailoring integrated care services for high-risk patients with multiple chronic conditions: a risk stratification approach using cluster analysis

Background

The purpose of this study was to produce a risk stratification within a population of high-risk patients with multiple chronic conditions who are currently treated under a case management program and to explore the existence of different risk subgroups. Different care strategies were then suggested for healthcare reform according to the characteristics of each subgroup.

Methods

All high-risk multimorbid patients from a case management program in the Navarra region of Spain were included in the study (n = 885). A 1-year mortality risk score was estimated for each patient by logistic regression. The population was then divided into subgroups according to the patients’ estimated risk scores. We used cluster analysis to produce the stratification with Ward’s linkage hierarchical algorithm. The characteristics of the resulting subgroups were analyzed, and post hoc pairwise tests were performed.

Results

Three distinct risk strata were found, containing 45, 38 and 17% of patients. Age increased from cluster to cluster, and functional status, clinical severity, nursing needs and nutritional values deteriorated. Patients in cluster 1 had lower renal deterioration values, and patients in cluster 3 had higher rates of pressure skin ulcers, higher rates of cerebrovascular disease and dementia, and lower prevalence rates of chronic obstructive pulmonary disease.

Conclusions

This study demonstrates the existence of distinct subgroups within a population of high-risk patients with multiple chronic conditions. Current case management integrated care programs use a uniform treatment strategy for patients who have diverse needs. Alternative treatment strategies should be considered to fit the needs of each patient subgroup.

COPD phenotypes and machine learning cluster analysis: A systematic review and future research agenda

Chronic Obstructive Pulmonary Disease (COPD) is a highly heterogeneous condition projected to become the third leading cause of death worldwide by 2030. To better characterize this condition, clinicians have classified patients sharing certain symptomatic characteristics, such as symptom intensity and history of exacerbations, into distinct phenotypes. In recent years, the growing use of machine learning algorithms, and cluster analysis in particular, has promised to advance this classification through the integration of additional patient characteristics, including comorbidities, biomarkers, and genomic information. This combination would allow researchers to more reliably identify new COPD phenotypes, as well as better characterize existing ones, with the aim of improving diagnosis and developing novel treatments. Here, we systematically review the last decade of research progress, which uses cluster analysis to identify COPD phenotypes. Collectively, we provide a systematized account of the extant evidence, describe the strengths and weaknesses of the main methods used, identify gaps in the literature, and suggest recommendations for future research.

Applying the exposome concept in birth cohort research: a review of statistical approaches

The exposome represents the totality of life course environmental exposures (including lifestyle and other non-genetic factors), from the prenatal period onwards. This holistic concept of exposure provides a new framework to advance the understanding of complex and multifactorial diseases. Prospective pregnancy and birth cohort studies provide a unique opportunity for exposome research as they are able to capture, from prenatal life onwards, both the external (including lifestyle, chemical, social and wider community-level exposures) and the internal (including inflammation, metabolism, epigenetics, and gut microbiota) domains of the exposome. In this paper, we describe the steps required for applying an exposome approach, describe the main strengths and limitations of different statistical approaches and discuss their challenges, with the aim to provide guidance for methodological choices in the analysis of exposome data in birth cohort studies. An exposome approach implies selecting, pre-processing, describing and analyzing a large set of exposures. Several statistical methods are currently available to assess exposome-health associations, which differ in terms of research question that can be answered, of balance between sensitivity and false discovery proportion, and between computational complexity and simplicity (parsimony). Assessing the association between many exposures and health still raises many exposure assessment issues and statistical challenges. The exposome favors a holistic approach of environmental influences on health, which is likely to allow a more complete understanding of disease etiology.

Association of Machine Learning-Derived Phenogroupings of Echocardiographic Variables with Heart Failure in Stable Coronary Artery Disease: The Heart and Soul Study

BACKGROUND

Many individual echocardiographic variables have been associated with heart failure (HF) in patients with stable coronary artery disease (CAD), but their combined utility for prediction has not been well studied.

