Hospital healthcare flows: A longitudinal clustering approach of acute coronary syndrome in women over 45 years.

Acute coronary syndrome (ACS) in women is a growing public health issue and a death leading cause. We explored whether the hospital healthcare trajectory was characterizable using a longitudinal clustering approach in women with ACS. From the 2009-2014 French nationwide hospital database, we extracted spatio-temporal patterns in ACS patient trajectories, by replacing the spatiality by their hospitalization cause. We used these patterns to characterize hospital healthcare flows in a visualization tool. We clustered these trajectories with kmlShape to identify time gap and tariff profiles. ACS hospital healthcare flows have three key categories: Angina pectoris, Myocardial Infarction or Ischemia. Elderly flows were more complex. Time gap profiles showed that readmissions were closer together as time goes by. Tariff profiles were different according to age and initial event. Our approach might be applied to monitoring other chronic diseases. Further work is needed to integrate these results into a medical decision-making tool.

A Quadriparametric Model to Describe the Diversity of Waves Applied to Hormonal Data.

BACKGROUND: Even in normally cycling women, hormone level shapes may widely vary between cycles and between women. Over decades, finding ways to characterize and compare cycle hormone waves was difficult and most solutions, in particular polynomials or splines, do not correspond to physiologically meaningful parameters. OBJECTIVE: We present an original concept to characterize most hormone waves with only two parameters. METHODS: The modelling attempt considered pregnanediol-3-alpha-glucuronide (PDG) and luteinising hormone (LH) levels in 266 cycles (with ultrasound-identified ovulation day) in 99 normally fertile women aged 18 to 45. The study searched for a convenient wave description process and carried out an extended search for the best fitting density distribution. RESULTS: The highly flexible beta-binomial distribution offered the best fit of most hormone waves and required only two readily available and understandable wave parameters: location and scale. In bell-shaped waves (e.g., PDG curves), early peaks may be fitted with a low location parameter and a low scale parameter; plateau shapes are obtained with higher scale parameters. I-shaped, J-shaped, and U-shaped waves (sometimes the shapes of LH curves) may be fitted with high scale parameter and, respectively, low, high, and medium location parameter. These location and scale parameters will be later correlated with feminine physiological events. CONCLUSION: Our results demonstrate that, with unimodal waves, complex methods (e.g., functional mixed effects models using smoothing splines, second-order growth mixture models, or functional principal-component- based methods) may be avoided. The use, application, and, especially, result interpretation of four-parameter analyses might be advantageous within the context of feminine physiological events.

PaFloChar: An Innovating Approach to Characterise Patient Flows in Myocardial Infarction.

A better knowledge of patient flows would improve decision making in health planning. In this article, we propose a method to characterise patients flows and also to highlight profiles of care pathways considering times and costs. From medico-administrative data, we extracted spatio-temporal patterns. Then, we clustered time between hospitalisations and cost trajectories in order to identify profiles of change over time. This approach may support renewed management strategies.

Characterization of hormonal profiles during the luteal phase in regularly menstruating women.

OBJECTIVE: To characterize the variability of hormonal profiles during the luteal phase in normal cycles. DESIGN: Observational study. SETTING: Not applicable. PATIENT(S): Ninety-nine women contributing 266 menstrual cycles. INTERVENTION(S): The women collected first morning urine samples that were analyzed for estrone-3-glucuronide, pregnanediol-3-alpha-glucuronide (PDG), FSH, and LH. The women had serum P tests (twice per cycle) and underwent ultrasonography to identify the day of ovulation. MAIN OUTCOME MEASURE(S): The luteal phase was divided into three parts: the early luteal phase with increasing PDG (luteinization), the midluteal phase with PDG ≥10 µg/mg Cr (progestation), and the late luteal phase (luteolysis) when PDG fell below 10 µg/mg Cr. RESULT(S): Long luteal phases begin with long luteinization processes. The early luteal phase is marked by low PDG and high LH levels. Long luteinization phases were correlated with low E1G and low PDG levels at day 3. The length of the early luteal phase is highly variable between cycles of the same woman. The duration and hormonal levels during the rest of the luteal phase were less correlated with other characteristics of the cycle. CONCLUSION(S): The study showed the presence of a prolonged pituitary activity during the luteinization process, which seems to be modulated by an interaction between P and LH. This supports a luteal phase model with three distinct processes: the first is a modulated luteinization process, whereas the second and the third are relatively less modulated processes of progestation and luteolysis.

An alternative classification to mixture modeling for longitudinal counts or binary measures.

