PSICA: Decision trees for probabilistic subgroup identification with categorical treatments

Personalized medicine aims at identifying best treatments for a patient with given characteristics. It has been shown in the literature that these methods can lead to great improvements in medicine compared to traditional methods prescribing the same treatment to all patients. Subgroup identification is a branch of personalized medicine, which aims at finding subgroups of the patients with similar characteristics for which some of the investigated treatments have a better effect than the other treatments. A number of approaches based on decision trees have been proposed to identify such subgroups, but most of them focus on two-arm trials (control/treatment) while a few methods consider quantitative treatments (defined by the dose). However, no subgroup identification method exists that can predict the best treatments in a scenario with a categorical set of treatments. We propose a novel method for subgroup identification in categorical treatment scenarios. This method outputs a decision tree showing the probabilities of a given treatment being the best for a given group of patients as well as labels showing the possible best treatments. The method is implemented in an R package psica available on CRAN. In addition to a simulation study, we present an analysis of a community-based nutrition intervention trial that justifies the validity of our method.     

Identification of airborne pollen allergens from two avenue trees of India

An attempt has been made to detect airborne pollen of Lagerstroemia speciosa (LS) and Spathodea campanulata (SC) – two common avenue trees of India as potential sources of aeroallergens and also to identify the major IgE-reactive components present in them. The airborne pollen concentration was assessed using a Burkard sampler. A detailed questionnaire on clinical data of 1490 patients was recorded based on hospital data. We assessed the allergenicity of pollen by in vivo and in vitro tests. The correlation among meteorological factors, pollen seasons and allergenic potency of patients was assessed by multiple regression analysis. The sensitivity of patients to pollen antigens was highly correlated with pollen seasons. In SDS–PAGE, 15 protein bands were detected from LS pollen, while 14 bands from SC. The IgE-specific immunoblotting with patients’ sera allergic to LS displayed five major allergens, while four major allergens were detected from SC. This would be the first report from India to prove the allergenic potentiality of airborne pollen of these two common avenue trees of India.     

Ageing in India: Financial hardship from health expenditures

India's rapidly ageing population raises concerns about the burden of health care payments among older individuals who may have both limited income and greater health care needs. Using a nationally representative household survey, we investigate the association between age and financial hardship due to health expenditures. We find that both the probability of experiencing health problems and mean total out‐of‐pocket health expenditures increase with age. Second, the probability of households experiencing catastrophic health expenditures increases with each additional member aged 60 and above—33% of households with one 60+ member and 38% of households with 2 or more 60+ members experienced catastrophic health expenditures, compared to only 20% in households with all members under the age of 60 years. Lastly, we show that individuals aged 60 and above had a much higher probability of becoming impoverished as a result of health expenditures—the probability of impoverishment for 60+ individuals was 3 percentage points higher than for individuals under the age of 60. Overall, around 4.8% of the older population, representing 4.1 million people, fell into poverty. The results suggest that there is an urgent need for public investments in financial protection programs for older people in India.     

Evaluation of local treatment effect by borrowing information from similar countries in multi‐regional clinical trials

One key objective of a multi‐regional clinical trial (MRCT) is to quantify country‐specific treatment effects to support local registration. Naïve estimate of the treatment effects based on country‐specific subsets of data are inefficient because of relatively sparse country‐specific sample sizes, even in large MRCTs, and subject to many limitations. The recently developed shrinkage estimate approach improves efficiency by incorporating data from other countries but treating all other countries equally, thus ignoring similarity to the country of interest. Ideally, for the estimation of treatment effect in a country of interest, it is more efficient to use all relevant data such as that from countries that are similar to the one of interest. We propose a tree‐based approach to incorporate similarity, with branches or forks used to represent the relationship of treatment effects between countries. Similarity in certain characteristics among countries in either continuous scale or categorical scale is used to build the tree. By allowing local treatment effects for countries to be random effects and follow a Gaussian process along the tree, we demonstrate that countries sharing the same internal parent nodes are more correlated and a country borrows more information from the neighbors sharing the same parent node. We illustrated this statistical framework using a MRCT with a continuous endpoint. Copyright © 2015 John Wiley & Sons, Ltd.     

A comparative study of variable selection methods in the context of developing psychiatric screening instruments

The development of screening instruments for psychiatric disorders involves item selection from a pool of items in existing questionnaires assessing clinical and behavioral phenotypes. A screening instrument should consist of only a few items and have good accuracy in classifying cases and non‐cases. Variable/item selection methods such as Least Absolute Shrinkage and Selection Operator (LASSO), Elastic Net, Classification and Regression Tree, Random Forest, and the two‐sample t‐test can be used in such context. Unlike situations where variable selection methods are most commonly applied (e.g., ultra high‐dimensional genetic or imaging data), psychiatric data usually have lower dimensions and are characterized by the following factors: correlations and possible interactions among predictors, unobservability of important variables (i.e., true variables not measured by available questionnaires), amount and pattern of missing values in the predictors, and prevalence of cases in the training data. We investigate how these factors affect the performance of several variable selection methods and compare them with respect to selection performance and prediction error rate via simulations. Our results demonstrated that: (1) for complete data, LASSO and Elastic Net outperformed other methods with respect to variable selection and future data prediction, and (2) for certain types of incomplete data, Random Forest induced bias in imputation, leading to incorrect ranking of variable importance. We propose the Imputed‐LASSO combining Random Forest imputation and LASSO; this approach offsets the bias in Random Forest and offers a simple yet efficient item selection approach for missing data. As an illustration, we apply the methods to items from the standard Autism Diagnostic Interview‐Revised version. Copyright © 2013 John Wiley & Sons, Ltd.     