METHODS

Unsupervised model-based cluster analysis was performed by researchers blinded to the study outcome in 1,000 patients with stable CAD on 15 transthoracic echocardiographic variables. We evaluated associations of cluster membership with HF hospitalization using Cox proportional hazards regression analysis.

RESULTS

The echo-derived clusters partitioned subjects into four phenogroupings: phenogroup 1 (n = 85) had the highest levels, phenogroups 2 (n = 314) and 3 (n = 205) displayed intermediate levels, and phenogroup 4 (n = 396) had the lowest levels of cardiopulmonary structural and functional abnormalities. Over 7.1 ± 3.2 years of follow-up, there were 198 HF hospitalizations. After multivariable adjustment for traditional cardiovascular risk factors, phenogroup 1 was associated with a nearly fivefold increased risk (hazard ratio [HR] = 4.8; 95% CI, 2.4-9.5), phenogroup 2 was associated with a nearly threefold increased risk (HR = 2.7; 95% CI, 1.4-5.0), and phenogroup 3 was associated with a nearly twofold increased risk (HR = 1.9; 95% CI, 1.0-3.8) of HF hospitalization, relative to phenogroup 4.

CONCLUSIONS

Transthoracic echocardiographic variables can be used to classify stable CAD patients into separate phenogroupings that differentiate cardiopulmonary structural and functional abnormalities and can predict HF hospitalization, independent of traditional cardiovascular risk factors.

Modifying effect of metabotype on diet-diabetes associations

Purpose

Inter-individual metabolic differences may be a reason for previously inconsistent results in diet–diabetes associations. We aimed to investigate associations between dietary intake and diabetes for metabolically homogeneous subgroups (‘metabotypes’) in a large cross-sectional study.

Methods

We used data of 1517 adults aged 38–87 years from the German population-based KORA FF4 study (2013/2014). Dietary intake was estimated based on the combination of a food frequency questionnaire and multiple 24-h food lists. Glucose tolerance status was classified based on an oral glucose tolerance test in participants without a previous diabetes diagnosis using American Diabetes Association criteria. Logistic regression was applied to examine the associations between dietary intake and diabetes for two distinct metabotypes, which were identified based on 16 biochemical and anthropometric parameters.

Results

A low intake of fruits and a high intake of total meat, processed meat and sugar-sweetened beverages (SSB) were significantly associated with diabetes in the total study population. Stratified by metabotype, associations with diabetes remained significant for intake of total meat (OR 1.67, 95% CI 1.04–2.67) and processed meat (OR 2.23, 95% CI 1.24–4.04) in the metabotypes with rather favorable metabolic characteristics, and for intake of fruits (OR 0.83, 95% CI 0.68–0.99) and SSB (OR:1.21, 95% CI 1.09–1.35) in the more unfavorable metabotype. However, only the association between SSB intake and diabetes differed significantly by metabotype (p value for interaction = 0.01).

Conclusions

Our findings suggest an influence of metabolic characteristics on diet–diabetes associations, which may help to explain inconsistent previous results. The causality of the observed associations needs to be confirmed in prospective and intervention studies.

Electronic supplementary material

The online version of this article (10.1007/s00394-019-01988-5) contains supplementary material, which is available to authorized users.

Identification of Comprehensive Metabotypes Associated with Cardiometabolic Diseases in the Population-Based KORA Study

SCOPE

"Metabotyping" describes the grouping of metabolically similar individuals. We aimed to identify valid metabotypes in a large cohort for targeted dietary intervention, for example, for disease prevention.

METHODS AND RESULTS

We grouped 1729 adults aged 32-77 years of the German population-based KORA F4 study (2006-2008) using k-means cluster analysis based on 34 biochemical and anthropometric parameters. We identified three metabolically distinct clusters showing significantly different biochemical parameter concentrations. Cardiometabolic disease status was determined at baseline in the F4 study and at the 7 year follow-up termed FF4 (2013/2014) to compare disease prevalence and incidence between clusters. Cluster 3 showed the most unfavorable marker profile with the highest prevalence of cardiometabolic diseases. Also, disease incidence was higher in cluster 3 compared to clusters 2 and 1, respectively, for hypertension (41.2%/25.3%/18.2%), type 2 diabetes (28.3%/5.1%/2.0%), hyperuricemia/gout (10.8%/2.3%/0.7%), dyslipidemia (19.2%/18.3%/5.6%), all metabolic (54.5%/36.8%/19.7%), and all cardiovascular (6.3%/5.5%/2.3%) diseases together.

