Adherence to hormone therapy among women with breast cancer

Background Despite the excellent results obtained with hormone therapy, the long treatment period and the side effects associated with its use make patient adherence difficult. Moreover, certain aspects of health care can mitigate or exacerbate non-adherence. This study aimed to identify the factors associated with adherence to hormone therapy for breast cancer, with the goal of contributing to the reformulation of the care process and to improvements in outcomes. Method This was a retrospective longitudinal study based on secondary data. The study integrated and analyzed data from a cohort of 5,861 women with breast cancer who were identified in the databases of the Brazilian National Cancer Institute [Instituto Nacional de Câncer - INCA] and the Unified Health System [Sistema Único de Saúde - SUS]. All of the patients were treated at INCA, which dispenses free medication, and the follow-up period lasted from 01/01/2004 to 10/29/2010. The outcome of interest was hormone treatment adherence, which was defined as the possession of medication, and a logistic regression model was employed to identify the socio-demographic, behavioral, clinical, and health care variables that were independently associated with the variations in this outcome. Results The proportion of women who adhered to hormone therapy was 76.3%. The likelihood of adherence to hormone therapy increased with each additional year of age, as well as among women with a secondary or higher level education, those with a partner, those who underwent surgery, those who had more consultations with a breast specialist and clinical oncologist, and those who underwent psychotherapy; the effect for the latter increased with each additional consultation. Conversely, the likelihood of adherence was lower among patients at a non-curable stage, those who were alcohol drinkers, those who received chemotherapy, those who had undergone more tests and had more hospitalizations, and those who used tamoxifen and combined aromatase inhibitors. Conclusion This study shows that approximately a quarter of the women with breast cancer did not adhere to hormone treatment, thus risking clinical responses below the expected standards. It also identifies the most vulnerable subgroups in the treatment process and the aspects of care that provide better results.


Background
Studies of endocrine therapy for breast cancer treatment have been increasingly performed throughout the world [1] due to the large patient volume [2], the long treatment duration [3], the optimal obtained results [4,5], and the adverse drug effects [6,7].
Adjuvant endocrine therapy or hormone therapy for breast cancer involves the use of hormone suppressants or similar substances to inhibit tumor growth [7]. This type of therapy is associated with lower rates of disease recurrence and metastasis and, thus, improves mortality rates and disease-free survival [4,8]. Consequently, its use and related expenses have increased significantly in recent decades [2].
The duration of hormone treatment is relatively long, and patients are required to use the medication daily for 5 years [3] to obtain the maximum benefits [9]. The increase in oral medication use has highlighted a potential problem in the adherence to cancer treatment in both health and economic terms. If patients do not take their medications, they will not benefit from them, thus resulting in increased mortality and morbidity as well as increased costs and consumption of care resources. Moreover, if the doctor is unaware of the patient's nonadherence to oral therapy, he/she might attribute the disease progression to the ineffectiveness of the drug and change the regimen unnecessarily [10] or prescribe a second-or third-line medicine with greater side effects or costs. In Brazil, breast cancer is the leading cause of cancer deaths among women (12,852 deaths in 2010), and it is estimated that in 2013, 52,680 women will be diagnosed with this type of tumor [11].
Despite the excellent results obtained with hormone therapy, the long treatment period and the side effects associated with its use make patient adherence difficult, which can lead to adverse clinical outcomes [12,13]. Moreover, certain aspects of health care (e.g., the doctor-patient relationship and the management of side effects) can mitigate or exacerbate non-adherence [1,8], even for patients with a good prognosis [14,15].
This study aimed to identify the factors associated with adherence to hormone therapy for breast cancer, with the goal of contributing to the reformulation of the care process and to improved outcomes.

Methods
A hospital-based population retrospective longitudinal study was conducted with secondary data from women with breast cancer who received hormone therapy prescriptions from the National Cancer Institute (Instituto Nacional de Câncer; INCA) of Brazil.
INCA is the Ministry of Health reference center for cancer policies and care [16] and as such, INCA integrates the public health system and provides all inpatient and outpatient oncological treatment modalities free of charge. It is also the largest breast cancer care provider in the state of Rio de Janeiro, which has the highest incidence of this disease in Brazil [11].