Classifying patients according to longitudinal measures, or trajectory classification, has become frequent in clinical research. The k-means algorithm is increasingly used for this task in case of continuous variables with standard deviations that do not depend on the mean. One feature of count and binary data modeled by Poisson or logistic regression is that the variance depends on the mean; hence, the within-group variability changes from one group to another depending on the mean trajectory level. Mixture modeling could be used here for classification though its main purpose is to model the data. The results obtained may change according to the main objective. This article presents an extension of the k-means algorithm that takes into account the features of count and binary data by using the deviance as distance metric. This approach is justified by its analogy with the classification likelihood. Two applications are presented with binary and count data to show the differences between the classifications obtained with the usual Euclidean distance versus the deviance distance.

CopyMean: A new method to predict monotone missing values in longitudinal studies.

BACKGROUND: Longitudinal studies are those in which the same variable is repeatedly measured at different times. More likely than others, these studies suffer from missing values. Because the missing values may impact the statistical analyses, it is important that they be dealt with properly. METHODS: In this paper, we present "CopyMean", a new method to impute (predict) monotone missing values. We compared its efficiency to sixteen imputation methods dedicated to the treatment of missing values in longitudinal data. All these methods were tested on four datasets, real or artificial, presenting markedly different caracteristics. RESULTS: The analysis showed that CopyMean was more efficient in almost all situations.

kmlShape: An Efficient Method to Cluster Longitudinal Data (Time-Series) According to Their Shapes.

BACKGROUND: Longitudinal data are data in which each variable is measured repeatedly over time. One possibility for the analysis of such data is to cluster them. The majority of clustering methods group together individual that have close trajectories at given time points. These methods group trajectories that are locally close but not necessarily those that have similar shapes. However, in several circumstances, the progress of a phenomenon may be more important than the moment at which it occurs. One would thus like to achieve a partitioning where each group gathers individuals whose trajectories have similar shapes whatever the time lag between them. METHOD: In this article, we present a longitudinal data partitioning algorithm based on the shapes of the trajectories rather than on classical distances. Because this algorithm is time consuming, we propose as well two data simplification procedures that make it applicable to high dimensional datasets. RESULTS: In an application to Alzheimer disease, this algorithm revealed a "rapid decline" patient group that was not found by the classical methods. In another application to the feminine menstrual cycle, the algorithm showed, contrarily to the current literature, that the luteinizing hormone presents two peaks in an important proportion of women (22%).

A graphical method to assess distribution assumption in group-based trajectory models.

Group-based trajectory models had a rapid development in the analysis of longitudinal data in clinical research. In these models, the assumption of homoscedasticity of the residuals is frequently made but this assumption is not always met. We developed here an easy-to-perform graphical method to assess the assumption of homoscedasticity of the residuals to apply especially in group-based trajectory models. The method is based on drawing an envelope to visualize the local dispersion of the residuals around each typical trajectory. Its efficiency is demonstrated using data on CD4 lymphocyte counts in patients with human immunodeficiency virus put on antiretroviral therapy. Four distinct distributions that take into account increasing parts of the variability of the observed data are presented. Significant differences in group structures and trajectory patterns were found according to the chosen distribution. These differences might have large impacts on the final trajectories and their characteristics; thus on potential medical decisions. With a single glance, the graphical criteria allow the choice of the distribution that best capture data variability and help dealing with a potential heteroscedasticity problem.

Social context of preterm delivery in France in 2011 and impact on short-term health outcomes: the EPIPAGE 2 cohort study.

BACKGROUND: Low socio-economic context increases the risk of preterm delivery and may affect short-term outcomes in children born preterm. We described the social context of preterm delivery in France in 2011 and compared it with the general population of deliveries over the same period. We also studied how social context influenced pregnancy and delivery characteristics in the preterm population, and how it affected mortality and short-term morbidity in liveborn preterm children (<35 weeks). METHODS: We created an individual socio-economic vulnerability index, derived from multiple correspondence analysis based on maternal social information in the French National Perinatal Survey (NPS-2010). Weighted coordinates were applied to families from the EPIPAGE 2 study, a population-based cohort of preterm infants born in 2011, to quantify the infant's exposure to socio-economic vulnerability. Multivariable logistic models were used to relate the socio-economic context to pregnancy and delivery characteristics, and to assess its impact on short-term outcomes of the infants. RESULTS: Among mothers of preterm infants, gestational age decreased as socio-economic conditions worsened. In the most deprived group, women had more irregular pregnancy care, a higher prevalence of infection during pregnancy, and a lower rate of antenatal corticosteroid administration. The most deprived group was associated with a higher risk of severe morbidity for the preterm neonates. CONCLUSION: Our results emphasise the need for a large population-based surveillance system to identify the most deprived mothers, and to propose appropriate follow-up and care to these women and their infants in order to enhance long-term health.

Impact of variability in adherence to HIV antiretroviral therapy on the immunovirological response and mortality.