CONCLUSION

Cluster analysis based on an extensive set of biochemical and anthropometric parameters allows the identification of comprehensive metabotypes that were distinctly different in cardiometabolic disease occurrence. As a next step, targeted dietary strategies should be developed with the goal of preventing diseases, especially in cluster 3.

Patterns of cleaning product exposures using a novel clustering approach for data with correlated variables

PURPOSE

Clustering methods may be useful in epidemiology to better characterize exposures and account for their multidimensional aspects. In this context, application of clustering models allowing for highly dependent variables is of particular interest. We aimed to characterize patterns of domestic exposure to cleaning products using a novel clustering model allowing for highly dependent variables.

METHODS

To identify domestic cleaning patterns in a large population of French women, we used a mixture model of dependency blocks. This novel approach specifically models within-class dependencies, and is an alternative to the latent class model, which assumes conditional independence. Analyses were conducted in 19,398 participants of the E3N study (women aged 61-88 years) who completed a questionnaire regarding household cleaning habits.

RESULTS

Seven classes were identified, which differed with the frequency of cleaning tasks (e.g., dusting/sweeping/hoovering) and use of specific products (e.g., bleach, sprays). The model also grouped the variables into conditionally independent blocks, providing a summary of the main dependencies among the variables.

CONCLUSIONS

The mixture model of dependency blocks, a useful alternative to the latent class model, may have broader application in epidemiology, in particular, in the context of exposome research and growing need for data-reduction methods.

Intensive prolonged exposure therapy for chronic PTSD patients following multiple trauma and multiple treatment attempts

Background: Suboptimal response and high dropout rates leave room for improvement of trauma-focused treatment (TFT) effectiveness in ameliorating posttraumatic stress disorder (PTSD) symptoms. Objective: To explore the effectiveness and safety of intensive prolonged exposure (iPE) targeting chronic PTSD patients with a likely diagnosis of ICD-11 Complex PTSD following multiple interpersonal trauma and a history of multiple treatment attempts. Method: Participants (N = 73) received iPE in 12 × 90-minute sessions over four days (intensive phase) followed by four weekly 90-minute booster prolonged exposure (PE) sessions (booster phase). The primary outcomes, clinician-rated severity of PTSD symptoms, and diagnostic status (Clinician-Administered PTSD Scale; CAPS-IV) were assessed at baseline, post-treatment, and at three and six months. Treatment response trajectories were identified and predictors of these trajectories explored. Results: Mixed model repeated measures analysis of CAPS-IV scores showed a baseline-to-posttreatment decrease in PTSD symptom severity (p < .001) that persisted during the three- and six-month follow-ups with large effect sizes (Cohen's d > 1.2); 71% of the participants responded. None of the participants dropped out during the intensive phase and only 5% during the booster phase. Adverse events were extremely low and only a minority showed symptom exacerbation. Cluster analysis demonstrated four treatment response trajectories: Fast responders (13%), Slow responders (26%), Partial responders (32%), and Non-responders (29%). Living condition and between-session fear habituation were found to predict outcome. Participants living alone were more likely to belong to the Partial responders than to the Non-responders cluster, and participants showing more between-session fear habituation were more likely to belong to the Fast responders than to the Non-responders cluster. Conclusions: The results of this open study suggest that iPE can be effective in PTSD patients with multiple interpersonal trauma and after multiple previous treatment attempts. In addition, in this chronic PTSD population iPE was safe.

Linear growth trajectories in Zimbabwean infants

BACKGROUND

Undernutrition in early life underlies 45% of child deaths globally. Stunting malnutrition (suboptimal linear growth) also has long-term negative effects on childhood development. Linear growth deficits accrue in the first 1000 d of life. Understanding the patterns and timing of linear growth faltering or recovery during this period is critical to inform interventions to improve infant nutritional status.

