Ramon Oller

Tutorial on methods for interval-censored data and their implementation in R

Interval censoring is encountered in many practical situations when the event of interest cannot be observed and it is only known to have occurred within a time window. The theory for the analysis of interval-censored data has been developed over the past three decades and several reviews have been written. However, it is still a common practice in medical and reliability studies to simplify the interval censoring structure of the data into a more standard right censoring situation by, for instance, imputing the midpoint of the censoring interval. The availability of software for right censoring might well be the main reason for this simplifying practice. In contrast, several methods have been developed to deal with interval-censored data and the corresponding algorithms to make the procedures feasible are scattered across the statistical software or remain behind the personal computers of many researchers. The purpose of this tutorial is to present, in a pedagogical and unified manner, the methodology and the available software for analyzing interval-censored data. The paper covers frequentist non-parametric, parametric and semiparametric estimating approaches, non-parametric tests for comparing survival curves and a section on simulation of interval-censored data. The methods and the software are described using the data from a dental study.

Frequentist and Bayesian approaches for interval-censored data

Interval censoring appears when the event of interest is only known to have occurred within a random time interval. Estimation and hypothesis testing procedures for interval-censored data are surveyed. We distinguish between frequentist and Bayesian approaches. Computational aspects for every proposed method are described and solutions with S-Plus, whenever are feasible, are mentioned. Three real data sets are analyzed.

Identifying patients with advanced chronic conditions for a progressive palliative care approach: a cross-sectional study of prognostic indicators related to end-of-life trajectories

Objectives 2 innovative concepts have lately been developed to radically improve the care of patients with advanced chronic conditions (PACC): early identification of palliative care (PC) needs and the 3 end-of-life trajectories in chronic illnesses (acute, intermittent and gradual dwindling). It is not clear (1) what indicators work best for this early identification and (2) if specific clinical indicators exist for each of these trajectories. The objectives of this study are to explore these 2 issues. Setting 3 primary care services, an acute care hospital, an intermediate care centre and 4 nursing homes in a mixed urban–rural district in Barcelona, Spain. Participants 782 patients (61.5% women) with a positive NECPAL CCOMS-ICO test, indicating they might benefit from a PC approach. Outcome measures The characteristics and distribution of the indicators of the NECPAL CCOMS-ICO tool are analysed with respect to the 3 trajectories and have been arranged by domain (functional, nutritional and cognitive status, emotional problems, geriatric syndromes, social vulnerability and others) and according to their static (severity) and dynamic (progression) properties. Results The common indicators associated with early end-of-life identification are functional (44.3%) and nutritional (30.7%) progression, emotional distress (21.9%) and geriatric syndromes (15.7% delirium, 11.2% falls). The rest of the indicators showed differences in the associations per illness trajectories (p<0.05). 48.2% of the total cohort was identified as advanced frailty patients with no advanced disease criteria. Conclusions Dynamic indicators are present in the 3 trajectories and are especially useful to identify PACC for a progressive PC approach purpose. Most of the other indicators are typically associated with a specific trajectory. These findings can help clinicians improve the identification of patients for a palliative approach.

Utility of the NECPAL CCOMS-ICO© tool and the Surprise Question as screening tools for early palliative care and to predict mortality in patients with advanced chronic conditions: A cohort study

Background: The Surprise Question (SQ) identifies patients with palliative care needs. The NECPAL CCOMS-ICO© (NECPAL) tool combines the Surprise Question with additional clinical parameters for a more comprehensive assessment. The capacity of these screening tools to predict mortality is still unknown. Aim: To explore the predictive validity of the NECPAL and SQ to determine 12- to 24-month mortality. Design: Longitudinal, prospective and observational cohort study. Setting/participants: Three primary care centres, one general hospital, one intermediate care centre, and four nursing homes. Population cohort with advanced chronic conditions and limited life prognosis. Patients were classified according to SQ and NECPAL criteria and followed for 24 months. Results: Data available to assess 1059 of 1064 recruited patients (99.6%) at 12 and 24 months: 837 patients were SQ+ and 780 were NECPAL+. Mortality rates at 24 months were as follows: 44.6% (SQ+) versus 15.8% (SQ−) and 45.8% (NECPAL+) versus 18.3% (NECPAL−) (p = 0.000). SQ+ and NECPAL+ identification was significantly correlated with 24-month mortality risk (hazard ratios: 2.719 and 2.398, respectively). Both tools were highly sensitive (91.4, CI: 88.7–94.1 and 87.5, CI: 84.3–90.7) with high negative predictive values (84.2, CI: 79.4–89.0 and 81.7, CI: 77.2–86.2), with low specificity and positive predictive value. The prognostic accuracy of SQ and NECPAL was 52.9% and 55.2%, respectively. The predictive validity was slightly better for NECPAL. Conclusion: SQ and NECPAL are valuable screening instruments to identify patients with limited life prognosis who may require palliative care. More research is needed to increase its prognostic utility in combination with other parameters.