BACKGROUND: Several previous studies have shown relationships between adherence to HIV antiretroviral therapy (ART) and the viral load, the CD4 cell count, or mortality. However, the impact of variability in adherence to ART on the immunovirological response does not seem to have been investigated yet. METHODS: Monthly adherence data (November 1999 to April 2009) from 317 HIV-1 infected patients enrolled in the Senegalese ART initiative were analyzed. Latent-class trajectory models were used to build typical trajectories for the average adherence and the standardized variance of adherence. The relationship between the standardized variance of adherence and each of the change in CD4 cell count, the change in viral load, and mortality were investigated using, respectively, a mixed linear regression, a mixed logistic regression, and a Cox model with time-dependent covariates. All the models were adjusted on the average adherence. RESULTS: Three latent trajectories for the average adherence and three for the standardized variance of adherence were identified. The increase in CD4 cell count and the increase in the percentage of undetectable viral loads were negatively associated with the standardized variance of adherence but positively associated with the average adherence. The risk of death decreased significantly with the increase in the average adherence but increased significantly with the increase of the standardized variance of adherence. CONCLUSIONS: The impacts of the level and the variability of adherence on the immunovirological response and survival justify the inclusion of these aspects into the process of patient education: adherence should be both high and constant.

Socialization instances linked to cannabis experimentation among French teenagers.

France presents one of the highest prevalence of teenagers aged 15-year-olds who report they already have experienced cannabis in Europe. Data from the French 2010 Health Behavior in School-aged Children (HSBC) survey and environmental parameters typifying schools' neighborhoods were used to study cannabis experimentation. We conducted a two-level logistic regression (clusters being schools) on 4,175 French 8th-10th graders from 156 schools. Several individual parameters were linked to cannabis experimentation. Living in a non-intact family, feeling insufficiently monitored, having poor communication with mother and being from a family with a high socio-economic status (SES) were all associated with increased risk of cannabis experimentation. At environmental level, only being in a priority education area was linked to this behavior, without explaining differences among schools.

The developmental course of childhood inattention symptoms uniquely predicts educational attainment: a 16-year longitudinal study.

In this 16-year longitudinal study, a new trajectory estimation approach was used to verify whether the developmental course of childhood inattention significantly predicted functional impairment. A rising childhood inattention trajectory significantly predicted graduation failure (OR: 1.76 [1.32-2.34]) independently of averaged inattention levels. Rising inattention is, in itself, important for prognosis.

Trends in prevalence and characteristics of post-neonatal cerebral palsy cases: a European registry-based study.

The present paper aims to analyze trends over time in prevalence of cerebral palsy of post-neonatal origin, to investigate whether changes are similar according to severity and to describe the disability profile by etiology. Post-neonatal cases, birth years 1976 to 1998, were identified from the Surveillance of Cerebral Palsy in Europe collaboration (19 population-based registries). A recognized causal event occurring between 28 days and 24 months of age was considered to define the cases. Trends in prevalence were explored using graphical methods (Lowess and Cusum control chart) and modeled with negative binomial regressions. Over the study period, 404 cases were identified as post-neonatal cases (5.5% of the total). Mean prevalence rate was 1.20 per 10,000 live births (95% CI [1.08-1.31]). A significant downward trend was observed (p=0.001), with an accentuated decrease in the 1990 s. The prevalence of severe cases which account for around one third of the total also significantly decreased over time (p<0.001). In 46% of cases, an infectious aetiology was reported; the corresponding prevalence significantly decreased since 1989. No significant decrease was observed for the rate of cases due to a vascular episode or of traumatic origin. Our results emphasize the need of large population-based surveillance systems to reliably monitor trends in prevalence in rare subgroups of children like those with acquired cerebral palsy. The decrease of the overall prevalence as well as those of the most severe cases may be partly due to public health actions targeted to prevent such events.

Developmental trajectories of body mass index in early childhood and their risk factors: an 8-year longitudinal study.

OBJECTIVES: To identify groups of children with distinct developmental trajectories of body mass index (BMI), calculated as weight in kilograms divided by height in meters squared, between the ages of 5 months and 8 years and identify early-life risk factors that distinguish children in an atypically elevated BMI trajectory group. DESIGN: Prospective cohort study. SETTING: Families with a child born between October 1997 and July 1998 in the province of Quebec, Canada. PARTICIPANTS: A representative sample of children (N = 2120) selected through birth registries for the Quebec Longitudinal Study of Child Development. Children for whom BMI data were available for at least 5 time points were retained in the present study (n = 1957). MAIN EXPOSURES: Early-life factors putatively associated with BMI, assessed by maternal report. OUTCOME MEASURE: Group-based trajectories of children's BMI, identified with a semiparametric modeling method from raw BMI values at each age. RESULTS: Three trajectories of BMI were identified: low-stable (54.5% of children), moderate (41.0%), and high-rising (4.5%). The high-rising group was characterized by an increasing average BMI, which exceeded international cutoff values for obesity by age 8 years. Two maternal risk factors were associated with the high-rising group as compared with the low-stable and moderate groups combined: maternal BMI (odds ratio, 2.38; 95% confidence interval, 1.38-4.54 for maternal overweight and 6.33; 3.82-11.85 for maternal obesity) and maternal smoking during pregnancy (2.28; 1.49-4.04). CONCLUSIONS: Children continuing on an elevated BMI trajectory leading to obesity in middle childhood can be distinguished from children on a normative BMI trajectory as early as age 3.5 years. Important and preventable risk factors for childhood obesity are in place before birth.

