Potential drug therapy problems: identification and associated factors among Primary Health Care users

Cardiovascular diseases (CVD) and diabetes mellitus (DM) are chronic non-communicable diseases with high prevalence. Several factors contribute to its lack of control, especially those related to pharmacotherapy, often leading to problems related to the use of medication (DTP). Objective: To identify potential DTPs for the treatment of CVD and DM, as well as associated factors, using big data from users of the Unified Health System (SUS) in a municipality. Methods: A cross-sectional study was carried out based on big data of patients for whom at least one medication was dispensed in SUS pharmacies in the primary healthcare network in April 2019 (n = 4,800). Potential DTPs involving the treatment of CVD or DM were identified based on data on medications dispensed, demographic and clinical characteristics. To classify these potential PRMs, the Pharmacotherapy Workup method was used. Logistic regression analyzes were performed to identify factors associated with identifying at least one potential MRP. Results: The results showed that 25% of the population had at least one potential DTP, with a total of 1,914 potential PRMs being identified. In the multivariate model, age group was statistically associated with the identification of at least one potential DTP. Conclusions: This study allows tracing the frequency of important potential DTPs and associated factors, pointing out some priorities that must be addressed by the public primary health care system, supporting the planning of the implementation of drug therapy management services aimed at the studied population.


Introduction
Cardiovascular diseases (CVD) and diabetes mellitus (DM) are chronic non-communicable diseases (NCDs) diseases of high prevalence that stand out among the others and generate considerable impacts on health systems (Filha et al., 2015). In 2019, it was estimated that 18.6 million people died from cardiovascular disease worldwide, representing the leading cause of disease burden in the world as it has been in Brazil Oliveira et al., 2020). Also in 2019, approximately 463 million people were living with DM in the world and 16,8 adults in Brazil, which puts the country in the fifth place among the top ten countries with the highest number of individuals with the disease (Atlas IDF, 2019). In addition, deaths and complications resulting from these NCDs range from vascular lesions to heart diseases and severe kidney diseases, which end up requiring significant expenditures from a country's annual health budget (Dib, 2010;Brasil, 2010). The estimated cost of DM in Brazil, for example, was US$ 22 billion in 2015, with a projection of US$ 29 billion for 2040 (Bahia et al, 2011). For CVDs, in the same period, it was estimated at a cost of R$ 56.2 billion for the country (Stevens et al, 2018).
Several factors contribute to the lack of control for CVDs and DM, especially those related to medication use. As pharmacological treatments for these diseases are continuous and often complex, it is common to identify drug therapy problems (DTPs) among their bearers (Obreli-Neto et al., 2015). DTP are any undesirable event that involves, or is suspected to involve, pharmacotherapy and that interferes with achieving its desired goals. Their prevention, identification and resolution are one of the main focuses of comprehensive medication therapy management (CMM) services (Ramalho de Oliveira, 2011;Cipolle, Strand & Morley, 2012), that, in turn, are based on the theoretical framework of Pharmaceutical Care (Ramalho de Oliveira, 2011;Cipolle, Strand & Morley, 2012).
CMM services have demonstrated a considerable impact on the control of chronic diseases such as CVD and DM, contributing significantly to the improvement of patients' clinical results (Cid, 2008;Borges et al., 2010;Mendonça et al., In this context, this study aimed to assess the profile of medication dispensing for CVDs and DM, to identify potential DTPs for the treatment of CVD and DM, as well as the associated factors, using big data from users of the primary health care.

Type and Place of Study
This is a cross-sectional study based on big data of adult primary health care users residing in the municipality of Congonhas, Minas Gerais, in which it is expected that further studies can bring longitudinal perspectives on the impact of the GTM service in this Congonhas scenario (Yang, West-Strum, 2013). Congonhas is a medium-sized municipality in the central region of Minas Gerais, located 70 kilometers from Belo Horizonte, the state capital (IBGE, 2018).
Congonha's inhabitants have their registration in the primary health care of the Brazilian Unified Health System (Sistema Único de Saúde -SUS) updated annually by the community health agents. This registration contains sociodemographic and health characteristics, including self-reported diagnosis of diseases, such as CVD and DM. Medication dispensing occurs free of charge in the pharmacies of the primary health care of SUS, which are located at primary health care center. Dispensation is registered in an electronic dispensing system that is integrated throughout the municipality.

