Deprescribing: a Portrait and Out-comes of the Reduction of Polypharmacy in Portugal

Study design

This is a three-phase study:

1. Cross-sectional, analytical study of the prevalence and patterns of polypharmacy, namely sociodemographic and clinical profiles (age, genre, area of residence and years of study) and about medication (number of drugs and their active component), in older adults attending Primary Care in Portugal.
2. Cross-sectional, triangulation study of older adults' perception of Barriers to and Facilitators of Deprescribing, Willingness to be Deprescribed and Willingness to Self-medicate.
3. Non-pharmacological randomised clinical study of the impact of enablement of older adults in their willingness to be Deprescribed and related Quality of Life.

Phase I: prevalence of polypharmacy in older adults attending primary care in Portugal Design Cross-sectional, analytical study.

Setting: Primary Care Centres in Portugal will be randomly selected from the five main-land Portuguese Healthcare Administrative Regions and two Autonomous Regions (Madeira and Azores), in order to obtain a national geographical representative sample.

Sample size Since the prevalence of polypharmacy in older adults is unknown, the investigators used as base of population all older adults in Portugal. For the study, the investigators used a 95% confidence interval (CI) and a maximum precision error of 5%, so a minimum of 385 patients should be recruited.

Study procedures This phase of the study starts in November 2017. General Practitioners (GPs) sampling is made according to existing files of previous projects adherent GPs, in other epidemiological studies. After the selection of GPs, those who accept to participate will recruit their own patients. Assuming that a GP will be able to include at least 6 patients in a 3-week period, a total of 65 GPs will be enrolled in the study: 21 in North of Portugal (31.7%), 16 in Centre of Portugal (24.7%), 18 in Lisbon-Tejo Valley (27.4%), 5 in Alentejo (8.4%), 3 in Algarve (4.3%), 1 in Azores (1.6%) and 1 in Madeira (1.9%) in accordance with the distribution of Portuguese old adult population (≥65 years) in Portugal according with Pordata (www.pordata.pt).

Enrolled GPs will be instructed to collect all necessary data about their patients meeting the eligibility criteria.

Data collection The collection of the data will occur in November 2017. GPs will be responsible for collecting all data about patients' sociodemographic characteristics, as well as morbidity and medication, during their consultations.

Data will be electronically stored in a database specifically designed for this study using MS Access 2010. Data will be encrypted and password protected. Information will be treated in strict confidentiality to protect the privacy of patients. The investigators will have no access to the data of the patient, except the one provided by the GP meaning that the only person to know who is being studied is the GP.

Before the collection of data, there will be online reunions with the GPs participating in the study.

Statistical analysis A descriptive analysis will be performed to all study variables, namely the number of valid observations, mean±SD, median and range for quantitative variables and absolute and relative frequencies for qualitative variables. Prevalence of polypharmacy (considering definition: ≥5 drugs vs ≥ the median number) will be calculated together with corresponding 95% CI. Moreover, the prevalence of polypharmacy will be estimated by subgroups, namely age, gender, residence area and formal education. Univariate analysis will be conducted to study the associations between those characteristics and polypharmacy using χ2 test (qualitative characteristics) or t test/Mann-Whitney (quantitative characteristics). Multiple logistic regressions will be carried out considering the presence of polypharmacy as the dependent variable and patients' characteristics as the independent variables in order to calculate odds ratio (ORs) and corresponding 95% CI. Total number of drugs taken by patient and their pharmacological classes will also be summarised together with 95% CI, and multiple regressions may be performed to analyse its association with patients' characteristics. All tests will be two-sided using a significance level of 0.05. Statistical analysis will be conducted using SPSS V.23.0 or higher.

Phase II: patients' perception of barriers to and facilitators of deprescribing, willingness to be deprescribed and actual self-medication in adult patients with polypharmacy attending primary care in Portugal Objectives To assert reasons and facilitators, willingness to be deprescribed and actual self-medication

Design Cross-sectional, analytical study.

Setting It will be the same of the phase I.

Sample size A minimum of 385 patients will be included in phase II in order to obtain a sample with a 95% CI and a maximum precision error of 5%.

Study procedures This phase of the study is expected to start in June 2018. Again, GPs sampling will be made according to existing files and those who accept to participate will recruit their own patients. Patients from phase I can be enrolled in phase II. Assuming that a GP will be able to include at least 6 patients in a 3-week period, a total of 65 GPs has to be enrolled in the study, with the same distribution of the phase I. Enrolled GPs will be instructed to invite all patients meeting the eligibility criteria.

Data collection The collection of the data will occur in June 2018. Patient's socio-demographic and clinical characteristics and medication will be registered using the same methodology as described in phase I. The investigators will also collect outcome measures of some tests.

Statistical analysis Descriptive statistics will be computed for all variables together with 95% CI whenever relevant and applicable. Associations between qualitative-independent variables will be tested using χ2 test. Comparisons between two or more independent groups regarding a quantitative variable are to be conducted using analysis of variance (ANOVA) or Kruskal-Wallis non-parametric test, if normality assumption is not met. ANCOVA may also be used to adjust for potential confounding factors. Associations between quantitative independent variables will be analysed using Pearson's or Spearman's correlation coefficient depending on normality assumption. All tests will be two-sided, considering a significance level of 0.05.

Phase III: impact of enablement of older adults in their willingness to deprescribe and quality of life Design Non-pharmacological randomised clinical study, intended to last for six months.

Setting Primary Care Centres in Portugal will be randomly selected from six Health Centres of Centre of Portugal (Aveiro, Castelo Branco, Coimbra, Guarda, Leiria and Viseu)

Sample size Will be created two groups with a minimum of 190 patients each (one will be the intervention group and the other the control).

Study procedures This phase of the study is expected to start in September 2019 and will last for 6 months.

Again, GPs sampling will be made according to existing files and those who accept to participate will recruit their own patients. Patients from previous phases can be enrolled in phase III. Assuming that a GP will be able to include at least 6 patients, a total of 64 GPs has to be enrolled in the study. Enrolled GPs will be instructed to invite all patients meeting the eligibility criteria.

Two groups will be created with a minimum of 190 patients each, one of which will be composed from patients from the region of Aveiro, Coimbra and Guarda and the other from patients from the region of Castelo Branco, Leiria and Viseu. In the intervention group the investigators will give enablement tools and talks with their GPs about how to issue the problem of polypharmacy. The information given in this group will result from the knowledge obtained in phase II in the shape of small leaflets and other information materials to be made according to the best practice, to be given and remembered at scheduled times to the intervention group.

Data collection The collection of the data will occur in the beginning and end of phase II. Patient's socio-demographic and clinical characteristics and medication will be registered using the same methodology as described in phase I. The investigators will also collect outcome measures of applied tests.

Statistical analysis Descriptive statistics will be computed for all variables together with 95% CI whenever relevant and applicable. Associations between qualitative-independent variables will be tested using χ2 test. Comparisons between two or more independent groups regarding a quantitative variable are to be conducted using analysis of variance (ANOVA) or Kruskal-Wallis non-parametric test, if normality assumption is not met. ANCOVA may also be used to adjust for potential confounding factors. Associations between quantitative independent variables will be analysed using Pearson's or Spearman's correlation coefficient depending on normality assumption. All tests will be two-sided, considering a significance level of 0.05.