Objective: To develop and evaluate the effectiveness of a mobile application to support decision making for prescription drugs to elderly patients. Method: Firstly, there will be a two-tier content validation of potentially inappropriate medicines (PIM) for the elderly based on Beers Criteria and STOPP. For the validation, a modified Delphi consensus technique will be applied. Twelve national experts will be invited. For each statement contained in the Beers criteria and STOPP a proposition will be prepared in order to allow each participant to issue an opinion on each of the propositions by assigning a value according to their agreement to the statement. Therefore, the Likert scale with established score of 1 to 5 (completely disagree - strongly agree) will be used. Statements with average scores above 4.0 will be considered consensus. All information regarding PIMs will be made available through an application for tablets. The information on risks and recommendations about the drugs will be available whenever the users enter the medicines name. The application will be developed for the Android platform, which is based on the Java programming language. To evaluate the use of this application 30 doctors working in basic health units and health units of the family in a municipality from Bahia will be asked to join the study. Upon acceptance, the proportion of PIMs' prescription by these professionals for a minimum period of two months will be analyzed. Later, 15 doctors will be randomized into the intervention group and will receive a tablet with the application installed. They will be trained to use it as well as to access to the Evidence Based Health Portal of the Ministry of Health. The control group of 15 other physicians will receive the tablet and will be trained to access the Evidence Based Health Portal. The study will be blind to the participants and to the main investigator. To continue, the information will be analyzed by sampling the PIMs' prescription in both groups. To access this information the prescriptions' duplicate original, filed in pharmacies, patient files or home visit will be required. The PIMs prescription ratios before and after the intervention will be compared.

Step 1:

Initially, data of the drugs contained in Beers and STOPP criteria will be reviewed. The availability of medicines in the Brazilian market will be analyzed through the website of the National Health Surveillance Agency (ANVISA). Subsequently, drug data will be placed in a questionnaire to be sent to national experts in order to establish a consensus on the information presented. The questionnaire assessment will happen in two rounds. The first round consists of sending propositions based on information contained in the Beers criteria, STOPP 2012 and 2006 in order to allow each participant to issue an opinion on each of the propositions by assigning a value according to their agreement to the statement. Therefore, the Likert scale with established score of 1 to 5 (completely disagree - strongly agree) will be used. For the second round, the scores will be analyzed and the propositions in which there is no consensus will again be sent along with the comments of each specialist. In these cases, each specialist may review or maintain their position. It is worth noting that during the application of questionnaires rounds experts do not have access to the identification of their peers. In the second round, Statements with average scores above 4.0 will be considered consensus. The final information will be used for the mobile device application development.

Experts will be selected for convenience. Twelve professionals with experience in geriatrics will be invited via email to participate as experts in the validation of the Beers Criteria and STOPP. This call will include a summary of the project and general information about the Delphi Technique. After acceptance, they will receive a Term of Informed Consent Form (ICF) and an Authorization to Release Name as Specialist / Senior, which must be signed and returned.

The application will be developed for the Android platform, which is based on the Java programming language.

Step 2:

Study design:

Clinical trial, randomized, single-blind, parallel-group. They will be invited to participate in this stage 30 physicians working in primary care of a municipality from Bahia. Upon acceptance to participate in the study and signing of the Informed Consent form (IC), the proportion of PIMs' prescription by these professionals for a minimum period of two months will be analyzed.

After, physicians will be randomized and included in one of two study groups:

Intervention: 15 doctors will receive the tablet with the application installed and will be trained to use it as well as to access the Evidence Based Health Portal of the Ministry of Health.

Control Group: 15 doctors will receive the tablet without the application and will also be trained to access the Evidence Based Health Portal.

After the intervention, the data will be examined again by sampling the proportion of PIM prescriptions by these professionals.

Hypothesis:

The author will evaluate the hypothesis that the frequency of PIM orders will be lower in the intervention group, compared to the control group.

Reference population: The number of participating physicians was determined by convenience. Currently the city has twenty health units (primary care) with internet access. The investigator chose to select two doctors from each unit to ensure the confidentiality of the professionals randomized into the intervention or control group.

Randomization: The random sequence to assign the study groups will be determined by the principal investigator based on a randomization table, constructed by the randomization method interchangeable blocks. A sequence of numbers will be generated and stored in sealed envelopes, individually arranged and marked with serial numbers for each participating physician. The envelope with the lowest number will be opened by the researcher for the inclusion of a new doctor in the study.

Prescription sampling:

In order to calculate the sample of prescriptions that will be verified before and after the intervention the following factors will be considered: expected proportion of PIM prescription before intervention 35%, expected difference between the intervention and control groups after the intervention: 30%. 5% significance level and 80% power of the test. Two-tailed hypothesis test and loss of 20% of prescriptions. The number of prescriptions per doctor should also take into consideration the proportion of elderly patients attended (in the previous three months) per doctor. A sensitivity analysis will be performed to compare the professional groups that meet different proportions of elderly in order to isolate the bias of demand. Patient prescriptions will be randomly selected.

Inclusion criteria:

Doctors who meet elderly people in basic health units, family health units or primary care polyclinics.

Exclusion criteria:

Doctors with residency in medicine-geriatrics or specialization recognized by the Brazilian Society of Geriatrics.

Dependent variables:

Number of orders with at least one PIM prescribed, frequency of PIMs prescribed, types of PIMs prescribed.

Primary outcome - Frequency of PIMs prescribed.

Independent variables:

Age; gender; training time; medical specialty.

Study Procedures:

In order to evaluate the PIM prescription frequency before the intervention, the prescriptions' duplicate originals stored in pharmacies of health facilities or patient records will be used. In occasions where this information is not available, home visits will be carried out to check the first copy of the prescription in the user's possession. In these situations, the user must sign a consent form. The visits will be carried out preferably in the company of a community health agent.

After the allocation of doctors, each group separately will receive training on access to the Evidence Based Health Portal of the Ministry of Health. For this, the investigators will use one of the computer labs on Universidade Federal da Bahia - Anísio Teixeira campus. In case of professionals' low compliance, the training will be held in the health clinics, in the day of the health teams meeting. Then, for doctors in the intervention group, there will be training to access the application as well as lectures related to PIMs.

After receiving the tablets, each doctor will be visited by a project team member at least twice a week during the 2 months of intervention. During these visits the struggles regarding the access to the Evidence Based Health Portal will be checked. It will be checked if doctors consult the application during the service or at other times, even outside working hours. The number of hits will be monitored also through a feature available in the application itself, for this purpose. Those who do not access the application at least once a week will be considered as a loss in the data analysis.

After the intervention period, a new sample of prescriptions for later analysis will be collected.

Data processing and statistical analysis:

The data will be posted and reviewed in a database built in SPSS® for Windows version 23.0, in order to make up the necessary statistical analysis. Statistical analysis will be composed by descriptive analysis of simple frequency and chi-square test to evaluate the difference between proportions of PIM prescriptions in both groups before and after the intervention. The statistical significance level is p <0.05 with 95% confidence interval.

Inclusion criteria:

-Doctors who meet elderly people in basic health units, family health units or primary care polyclinics.

Exclusion criteria:

-Doctors with residency in medicine-geriatrics or specialization recognized by the Brazilian Society of Geriatrics.