Objective, study design and setting
The objective of this multicenter, prospective, cross-sectional, and strictly observational study, was to identify those pharmaceutical products less accepted in the older population. The study was conducted between October 2016 and August 2018 in 8 French hospitals and 8 nursing homes.
Inclusion criteria required that patients were 65-years-old and older, hospitalized or residing in a nursing home, receiving any medicine. Patients receiving intravenous medication where the intravenous device is already in situ were excluded, as the insertion of such a device was considered as part of the acceptability. All patients answering these criteria, and who have assented verbally to participating in the study, were included without any randomization. The observer reports were recorded anonymously. Being a non-interventional study, patients were maintained under their current treatment, no modifications or additions to their prescriptions were made during the observational period.
CAST - ClinSearch Acceptability Score Test®, is a tool integrating many aspects of acceptability to discriminate between positively and negatively accepted medicines in vulnerable populations. CAST was initially developed for the pediatric population [17, 18], and has since been transposed for the older population .
This tool includes observer reports corresponding to several observations of behaviors/events performed during the medicine use: patient’s reaction during the administration (positive, neutral or negative reaction), result of the administration (the dose is fully, partly or not taken at all), time needed to prepare (from opening any packaging to having a required dose ready to use, including all handling and modifications), and time to administer the required dose of medication (from a required dose of medication ready to use to the end of the intake). The preparation and administration time was classified as 20 s or shorter, medium time, or longer than one minute. In addition, observers were required to report recourse to any methods used to ease/achieve administration. This might include dividing the intake of a dose which cannot be taken as a whole; altering the intended use (modify the dosage form such as tablet crushed or capsule opened; use another route/mode of administration; use a device not provided); use of food/drink to mask a taste or ease swallowing; use of restraint (the patient had to force himself or was opposed taking the medication). These observational variables were collected prospectively by observers in order to score acceptability.
Each evaluation of one medication taken by one patient corresponding to a particular combination of observed measures was related to information on the product, the patient and the context of use , as described briefly hereafter. Demographic parameters (e.g. sex) and comorbidities (e.g. swallowing disorders) were recorded from the patient’s medical record, while information on the pharmaceutical products (e.g. pharmaceutical form, presence of flavoring agent) were extracted from the summary of product characteristics (SmPC). These explanatory variables were collected in order to investigate their impact on acceptability.
To describe the many aspects of acceptability, previously described observations were included in a multivariate analysis without weighting. Using mapping and clustering processes, an intelligible acceptability reference framework was designed: a three-dimensional acceptability map was created that juxtaposed two clusters of evaluations, each defining a contrasting acceptability profile. All of the evaluations were positioned on the map depending on their similarity: the most similar evaluations converged. Positively connoted observations were over-represented in the first cluster of evaluations defining the “Positively accepted” profile (green area on the map), while negatively connoted evaluations were over-represented in the second cluster defining the “Negatively accepted” profile (red area on the map).
The reference framework was used to investigate the impact of characteristics of both users (gender and presence of swallowing alteration) and drug products (presence of flavoring agents) on acceptability of the oral liquid pharmaceutical products.
To obtain an acceptability score the barycenter of the evaluations of interest (e.g. those related to the flavored formulations) was positioned on the map. A barycenter, along with the entire 90% confidence ellipses surrounding it, belonging to the “positively accepted” profile, could be considered as accepted. Due to differences among patients, we consider a minimum of 30 evaluations to get a robust acceptability score. In cases with fewer than this threshold, we cannot draw any conclusion and only note any tendency that might be observed.
Pearson’s Chi-squared test was used to assess the significance of the differences observed between the subpopulations of patients related to the different acceptability scores, in terms of gender and swallowing disorders .
The R packages “FactoMineR”  and “MissMDA”  were used to perform multivariate analysis and to handle missing data, respectively (R version 3.4.4).