Study Population and Data Source
The present study was based on data collected through the crossing of two big data sources: the registration of individuals in the primary health care of Congonhas; and the reports of medication dispensing from the SUS´ primary health care pharmacies.
At the time of the study (2019), Congonhas´ inhabitants had their registration last updated in April 2019 and thisdatabase showed 37,085 adults (aged 18 years or more). In the same month of their registration updates(from April 1 to 30, 2019), all inhabitants thathad at least one medication dispensed to them in the SUS primary health care pharmacies of Congonhas were identified, accounting for a total of 4.800 patients, that composed the study population.
The data on mediation dispensing (active ingredients, dosage, pharmaceutical form and quantity dispensed) were taken from the reports of the eletronic dispensing system of Congonhas (Viver System). In the studied period, a total of 11,749 dispensations were identified for the 4,800 patients included in the present study. The data on socio-demographics and of selfreported diseases were collected from registration of the individuals in the primary health care.

Study Variables
During the joining of the individual registration and medication dispensing databases in a Microsoft Excel® software spreadsheet, the following variables were maintained: age; sex; self-report of HT, AMI or DM; number and types of medications dispensed. From the data on medications and self-reports of diseases in the unified database, potential DTPs were identified, taking into account the classification proposed in the Pharmacotherapy Workup (PW) method (Cipolle, Strand & Morley, 2012). To this end, 11 types of potential DTPs were investigated, which are described according to their classification within the PW method in Table 1.

Potential DTPs -Indication
Potential DTP 1 -Unnecessary drug therapy 1.1) Potential therapeutic duplicity identified when there was a simultaneous dispensing of ACEI and ARB, whose simultaneous use does not bring therapeutic advantage (SBC, 2016; AHA, 2017). 1.2) Potential unnecessary medication identified when the patient was over 70 years old, without self-report of AMI or stroke and collected ASA, whose use is not recommended for primary prevention in this age group (ASCEND, 2018; ARRIVE, 2018; ASPREE, 2018). Potential DTP 2 -Needs additional drug therapy 2.1) Potential additional drug need identified when the patient had a history of AMI, but did not collect a β-blocker at the SUS pharmacy, whose use is recommended in these circumstances (SBC, 2016; AHA, 2017). 2.2) Potential additional drug need identified when the patient had a history of AMI, but did not collect ASA, whose use is recommended in these circumstances (SBC, 2015). 2.3) Potential additional drug need identified when the patient had a history of AMI, but did not collect statin, whose use is recommended in these circumstances (ACC, 2019). 2.4) Potential additional drug need identified when no dispensing of medication for HT was identified, but there was a self-report of HT in the patient's record (AHA, 2017). 2.5) Potential additional drug need identified when dispensing medication for DM was not identified, but there was self-report of DM in the patient record (ADA, 2020). 2.6) Potential additional drug need identified when no drug dispensing was identified for ACEI or ARB, but there was self-report of HT and CKD in the patient's record (SBC, 2016; AHA, 2017). 2.7) Potential additional drug need identified when no medication dispensing for ACE inhibitors or ARB was identified, but there was self-report of HT and DM in the patient's record (SBC, 2016;AHA, 2017). Potential DTP -Effectiveness Potential DTP 3 -Ineffective drug 3.1) Potential effectiveness problem identified when the patient presented self-report of HT, but not AMI, and the dispensing of at least one of the drugs was not identified: ACEI, ARB, thiazide diuretic or dihydropyridine CCB, which are drugs of first choice for the treatment of HT (SBC, 2016; AHA, 2017). Potential DTP -Safety Potential DTP 5 -Adverse drug reaction 5.1) Potential safety problem identified when the patient aged 60 years or more and self-reported DM collected glyburide, the use of which is potentially dangerous for the elderly (AGS, 2019). Through the Pharmacotherapy Workup (PW) method, with technical knowledge of pharmacotherapy, the professional is able to assess the indication, effectiveness, safety and convenience of medications being used by the patient, identifying problems related to the use of medications. Those identified in this study were described in Table 1.
The identification of at least one potential DTP for a patient was defined as the dependent variable. As independent variables, the following were investigated: • Sex: female versus male; • Age: the variable was divided into three categories according to the median and 75% interquartile range, generating the categories 18 to 54 years old, 55 to 64 years old, and 65 or more years old; • Number of drugs dispensed: the variable was divided into two categories according to the 75% interquartile range, generating the categories of 0 to 2 drugs, and 3 or more drugs.