The study included all women with breast cancer who appeared in the hospital-based cancer registry (hospitalbased cancer registry; RHC) between 2002 and 2008 who started hormone treatment with tamoxifen (TMX) and/or the aromatase inhibitors (AIS) letrozole or anastrozole on or after 01/01/2004, and who received dispensed medication more than once before 10/29/2010, according to the INCA pharmacy (the only medicine dispensary for those women).
An integration and analysis of the information found in the following databases was performed as follows: After combining the databases, the differences in dates between the beginning of hormone treatment and the diagnosis of breast cancer were calculated. There were 198 cases with negative values; these were most likely due to typographical errors, and the data were subsequently corrected according to the following procedures.
(1) If the initial hormone treatment date was ≤ 3 months earlier than the date of diagnosis, the diagnosis and the initiation of hormone treatment were assumed to coincide (i.e., the difference was equal to 0). (2) If the negative difference between the start of hormone therapy and the diagnosis was > 3 months, and if the second medication dispensing date was consistent with the diagnosis date, the first dispensing date was ignored and replaced with the second date, and the amount dispensed on the discarded date was deducted from the total. With these procedures, it was possible to retain 185 cases in the analysis; however, 13 women were eliminated due to a complete lack of consistency in the data.
Thus, data for 5,861 women were retained for analysis in the integrated database. There was no systematic association between the excluded cases and the variables of interest; thus, there is a low probability that data exclusion due to operational issues might have introduced bias into the study.
Adherence is defined as the extent to which a patient acts in accordance with the prescribed interval and dose of a dosing regimen, and is operationalized, in a retrospective assessment, as the number of doses dispensed in relation to the dispensing period, often called the "medication possession ratio (MPR)" [12,[17][18][19].
It was calculated, for each woman, by summing up all quantities dispensed, and dividing the sum by her time in the cohort, that, in turn, was given by the difference between the last and the first dispensing date summed to the last dispensed quantity, for those that did not die, and between the death and the first dispensing date, for those that died [12,[17][18][19]. We considered the recommendation of a daily hormone therapy (HT) pill for five years. The drugs that INCA dispensed free of charge were TMX and AIS.
The independent variables of interest included sociodemographic variables (RHC) such as the age at diagnosis, education, and marital status; clinical variables (RHC) such as the histological tumor type, stage, laterality, family history of cancer, smoking, and alcohol consumption; and treatment and health care variables (SHI/Absolute). The latter included the type of hormone therapy (e.g., TMX only, AIS only, letrozole or anastrozole, or both TMX and AIS, which indicated a switch from one form of therapy to the other), surgery, chemotherapy (CT), radiation therapy (RT), hospitalizations, clinical consultations with mastologists, clinical oncologists and other doctors, psychotherapy, multi-professional therapeutic support (MTS) (including nursing ambulatory care, nutrition services, physical therapy, speech therapy and psychology), services from a social worker, dentist or pharmacist, diagnostic and therapeutic services (DTS), and the time between the diagnosis and the initiation of hormone treatment.
Bivariate analyses of adherence were conducted with the chi-squared (χ 2 ) test for categorical independent variables and Student's t-test for continuous variables. Multivariate logistic regression was also used, while excluding variables without independent effects (p ≥ 0.15) for parsimoniousness. The analyses were performed with the SAS® statistical software package, version 9.1 (SAS Institute Inc., 2003; Cary, NC, USA).
The study was approved by the Research Ethics Committee of INCA under number 84/2010.

Results
The patient age at diagnosis ranged from 21 to 103 years, with a mean of 57.5 (SD = 13.6) years and a median of 56.6 years. Approximately 50% of the women were between 40 and 59 years old, and only a minority was younger than 40 years. Half of the women in the cohort had a low level of education (e.g., they were illiterate or had an incomplete elementary education), and 10% had completed a higher education level.