OBJECTIVE

We aimed to identify the pattern and determinants of linear growth trajectories from birth through 24 mo of age in a cohort of Zimbabwean infants.

DESIGN

We performed a secondary analysis of longitudinal data from a subset of 3338 HIV-unexposed infants in the Zimbabwe Vitamin A for Mothers and Babies trial. We used k-means clustering for longitudinal data to identify linear growth trajectories and multinomial logistic regression to identify covariates that were associated with each trajectory group.

RESULTS

For the entire population, the mean length-for-age z score declined from -0.6 to -1.4 between birth and 24 mo of age. Within the population, 4 growth patterns were identified that were each characterized by worsening linear growth restriction but varied in the timing and severity of growth declines. In our multivariable model, 1-U increments in maternal height and education and infant birth weight and length were associated with greater relative odds of membership in the least-growth restricted groups (A and B) and reduced odds of membership in the more-growth restricted groups (C and D). Male infant sex was associated with reduced odds of membership in groups A and B but with increased odds of membership in groups C and D.

CONCLUSION

In this population, all children were experiencing growth restriction but differences in magnitude were influenced by maternal height and education and infant sex, birth weight, and birth length, which suggest that key determinants of linear growth may already be established by the time of birth. This trial was registered at clinicaltrials.gov as NCT00198718.

Fraction of Missing Information (γ) at Different Missing Data Fractions in the 2012 NAMCS Physician Workflow Mail Survey

In his 1987 classic book on multiple imputation (MI), Rubin used the fraction of missing information, γ, to define the relative efficiency (RE) of MI as RE = (1 + γ/m)^-1/2, where m is the number of imputations, leading to the conclusion that a small m (≤5) would be sufficient for MI. However, evidence has been accumulating that many more imputations are needed. Why would the apparently sufficient m deduced from the RE be actually too small? The answer may lie with γ. In this research, γ was determined at the fractions of missing data (δ) of 4%, 10%, 20%, and 29% using the 2012 Physician Workflow Mail Survey of the National Ambulatory Medical Care Survey (NAMCS). The γ values were strikingly small, ranging in the order of 10^-6 to 0.01. As δ increased, γ usually increased but sometimes decreased. How the data were analysed had the dominating effects on γ, overshadowing the effect of δ. The results suggest that it is impossible to predict γ using δ and that it may not be appropriate to use the γ-based RE to determine sufficient m.

Exposure to Endocrine-Disrupting Chemicals during Pregnancy and Weight at 7 Years of Age: A Multi-pollutant Approach

BACKGROUND

Prenatal exposure to endocrine-disrupting chemicals (EDCs) may induce weight gain and obesity in children, but the obesogenic effects of mixtures have not been studied.

OBJECTIVE

We evaluated the associations between pre- and perinatal biomarker concentrations of 27 EDCs and child weight status at 7 years of age.

METHODS

In pregnant women enrolled in a Spanish birth cohort study between 2004 and 2006, we measured the concentrations of 10 phthalate metabolites, bisphenol A, cadmium, arsenic, and lead in two maternal pregnancy urine samples; 6 organochlorine compounds in maternal pregnancy serum; mercury in cord blood; and 6 polybrominated diphenyl ether congeners in colostrum. Among 470 children at 7 years, body mass index (BMI) z-scores were calculated, and overweight was defined as BMI > 85th percentile. We estimated associations with EDCs in single-pollutant models and applied principal-component analysis (PCA) on the 27 pollutant concentrations.

RESULTS

In single-pollutant models, HCB (hexachlorobenzene), βHCH (β-hexachlorocyclohexane), and polychlorinated biphenyl (PCB) congeners 138 and 180 were associated with increased child BMI z-scores; and HCB, βHCH, PCB-138, and DDE (dichlorodiphenyldichloroethylene) with overweight risk. PCA generated four factors that accounted for 43.4% of the total variance. The organochlorine factor was positively associated with BMI z-scores and with overweight (adjusted RR, tertile 3 vs. 1: 2.59; 95% CI: 1.19, 5.63), and these associations were robust to adjustment for other EDCs. Exposure in the second tertile of the phthalate factor was inversely associated with overweight.