Data analysis
Two unit of analysis were used in the present study: dispensed medication, to describe the profile of medications that were dispensed in the studied period; and individuals, to describe the population characteristics, the prevalence of medication use or the prevalence of the identification of DTP. Categorical variables were presented through the distribution of absolute and relative frequencies. Quantitative variables were described using numerical synthesis measures -mean, minimum and maximum standard deviation (SD).
The factors associated with the presence of at least one potential DTP (dependent variable) were analyzed using univariate and multivariate models of logistic regression and logistic regression with sequential deletion respectively, with estimation of the odds ratios (OR) and their respective confidence intervals of 95% (CI 95%). The significance level of 0.05 was adopted in all tests. All analyses were performed using the R® statistical software, version 4.0, to which the data contained in the unified database in Microsoft Excel® software were transferred.

Ethical aspects
This study is an integral part of the project "Clinical and economic results, humanistic, cultural and educational aspects of medication therapy management services in the Unified Health System", approved by the Federal University of   The prevalence of use among the patients and dispensing frequency of the medications used to treat CVDs and DM are described in Table 3. The most frequently dispensed medications for patients with HT were losartan and hydrochlorothiazide; and for DM, metformin. A total of 1,914 potential DTPs were identified. The identification of non-dispensing of antihypertensive drugs for patients with HT (727 potential DTP; 38.0% of the total potential DTP) or non-dispensing of antidiabetics for patients with diabetes (n= 289; 15.1%) represented the majority of the potential DTP identified (sum= 1,016; 53.08%). Also noteworthy was the non-prescription of an angiotensin II converting enzyme inhibitor (ACE inhibitor) or AT1 subtype angiotensin II receptor antagonist (ARB) for patients with DM or CKD (sum= 432 potential DTP; 22.57%) ( Table 4). Approximately 25% of the studied population had at least one potential DTP (n=1.210), with the majority of these having only one potential DTP (n= 707; 14.73%) ( Table 5). In the multiple logistic regression model, only the age group variable was positively and independently associated with the identification of at least one potential DTPin a statistically significant way (Table 6).