The adherence rate was compiled from the information for each time frame as follows: The mean and median follow-up periods were 1225 (SD = 631) and 1209 days, respectively. The cohort showed a mean and median adherence of 86.3% and 94.3%, respectively. If the patients with at least 80% adherence to treatment were assumed to be adherent, the proportion of adherent patients was 76.3%.
The group of adherent women was slightly older than the group of non-adherent women, with mean ages of 58.0 and 56.0 years, respectively (student's t-test, p < 0.0001). Based on the chi-square test, we found associations of adherence with all categorical variables except the histological primary tumor type (p = 0.6049) and laterality (p = 0.2690; Tables 1, 2 and 3). Table 1 presents the results according to sociodemographic variables and shows the lower likelihood of adherence among younger women (<40 years). Table 1 also shows the higher likelihood of adherence among women who had completed second grade or higher and among those with a partner; these factors were found to have borderline associations.
In relation to the clinical variables (Table 2), the results show a higher likelihood of adherence among women with a family history of cancer and a lower likelihood of adherence among women who were alcohol drinkers or smokers. The likelihood of adherence was especially low among those who were diagnosed with cancer at a noncurable stage (III and IV).
With regard to the variables related to patient care (Table 3), there was a higher likelihood of adherence in women who were treated with TMX alone, as well as those who underwent surgery. In contrast, women who were treated with both types of hormone therapy (switching from 1 form of therapy to the other), who underwent radiotherapy, who had more consultations with clinical oncologists, or who underwent seven or more chemotherapy procedures, hospitalizations, and tests all had a lower likelihood of adherence. The results also revealed a lower frequency of adherence among women who had not consulted with a mastologist and among those who had not received professional therapeutic support or psychotherapy. Table 4 shows the multiple logistic regression model that identified the effects of the explanatory variables on adherence to hormone therapy.

Discussion
The observed level of adherence to hormone therapy (76.3%) in this study was consistent with those found in other published studies, that indicated, considering the same criteria to define adherence, proportions of 75% [20] and 72% [22], or a mean adherence of 93% [12]. However, we underline the difficulties of correlating some available previous studies due to differences in adherence definitions, eligibility criteria (e.g., patients with early tumors or only either young or elderly women), analysis methods, and the types of drugs used (only TMX, only IAS, or both).
The method employed in this study, as in others, accounts for medication delivery as a proxy for medication use, what may result in an overestimation of adherence rates. It is a limitation, but it is believed that the bias  can be attenuated when estimations are made on the basis of secondary data from large populations [23]. Another limit of this study was the absence of individual information regarding treatment side effects, which could certainly affect hormone therapy adherence [24].
The truncation of adherence in cases where it reached levels above 100% is justified by the awareness of situations such as the occurrence of deaths, change of medication, and anticipated drug deliveries to meet patient needs.
Only women who received drugs at least twice were included in the study, once the MPR formula requires two dispensing dates. This procedure could have contributed to over or underestimations of the adherence rates, since who received the medication only one time could have been adherent or not to treatment in another place.
Moreover, we underline the difference between the concepts of treatment adherence and persistence, indicating that adherent patients may be non-persistent if they interrupt the treatment before the recommended period of five years.
This study was based on data from a single hospital, and we acknowledge that this might be a limitation. Nevertheless, it is important to emphasize that the hospital allows universal access, restricting entrance only to patients who was treated previously in other locations.  The observed association between adherence and age in this study agrees with some studies [12,19,21,25] yet contradicting others [26]. It can be speculated that the observed lower adherence rate among younger patients is related to the adverse effects of hormone therapy on the women's sexuality, which include fertility issues and menopausal symptoms [13]. Moreover, the association of greater treatment compliance with having a partner, also found in other studies [22,26], might be linked to support reception. This result is consistent with the idea that social support is highly predictive of adherence [1].
In this cohort, there was greater adherence among women with higher education levels, contrasting to findings of Wigertz et al. [26], who did not identify a statistically significant association between adherence and education. Anyway, it must be noted that it is very difficult to ensure that education in the Brazilian context is comparable to that of developed countries, and there are few other national studies on the subject. It is plausible the observation of greater adherence among patients with higher education while considering, among other things, that socioeconomic status has also been identified as predictive of adherence [21].