Childhood trajectories of inattention and hyperactivity and prediction of educational attainment in early adulthood: a 16-year longitudinal population-based study.

OBJECTIVE: Literature clearly documents the association between mental health problems, particularly attention deficit hyperactivity disorder (ADHD), and educational attainment. However, inattention and hyperactivity are generally not considered independently from each other in prospective studies. The aim of the present study was to differentiate the unique, additive, or interactive contributions of inattention and hyperactivity symptoms to educational attainment. METHOD: The authors randomly selected 2,000 participants from a representative sample of Canadian children and estimated developmental trajectories of inattention and hyperactivity between the ages of 6 and 12 years using yearly assessments. High school graduation status, at age 22-23 years, was obtained from official records. RESULTS: Four trajectories of inattention and four trajectories of hyperactivity were observed between the ages of 6 and 12 years. After controlling for hyperactivity and other confounding variables, a high inattention trajectory (compared with low inattention) strongly predicted not having a high school diploma at 22-23 years of age (odds ratio=7.66, 95% confidence interval [CI]=5.06-11.58). To a lesser extent, a declining or rising trajectory of inattention also made a significant contribution (odds ratios of 2.67 [95% CI=1.90-3.75] and 3.87 [95% CI=2.75-5.45], respectively). Hyperactivity was not a significant predictor once inattention was taken into account. CONCLUSIONS: Inattention rather than hyperactivity during elementary school significantly predicts long-term educational attainment. Children with attention problems, regardless of hyperactivity, need preventive intervention early in their development.

KmL: a package to cluster longitudinal data.

Cohort studies are becoming essential tools in epidemiological research. In these studies, measurements are not restricted to single variables but can be seen as trajectories. Thus, an important question concerns the existence of homogeneous patient trajectories. KmL is an R package providing an implementation of k-means designed to work specifically on longitudinal data. It provides several different techniques for dealing with missing values in trajectories (classical ones like linear interpolation or LOCF but also new ones like copyMean). It can run k-means with distances specifically designed for longitudinal data (like Frechet distance or any user-defined distance). Its graphical interface helps the user to choose the appropriate number of clusters when classic criteria are not efficient. It also provides an easy way to export graphical representations of the mean trajectories resulting from the clustering. Finally, it runs the algorithm several times, using various kinds of starting conditions and/or numbers of clusters to be sought, thus sparing the user a lot of manual re-sampling.

Associations between sleep duration patterns and overweight/obesity at age 6.

OBJECTIVE: To investigate whether longitudinal sleep duration patterns during early childhood is a risk factor of overweight or obesity at school entry while controlling for a variety of obesogenic environmental factors. DESIGN, SETTING, AND PARTICIPANTS: This is a prospective cohort study (March-December 1998 to December 2004) of a representative sample of infants born in 1997-1998 in the Canadian province of Quebec. Body mass index (BMI) was measured at ages 2.5 and 6 years. Sleep duration was reported yearly from 2.5 to 6 years of age by their mothers. Prenatal, postnatal (5 and 29 months), and lifestyle (6 y) potentially confounding factors for excess weight were assessed by interviews, questionnaires and hospital records. A group-based semiparametric mixture model was used to estimate developmental patterns of sleep duration. The relationship between sleep duration patterns and BMI was tested using multivariate logistic regression models to control for potentially confounding factors on 1138 children. RESULTS: Four sleep duration patterns were identified: short persistent (5.2%), short increasing (4.7%), 10-hour persistent (50.7%), and 11-hour persistent (39.4%). After controlling for potentially confounding factors, the risk for overweight or obesity was almost 4.2 times higher for short persistent sleepers (odds ratio [OR], 4.2; 95% confidence interval [CI], 1.6 to 11.1; P = 0.003) than for 11-hour persistent sleepers. CONCLUSIONS: Persistently short sleep duration (<10 h) during early childhood significantly increases the risk of excess weight or obesity in childhood, and appears to be independent of other obesogenic factors.