Discussion
The analyses on the medication dispensing profile for CVDs and/ or DM presented in this study contribute to expand the knowledge about the most used medications in the municipality for these conditions, in addition to allowing the identification of potential DTPs among primary health care users. In turn, through the inferences made about the potential DTPs identified, it is possible to assess indirectly the potential inadequacy in medication prescription and the factors associated with these problems.
The most dispensed medications for the treatment of HT (losartan and hydrochlorothiazide) and DM (metformine) corroborates whith the first line treatments propose in important guidelines (AHA, 2017; ADA, 2020). On the other hand, a considerable number of potential DTPs was identified (n= 1,914) and for a considerable part of the studied population (25%).
This scenario is worrying since the present study was limited to assessing few types of potential DTP and related only to CVD and DM. Thus, it can be projected that the proportion of the population with actual DTPs might be considerably higher, since for their identification in CMM services, a complete assessment of all health problems and all medications used by patients is Thus, some hypotheses can be raised as to why these patients have not taken their medication from the pharmacy. The first hypothesis would point out the need for health education strategies, since, considering that HT and DM can be asymptomatic or little symptomatic, patients often have little understanding of their risk or the relevance of regular use of medications to control these conditions (Osterberg, & Blaschke, 2005). The complexity of continuous pharmacotherapeutic regimes such as these diseases' can also contribute to hinder treatment adherence (Coleman et al., 2012;Coelho et al., 2017), as well as the number of drugs prescribed and the adverse effects resulting from their use. Failing to properly follow up or abandoning prescriptions leads to increase in the number of hospitalizations and treatment costs, decrease in effectiveness, loss of quality of life and less productivity for the country (Lessa, 2006).
Since the present study is based on secondary data, it is not possible to state whether patients with CVD and DM got their medications in any way other than SUS pharmacies. However, it is noteworthy that only registered patients who had collected at least one drug from municipal pharmacies were included in the study, which minimizes the chance of having included a population with low dependence on SUS. In addition, in the municipal registry, it was identified that a small portion of the respondents (about 24% -results not shown previously) had health insurance, which reinforces the notion that the population of Congonhas considerably relies on SUS.
Another frequent DTP was the non-use of ACEI or ARB among patients who self-reported CKD or DM (n= 432; 22.56%). The use of these classes of drugs is preferred among patients with such diseases, once the literature points out that they promote the reduction of morbimortality (Mishima et al., 2019;LV et al., 2018).
In the multivariate analysis, only age groups were positively and in a statistically significant way associated with the identification of at least one potential DTP (age group 55-64 years -OR= 3.93; 95% CI= 3.28-4.71; 65 or more -8.34; 95% CI= 7.01-9.91). This demonstrates a strong association between the possibility of identifying at least one potential DTP in the pharmacotherapy used and the different age group corresponding to older patients. Therefore, the present study reinforces the notion that a priority and holistic evaluation of the pharmacotherapy used by the elderly is extremely important. The definition of age groups associated with the identification of potential DTPs makes it possible to establishthe prioritization of these patients, being an important criterion for inclusion in CMM services (Santos et al., 2019). CMM enables the identification of situations that can cause DTPs among these patients with knowingly complex pharmacotherapy, guiding the implementation of preventive action to the occurrence of negative clinical results through a complete clinical pharmaceutical service (Zermanski et al., 2001).
The study has a limitation in fact that it is based on self-reported diagnoses recovered from a secondary database for patients who may be uncertain of their diagnosis or even uncomfortable in mentioning them during registration service.
Despite the limitations related to the use of secondary data, however, it is important to highlight it makes it possible to assess the full population of medication users in primary care and allowed the use of a considerable number of data coming from different databases to provide a diagnosis of the potential pharmacotherapeutic needs in the healthcare system. Also, even though the results are based on local Brazilian data, the steps used for the identification of potential DTP proposed in the present study delimits an easily reproducible methodology for evaluating large databases that can help the planning and implementation of CMM services, which will probably encounter CVD and DM as the most prevalent diseases among their patients.
Therefore, the study makes an important population diagnosis, highlighting the need for qualification of the prescription and points out the direction for the implementation of CMM services in the municipality directed at patients diagnosed with CVD or DM, that may then provide identification, prevention and resolution of actual DTPs.

Final Considerations
This study allowed tracing the frequency of potential DTPs and associated factors, pointing out some priorities that should be addressed by the public primary health care system. It also made an important population diagnosis, highlighting the need for qualifying prescriptions in the city and a holistic evaluation of the pharmacotherapy used by patients diagnosed with chronic non-communicable diseases.
With it, it was possible to perceive criteria that can be used to prioritize the inclusion of patients in the CMM service, supporting the planning of the implementation of these services aimed at the studied population.
Considering the elderly as a large part of drug consumers and with a strong association with at least one potential DTP, there is a need for investments in the management of pharmacotherapy for these patients, through a service that provides rationality in the use of medication.
Offering CMM services, a practice in the process of being implemented in several health establishments that have the performance of pharmacists, is challenging. The present study represents an important step in the demonstration of aspects that precede and contribute to the implementation of a new clinical service, identifying relevant inclusion criteria to it and, in addition, being able to use the results obtained for the introduction of a new work routine by pharmacists and their managers.