With regard to the disease stage, the observation of lower compliance among patients at non-curable stages (III and IV) contradicts Wigertz et al. [26], who found greater adherence among women with larger tumors. Kimmick et al. [20] found no association between these variables. Importantly, however, the majority of studies on this topic were based only on cohorts of women with early stage disease, which reflects the reality in developed countries, wherein 60% of tumors are diagnosed at a localized stage and thus women who are fighting the disease are at a great advantage.
In Brazil, despite the efforts of the Ministry of Health to implement national policies for earlier breast cancer detection, the proportion of advanced-stage diagnoses is very high, compared to that in developed countries [27]. The lower adherence rates among women with a noncurable stage at diagnosis emphasize the importance of investments in early diagnosis since it has a direct effect on patient survival [4,8] and also an indirect effect by facilitating treatment compliance. Bivariate analyses conducted in this study identified the association between both smoking and alcohol consumption and lower treatment adherence [21], while the multivariate analysis ratified only the effect of alcohol consumption on the dependent variable.
This study reinforces the importance of surgery in breast cancer treatment. Previously, surgery had been associated with longer survival [27,28], and herein, it was associated with greater patient adherence to hormone therapy. The current study also revealed that patient monitoring by mastologists and oncologists had a positive association effect on hormone therapy adherence [19]. However, a more intensive use of health care resources such as chemotherapy, exams and hospitalizations appeared to associate with lower adherence. This result was not directly observed in other studies but may be indirectly related to the association between lower adherence and comorbidities, observed in other studies [15,20]. Additionally, it should be emphasized that the independent effects of these clinical variables, even after controlling for the disease stage at diagnosis, might reflect the different levels of tumor aggressiveness, different levels of treatment responses, and the presence of comorbidities.
Another notable finding in this study was the independent positive effect of psychotherapy consultations on the likelihood of treatment adherence. This result might be related to those of studies that related psychological problems, particularly depression, to lower treatment adherence [1].
For Kim & Toge [29], a multidisciplinary approach to breast cancer treatment broke down barriers to access, facilitated cooperation between doctors to produce better treatment proposals, and improved the quality of care and patient satisfaction. Although the multidisciplinary team approach to oncology is required by the Brazilian Ministry of Health [16] and is a necessary condition to hospital certification for the provision of oncological care in the Unified Health System (Sistema Único de Saúde; SUS), the present study indicates that between 24.0% and 57.0% of women who were treated at INCA had no records of multidisciplinary therapeutic support consultations or psychotherapy, respectively. Thus, more consultations with nurses, psychologists, social workers and physiotherapists should be encouraged in the clinical monitoring phase.
Steps that facilitate the patient's understanding of the need for treatment and that promote and monitor adherence should be valued. Multidisciplinary therapeutic support consultations were associated with greater adherence to hormone therapy in the bivariate analysis, but were not significant in the multivariate model when psychotherapy consultations were considered separately. The role of such consultations should be further explored in other studies, which may provide more information about their effects and indicate potential changes in their objectives and uses.
With regard to the type of hormone therapy, the poor adherence among patients who used both TMX and AIS has been corroborated in others studies [19,26]. This finding might be due to the frequent musculoskeletal toxicity associated with AIS [30]. Other studies have found no statistically significant differences between the different types of hormone therapy and adherence [31,32], although a lower adherence rate has been noted for anastrozole than for TMX [32]. It is noteworthy that in the current study cohort, the majority (64.4%) of the women were treated with TMX alone, while 29.8% received a combination of TMX and AIS and a few women (5.8%) were treated only with AIS. The indication of each drug according to the patient's morbidity profile was not analyzed.

Conclusion
In this cohort, approximately 25% of the breast cancer patients were considered to be non-adherent to hormone treatment, thus risking a clinical response below the expected standards. Moreover, this study identified the patient characteristics that are associated with greater vulnerability while receiving hormone therapy for breast cancer and reveals the health care practices that facilitate adherence to treatment. Interventions based on these results may help improve patient adherence, quality of care and outcomes.