CONCLUSIONS

Prenatal exposure to organochlorines was positively associated with overweight at age 7 years in our study population. Other EDCs exposures did not confound this association.

Construction of quality of life change patterns: example in oncology in a phase III therapeutic trial (FFCD 0307)

Objective

Quality of life data in cancerology are often difficult to summarize due to missing data and difficulty to analyze the pattern of evolution in different groups of patients. The aim of this work was to apply a new methodology to construct Quality of Life (QoL) change patterns within patients included in a clinical trial comparing to regimen of treatment in locally advanced eosogastric cancer.

Materials and methods

In this trial, QoL was assessed every 2 months by self-reported EORTC QLQ-C30 questionnaire. Physical dimension scores were analyzed. After multiple imputation of missing data, 27 statistical measures aiming to describe the variation of QoL measures among follow-up were computed for each patient. Based on these measures, patient were grouped into homogenous groups in terms of QoL variation pattern using a K-Means classification method. The mean QoL score at each time was graphically represented in each obtained pattern. Finally, clinical characteristic of patients in each pattern of QoL were described and compared.

Results

The trial included 416 patients and 1023 questionnaire were collected. 74 % of patients were male with a mean ± SD age of 62 ± 11 years. 43 % of scores were missing. Patients were grouped into four classes of homogeneous QoL variation patterns. 1) a Pattern of 24 (6 %) patients showing improvement in QoL with a mean variation of +10.7 points on the 0–100 scale, 2) a Pattern of 171 (41 %) patients showing a stability 3) two Patterns of 78 (19 %) and 143 (34 %) patients respectively showing a deterioration of QoL with a mean variation of −67.2 and −67.6, respectively. There were no difference between patterns in terms of gender or age. Patients within “degradation” pattern had significantly lower performance status (p = 0.015), higher severe after-effects rate (p < 10-3) and death rate (p < 10-3).

Conclusion

This work opens up perspectives for longitudinal data analysis with a high probability of missing values while providing a relevant graphical summary. Patterns of QoL evolution with clinical relevance may help to interpret longitudinal QoL data in Cancer studies.

Electronic supplementary material

The online version of this article (doi:10.1186/s12955-015-0342-1) contains supplementary material, which is available to authorized users.

Quality of Life Assessment in Sarcoidosis

Quantifying health-related quality of life (HRQL) and specific patient symptoms has developed to include rigorous techniques to develop patient-reported outcome measures (PROs). PROs may assess objectively the impact of a therapeutic intervention in a clinical sarcoidosis trial, and may be useful in following HRQL. Item response theory may lead to the construction of PROs that allow for the development of short forms, the PRO to be focused on specific areas along the continuum of the trait being studied, and the development of computer-adaptive testing where HRQL can be assessed accurately using very few items.

Phenotyping asthma, rhinitis and eczema in MeDALL population-based birth cohorts: an allergic comorbidity cluster

BACKGROUND

Asthma, rhinitis and eczema often co-occur in children, but their interrelationships at the population level have been poorly addressed. We assessed co-occurrence of childhood asthma, rhinitis and eczema using unsupervised statistical techniques.

METHODS

We included 17 209 children at 4 years and 14 585 at 8 years from seven European population-based birth cohorts (MeDALL project). At each age period, children were grouped, using partitioning cluster analysis, according to the distribution of 23 variables covering symptoms 'ever' and 'in the last 12 months', doctor diagnosis, age of onset and treatments of asthma, rhinitis and eczema; immunoglobulin E sensitization; weight; and height. We tested the sensitivity of our estimates to subject and variable selections, and to different statistical approaches, including latent class analysis and self-organizing maps.

RESULTS

Two groups were identified as the optimal way to cluster the data at both age periods and in all sensitivity analyses. The first (reference) group at 4 and 8 years (including 70% and 79% of children, respectively) was characterized by a low prevalence of symptoms and sensitization, whereas the second (symptomatic) group exhibited more frequent symptoms and sensitization. Ninety-nine percentage of children with comorbidities (co-occurrence of asthma, rhinitis and/or eczema) were included in the symptomatic group at both ages. The children's characteristics in both groups were consistent in all sensitivity analyses.