A quadtree approach based on European geographic grids: reconciling data privacy and accuracy

Methods to preserve confidentiality when publishing geographic information conflict with the need to publish accurate data. The goal of this paper is to create a European geographic grid frame- work to disseminate statistical data over maps. We propose a methodology based on quadtree hierarchical geographic data structures. We create a varying size grid adapted to local area densities. High populated zones are disaggregated in small squares to allow dissemination of accurate data. Alternatively, information on low populated zones is published in big squares to avoid identification of individual data. The methodology has been applied to the 2014 population register data in Catalonia

Discrete choice and survival models in employee turnover analysis

The purpose of this paper is to assess the capacity of two methodological approaches – discrete choice and survival analysis models – to investigate the relationship between socio-economic characteristics and turnover in a retailing company. A comparison of the estimation results under each model and their interpretation is carried out. The study provides a guide to determine, assess and interpret the effects of different driving factors behind turnover.,The authors use a data set containing information about 1,199 workers followed up between January 2007 and December 2009. First, not distinguishing voluntary and involuntary resignation, a binary logistic regression model and a Cox proportional hazards (PH) model for univariate survival data are set up and estimated. Second, distinguishing voluntary and involuntary resignation, a multinomial logistic regression model and a Cox PH model for competing risk data are set up and estimated.,When no distinction is made, the results point that wage and age exert a negative effect on turnover. Risk of resignation is higher for male, single, not married and Spanish nationals. When the distinction is made, previous results hold for voluntary turnover: wage, age, gender, marital status and nationality are significant. However, when explaining involuntary turnover, all variables except wage lose explaining power. The survival analysis approach is better suited as it measures risk of resignation in a longitudinal way. Discrete choice models only study the risk at a particular cut-off point (24 months in case of this study).,This paper is a systematic application, evaluation and comparison of four different statistical models for analysing employee turnover in a single firm. This work is original because no systematic comparison has been done in the context of turnover.

Cada semana cuenta: uso de dispositivos sanitarios y costes relacionados de una cohorte comunitaria de personas con enfermedades crónicas avanzadas

Resumen Objetivo Describir el uso de dispositivos sanitarios de las personas con cronicidad avanzada, cuantificando y caracterizando el coste de dicho consumo para sugerir mejoras en los modelos de atención. Diseño Estudio observacional, analítico y prospectivo durante 3 años de una cohorte de personas con cronicidad avanzada. Emplazamiento Tres equipos de atención primaria (EAP) de Osona, Cataluña. Participantes Un total de 224 personas identificadas como enfermos avanzados mediante una estrategia poblacional sistemática. Mediciones principales Edad, sexo, tipo de domicilio, trayectoria final de vida; uso, tipo y coste de los recursos en atención primaria, urgencias, por equipos de paliativos o de hospitalización (en agudos o atención intermedia). Resultados Se realizaron una media de 1,1 ingresos al año (estancia media = 6 días), el 74% en hospitales de atención intermedia. El 93,4% del tiempo los pacientes vivieron en la comunidad, realizando un contacto semanal con el EAP (45,1% en domicilio). El coste medio diario fue 19,4 euros, siendo los principales capítulos la hospitalización de atención intermedia (36,5%), la actividad EAP (29,4%) y los ingresos en agudos (28,6%). Los determinantes de menor coste serían la trayectoria fragilidad/demencia (p < 0,001), vivir en una residencia (p < 0,001) y el sobreenvejecimiento (p < 0,001). Hay ciertas diferencias en el comportamiento de los EAP en el coste global y en recursos comunitarios (p < 0,05). Conclusiones Los consumos en hospitalización intermedia y atención primaria son más relevantes que las estancias en centros de agudos. Los contextos residencial y domiciliario son importantes para atender con efectividad y eficiencia, especialmente cuando los EAP se preparan para ello.

Kernel-Based Measure of Variable Importance for Genetic Association Studies

Abstract The identification of genetic variants that are associated with disease risk is an important goal of genetic association studies. Standard approaches perform univariate analysis where each genetic variant, usually Single Nucleotide Polymorphisms (SNPs), is tested for association with disease status. Though many genetic variants have been identified and validated so far using this univariate approach, for most complex diseases a large part of their genetic component is still unknown, the so called missing heritability. We propose a Kernel-based measure of variable importance (KVI) that provides the contribution of a SNP, or a group of SNPs, to the joint genetic effect of a set of genetic variants. KVI can be used for ranking genetic markers individually, sets of markers that form blocks of linkage disequilibrium or sets of genetic variants that lie in a gene or a genetic pathway. We prove that, unlike the univariate analysis, KVI captures the relationship with other genetic variants in the analysis, even when measured at the individual level for each genetic variable separately. This is specially relevant and powerful for detecting genetic interactions. We illustrate the results with data from an Alzheimer’s disease study and show through simulations that the rankings based on KVI improve those rankings based on two measures of importance provided by the Random Forest. We also prove with a simulation study that KVI is very powerful for detecting genetic interactions.