The impact of out‐of‐pocket health expenditure on household impoverishment: Evidence from Morocco

Health financing in Morocco relies mainly on out‐of‐pocket (OoP) payments. World Health Organization (WHO) has shown that these payments can expose households to catastrophic health expenditure (hereinafter CHE) and impoverish them. The study examines the financial burden of OoP health payments on Moroccan households. Two approaches—that developed by Wagstaff and Doeslear and the one advocated by WHO—are adopted to estimate the extent of CHE. These show that 1.77% of households incurred CHE at the 40% threshold for nonfood expenditure. At the 10% threshold for total consumption expenditure, 12.8% of households incurred CHE. We find that these OoP payments have made 1.11% of Moroccan households poorer. In analyzing the determinants of CHE, we estimated an ordered probit model. It appears that any of (a) hospitalization, (b) presence of an elderly person in the household, or (c) the level of poverty increases significantly the likelihood of health expenditure becoming catastrophic. On the other hand, we find that coverage by health insurance protects against CHE.     

Transitions from private to public health coverage among children: estimating effects on out-of-pocket medical costs and health insurance premium costs

Objective. To assess the effects of transitions from private to public health insurance by children on out‐of‐pocket medical expenditures and health insurance premium costs. Data Sources. Data are drawn from the 1996 and 2001 panels of the Survey of Income and Program Participation. We construct a nationally representative, longitudinal sample of children, ages 0–18, and their families for the period 1998–2003, a period in which states raised public health insurance eligibility rates for children. Study Design. We exploit the Survey of Income and Program Participation's longitudinal design to identify children in our sample who transition from private to public health insurance. We then use a bootstrapped instrumental variable approach to estimate the effects of these transitions on out‐of‐pocket expenditures and health insurance premium costs. Principal Findings. Children who transition from private to public coverage are relatively low‐income, are disproportionately likely to live in single‐mother households, and are more likely to be Black or of Hispanic origin. Child health status is highly predictive of transitions. We estimate that these transitions provide a cash‐equivalent transfer of nearly U.S.$1,500 annually for families in the form of reduced out‐of‐pocket and health insurance premium costs. Conclusions. Transitions from private to public health coverage by children can bring important social benefits to vulnerable families. This suggests that instead of being a net societal cost, such transitions may provide an important social benefit.     

Vaccination coverage rates for Diphtheria,Tetanus, Poliomyelitis and Pertussis booster vaccination in France between 2013 and 2017: Learnings from an analysis of National Health System Real-World Data

BackgroundMaintaining a high vaccination coverage rate (VCR) throughout the lifetime and complying with the National Immunization Program are essential to optimize the protection of the population. The study objectives were to evaluate the evolution of the VCRs and the compliance with the vaccination visits for the diphtheria, tetanus, poliomyelitis and pertussis boosters in France since the changes implemented in the 2013 National Immunization Program.MethodsCumulative booster VCRs were estimated at all vaccination visits, from 2013 to 2017, among persons eligible for a booster vaccination from a 1/97th random sample of French claims data. Broader age groups around the recommended ages by the vaccination schedule (6, 11–13, 25, 45, 65, 75, 85, 95y) were used: all persons aged 5 to 8, 10 to 15, 21 to 29, 41 to 49, 61 to 69, 71 to 79, 81 to 89 and 91 to 99.ResultsOver the study period, the diphtheria-tetanus-poliomyelitis booster VCRs increased, reaching in 2017: 73.3% at 8 years old, 75.6% at 15 years old, 46.6% at 29 years old, 38.4% at 49 years old, 36.3% at 69 years old, 30.8% at 79 years old, 22.1% at 89 years old and 11.0% at 99 years old. The pertussis VCRs were also increasing at all vaccination visits, in particular at the vaccination visits at 6 and 11–13 years old (from 16.4% to 63.8% and from 50.3% to 61.2%, respectively). Delayed vaccinations were observed at all vaccination visits.ConclusionVCRs for Diphtheria, Tetanus, Poliomyelitis and Pertussis booster vaccination increased from 2013 to 2017 while remaining suboptimal across all ages and lower in the adult populations. The analysis also shows that the introduction in 2013 of a pertussis vaccination at 6 years of age was relatively well-established in 2017 while other changes in recommendations were slowly or partially implemented.