CONCLUSION

At 4 and 8 years, at the population level, asthma, rhinitis and eczema can be classified together as an allergic comorbidity cluster. Future research including time-repeated assessments and biological data will help understanding the interrelationships between these diseases.

Descriptive data analysis examining how standardized assessments are used to guide post-acute discharge recommendations for rehabilitation services after stroke

BACKGROUND

Use of standardized assessments in acute rehabilitation is continuing to grow, a key objective being to assist clinicians in determining services needed postdischarge.

OBJECTIVE

The purpose of this study was to examine how standardized assessment scores from initial acute care physical therapist and occupational therapist evaluations contribute to discharge recommendations for poststroke rehabilitation services.

DESIGN

A descriptive analysis was conducted.

METHODS

A total of 2,738 records of patients admitted to an acute care hospital with a diagnosis of stroke or transient ischemic attack were identified. Participants received an initial physical therapist and occupational therapist evaluation with standardized assessments and a discharge recommendation of home with no services, home with services, inpatient rehabilitation facility (IRF), or skilled nursing facility (SNF). A K-means clustering algorithm determined if it was feasible to categorize participants into the 4 groups based on their assessment scores. These results were compared with the physical therapist and occupational therapist discharge recommendations to determine if assessment scores guided postacute care recommendations.

RESULTS

Participants could be separated into 4 clusters (A, B, C, and D) based on assessment scores. Cluster A was the least impaired, followed by clusters B, C, and D. In cluster A, 50% of the participants were recommended for discharge to home without services, whereas 1% were recommended for discharge to an SNF. Clusters B, C, and D each had a large proportion of individuals recommended for discharge to an IRF (74%-80%). There was a difference in percentage of recommendations across the clusters that was largely driven by the differences between cluster A and clusters B, C, and D.

LIMITATIONS

Additional unknown factors may have influenced the discharge recommendations.

CONCLUSIONS

Participants poststroke can be classified into meaningful groups based on assessment scores from their initial physical therapist and occupational therapist evaluations. These assessment scores, in part, guide poststroke acute care discharge recommendations.

An empirical typology of private child and family serving agencies

Differences in how services are organized and delivered can contribute significantly to variation in outcomes experienced by children and families. However, few comparative studies identify the strengths and limitations of alternative delivery system configurations. The current study provides the first empirical typology of private agencies involved with the formal child welfare system. Data collected in 2011 from a national sample of private agencies were used to classify agencies into five distinct groups based on internal management capacity, service diversification, integration, and policy advocacy. Findings reveal considerable heterogeneity in the population of private child and family serving agencies. Cross-group comparisons suggest that differences in agencies' strategic and structural characteristics correlated with agency directors' perceptions of different pressures in their external environment. Future research can use this typology to better understand local service systems and the extent to which different agency strategies affect performance and other outcomes. Such information has implications for public agency contracting decisions and could inform system-level assessment and planning of services for children and families.

Identification of clinical phenotypes using cluster analyses in COPD patients with multiple comorbidities

Chronic obstructive pulmonary disease (COPD) is characterized by persistent airflow limitation, the severity of which is assessed using forced expiratory volume in 1 sec (FEV1, % predicted). Cohort studies have confirmed that COPD patients with similar levels of airflow limitation showed marked heterogeneity in clinical manifestations and outcomes. Chronic coexisting diseases, also called comorbidities, are highly prevalent in COPD patients and likely contribute to this heterogeneity. In recent years, investigators have used innovative statistical methods (e.g., cluster analyses) to examine the hypothesis that subgroups of COPD patients sharing clinically relevant characteristics (phenotypes) can be identified. The objectives of the present paper are to review recent studies that have used cluster analyses for defining phenotypes in observational cohorts of COPD patients. Strengths and weaknesses of these statistical approaches are briefly described. Description of the phenotypes that were reasonably reproducible across studies and received prospective validation in at least one study is provided, with a special focus on differences in age and comorbidities (including cardiovascular diseases). Finally, gaps in current knowledge are described, leading to proposals for future studies.