Arkaitz Galbete

On the use of randomization tests following adaptive designs

ABSTRACT Randomization tests (sometimes referred to as “re-randomization” tests) are used in clinical trials, either as an assumption-free confirmation of parametric analyses, or as an independent analysis based on the principle of randomization-based inference. In the context of adaptive randomization, either restricted or response-adaptive procedures, it is unclear how accurate such Monte Carlo approximations are, or how many Monte Carlo sequences to generate. In this paper, we describe several randomization procedures for which there is a known exact or asymptotic distribution of the randomization test. For a special class of procedures, called , and binary responses, the exact test statistic has a simple closed form. For the limited subset of existing procedures with known exact and asymptotic distributions, we can use these as a benchmark for the accuracy of Monte Carlo randomization techniques. We conclude that Monte Carlo tests are very accurate, and require minimal computation time. For simple tests with binary response in the class of procedures, the exact distribution provides the best test, but Monte Carlo approximations can be used when the exact distribution is difficult to compute.

Randomization Based Inference for the Drop-The-Loser Rule

In the framework of clinical trials, legal and ethical restrictions make a population model unrealistic for sampling. Randomization tests are a viable alternative to classical inference. Their theoretical properties depend heavily on the random rule used to allocate patients to treatments, so that Ad-Hoc theoretical studies are necessary for each allocation design. In this paper, we obtain theoretical results for randomization tests when the drop-the-loser rule is used.

Quality of life evolution in elderly survivors with localized breast cancer treated with radiotherapy over a three-year follow-up

PURPOSE This paper studies the Quality of Life (QL) of elderly early-stage breast cancer survivors. The aims are to compare the QL scores of these patients after follow-up with their scores before the start of radiotherapy (RT) and compare QL among different axillary treatment groups. METHODS Of 173 patients over 65 who began treatment and completed the EORTC QLQ-C30 and QLQ-BR23 and the Interview for Deterioration in Daily Living Activities in Dementia (IDDD) before RT, 138 also completed these questionnaires three years after RT. Longitudinal changes in QL were assessed for the whole sample using linear mixed-effect models. Also assessed were differences in QL scores between axillary treatment groups (axillary node dissection ALND, sentinel lymph node biopsy SLNB, and no surgery) at the end of the follow-up (Anova or Kruskal-Wallis) and differences in the evolution of QL from baseline among these groups (linear regression models). RESULTS QL scores in the follow-up were high (>70/100 points) in most QL areas, with moderate limitations (>30) in sexual functioning and enjoyment. Five areas (pain, nausea/vomiting, financial impact, breast symptoms and IDDD) improved significantly (<6 points) in the follow-up. The ALND group scored less for the future perspective item (15-20 points) in the follow-up than the other two axillary treatment groups. No differences between the pre-treatment and follow up assessments regarding treatment were found among the axillary treatment groups. CONCLUSIONS Our results suggest that early-stage breast cancer patients adapted well both to their disease and treatments over the follow-up period and to the administration of RT. There were few QL differences between the axillary treatment groups.

Socioeconomic inequalities in cardiometabolic control in patients with type 2 diabetes

BackgroundThe aim of this study was to determine if the achievement of control targets in patients with type 2 diabetes was associated with personal socioeconomic factors and if these associations were sex-dependent.MethodsThis cross-sectional, population-based study was conducted in Spain. Glycated haemoglobin (HbA1c) level and other clinical parameters were obtained from electronic primary care records (n = 32,638 cases). Socioeconomic status was determined using education level and yearly income. Among patients, having their HbA1c level checked during the previous year was considered as an indirect measure of the process of care, whereas tobacco use and clinical parameters such as HbA1c, low-density lipoprotein cholesterol (LDL-c) and blood pressure (BP) were considered intermediate control outcomes. General linear mixed effect models were used to assess associations.ResultsThe achievement of metabolic and cardiovascular control targets in patients with type 2 diabetes was associated with educational level and income, and socioeconomic gradients differed by sex. The probability of having had an HbA1c test performed in the previous year was higher in patients with lower education levels. Patients in the lowest income and education level categories were less likely to have reached the recommended HbA1c level. Males in the lowest education level categories were less likely to be non-smokers or to have achieved the blood pressure targets. In contrast, patients within the low income categories had a higher probability of reaching the recommended LDL-c level.ConclusionsOur results suggest the presence of socioeconomic inequalities in the achievement of cardiovascular and metabolic control that differed in direction and magnitude depending on the measured outcome and sex of the patient. These findings may help health professionals focus on high-risk individuals to decrease health inequalities.

Asymptotic Behavior of a Randomization Test for a Response-Adaptive Design

Abstract Response-adaptive designs allow the incorporation of ethical goals in the performance of a clinical trial, and they have been thoroughly studied assuming that treatment responses follow a population model. However, in some clinical trials, population models are not appropriate and randomization tests appear as a plausible alternative to make inference. Randomization-based tests can be devised but the calculation of their exact p-values when a response-adaptive design is used to allocate patients is either time consuming or not feasible for moderate to large sample sizes and so asymptotic results become helpful. Nevertheless, these asymptotic results are not available for response-adaptive designs with good properties. The Klein allocation rule is a response-adaptive design, with good ethical and inferential properties, that generalizes the classical Ehrenfest urn design by making the replacement policy dependent on the response of the last patient. The goal of this article is to study the asymptotic distribution of a test statistic under a randomization-based approach when patients are allocated by using the Klein allocation rule.