Application of Item Response Theory to Model Disease Progression and Agomelatine Effect in Patients with Major Depressive Disorder

Abstract

Introduction: In this paper, we studied the effect over time of agomelatine, an antidepressant drug administered in patient with major depressive disorder, through item response theory (IRT), taking into account a strong placebo effect and missing not at random data. We also assessed the informativeness of the HAMD-17 scale's item. Materials and Methods: The data includes five phase III clinical trials sponsored by Servier Institute, totalling 1549 patients followed during a maximum of 1 year. At each observation, individual scores for the 17 items of the HAMD scale were recorded. The probability for each score was modelled with IRT. A non-linear mixed effects model was used to describe the evolution of the disease and was coupled with a time to event model to predict dropout. Clinical trial simulations were then used to compare placebo and active treatment. Informativeness of each item was evaluated using the Fisher information theory. Results: The best model combined an IRT model, a longitudinal model for underlying depression which describes the remission and then a possible relapse, and a hazard model for dropout depending on the evolution from baseline. The drug effect was best modelled as an effect on the remission and the relapse phases. The median predicted drop in HAMD between baseline and 6 weeks was 8.8 (90% PI, 8.3-9.2) when on placebo and 13.1 (90% PI, 12.8-13.4) when treated. Nine items were found to be the most informative. Conclusion: The IRT framework allowed to characterise the evolution of depression with time and estimate the effect of agomelatine, as well as the link between symptoms and disease.

Modelling the association between biomarkers and clinical outcome: An introduction to nonlinear joint models

“…Most published studies focus on simple link functions derived from the biomarker value or the random effects (Table 1). However, mechanistic approaches consider link functions based on latent variables 5,13,21 or any relevant combinations of individual parameters, such as the time to tumour growth in oncology. 2 In our simulated data, we assumed an association between risk of death and the current value of tumour size:…”

“…However, they remain limited to baseline factors and do not adequately take into account the effects of a time‐dependent biomarker which reflects the evolution of the patient's condition (see 1 for discussion of Cox models and biomarker). There are many examples of this sort, such as the association between tumour size (TS) dynamics and survival after treatment initiation in oncology, 2,3 or, in psychiatry, the association between the Hamilton Depression Rating Scale score and the time when patients dropout of clinical trials 4,5 . In infectious diseases, CD4 count trajectory has been associated with overall survival 6 and more recently in patients with SARS‐CoV‐2 an increase in viral load has been shown to increase the risk of death 7 .…”

Modeling repeated self-reported outcome data: a continuous-time longitudinal Item Response Theory model

“…Let denote θ the total vector of parameters defined in the structural part of the model described in (1) and in the K measurement equations described in (2). This vector includes:…”

“…Item Response Theory (IRT) models, which exploit the information provided by each individual's items responses, have received growing interest in health science for capturing latent constructs of interest such as depression, anxiety, fatigue, quality of life, or cognitive functioning [1,2,3,4]. IRT models have interesting properties compared to models coming from classical measurement theory such as Classical Test Theory (CTT) models which aggregate the items into a global score or score per domain.…”

Item response theory in early phase clinical trials: Utilization of a reference model to analyze the Montgomery‐Åsberg Depression Rating Scale

“…The benefit of IRT‐based analysis versus total score analysis of clinical trial data has already been demonstrated by studies using both empirical and simulation data to show that estimation of the treatment effect is less biased when applying IRT 6,7 . IRT has also increasingly been applied in the field of nonlinear mixed effects (NLME) modeling, where longitudinal models describing disease progression and treatment effect are now developed based on the change in ψ over time instead of the total score 3,8–14 . Many of these studies have shown that the statistical power to determine treatment effect using an IRT approach is higher when compared to using the total score data in NLME modeling 9,10,13 .…”