Introduction
Nonadherence rates in adjuvant endocrine therapy (AET) for breast cancer range from 41 to 72% at 5 years [
36] including 30–47% patients who have discontinued [
5,
25]. Nonadherence lead to a higher risk of all-cause mortality and recurrence compared to completion of treatment [
16,
33]. Currently, three treatments are available: Tamoxifen, Aromatase Inhibitors, and Fulvestrant (used in the metastatic setting). With breast cancer being the most common cancer for women worldwide [
27] and AET assigned for approximately 75% of all breast cancer cases [
37], it is important to understand which factors are related to nonadherence. Meta-analyses and systematic reviews have reported that the most consistent demographic, clinical and treatment characteristics related to nonadherence in AET are younger age, follow-up care by a general practitioner versus an oncologist or gynecologist, increased costs, medication switch, higher frequency of hospitalizations, and treatment side effects [
9,
34,
36].
Side effects include vasomotor symptoms, arthralgia, nausea, weight gain, vaginal dryness, and many others. Incidence rates of 21–43% were found for hot flushes and arthralgia in clinical trials [
11,
15,
43] and indicate that side effects are common, can be burdensome, and for some, to such a degree that quality-of-life is largely affected [
11]. However, as noted by Cahir et al. [
9], it remains unknown whether it is the side effects per se or poor side-effect coping that leads to nonadherence. Aside from pharmacodynamics of the drug, side effects can be influenced by one’s expectations of side effects occurrence and intensity, a phenomenon known as the ‘nocebo-effect’ [
14]. Studies with breast cancer patients undergoing chemotherapy or AET have found that expecting more side effects can lead to more actual side effects [
12,
38,
41]. Given the relations between nonadherence and side effects, and side-effect and side-effect expectations, investigating the immediate relationship between side-effect expectations and adherence is warranted.
One expectation factor which has been frequently researched with regard to adherence is the patient’s perceived treatment necessity (e.g., to prevent recurrence) and related concerns (e.g., side effects, potential drug dependency). The underlying theoretical framework suggests that patients individually weigh their perceived necessity against their concerns and derive corresponding behavior [
23]. Accordingly, if concerns outweigh necessity, nonadherent behavior would be more probable. The link between necessity–concern beliefs and adherence to AET has been found in a small body of studies [
8,
18,
19,
28,
30,
42], including two longitudinal studies with a follow-up of 2 and 5 years [
18,
30]. However, a recent review argues that the overall evidence remains tentative since most studies were cross-sectional [
34] and some studies used unvalidated measures of necessity–concern beliefs [
8,
18,
30,
42].
The goal of this study was to examine stability and predictive power of initial treatment expectations, i.e., expected side-effect severity, expected coping with side effects, and necessity–concern beliefs on adherence to AET after 24 months in a multivariate model along with clinical, demographic, and psychosocial characteristics. Also, up to now, it remains unknown whether treatment expectations change or remain stable over the course of the treatment. Furthering our understanding of modifiable factors like expectations could be helpful to design interventions with the objective to increase adherent behavior.
Discussion
In a prospective cohort study in primary care with 116 patients, we documented a nonadherence rate of 14.7% after 2 years. Initial side-effect severity and necessity–concern beliefs predicted long-term adherence over and above the influence of sociodemographic, other medical, and psychosocial characteristics. Expected side-effect severity and expected coping with side effects did not predict adherence. Necessity–concern beliefs remained stable on an overall medium level over the course of 2 years, whereas side-effect and coping expectations became less optimistic over time. The trajectories of coping expectations differed by adherence status at 24 months with expected coping being less optimistic among the nonadherent than the adherent patients.
Our findings indicate that, given a scale range from − 4 to 4, a patient with a necessity–concern beliefs value of 1 (i.e., necessity slighter higher than concerns) is two times more likely to be adherent than a patient with necessity–concern beliefs of 0 (i.e., necessity and concerns equally high). A patient with necessity–concern beliefs of 2, again, is four times more likely to be adherent. As for side effects, a patient with mild side effects is four times more likely to be adherent than a patient with moderate side effects. Our finding adds to the line of studies which linked necessity–concern beliefs to AET adherence [
2,
8,
18,
19,
28,
30,
42,
46,
52]. Hereby, our study is the first to report this relationship in a prospective design over 24 months using a validated questionnaire of necessity–concern beliefs. These results also indicate that side-effect severity and necessity–concern beliefs play a pivotal role in AET adherence over and above demographic and clinical characteristics, which did not contribute to the prediction. Except for younger age, the latter finding aligns with reviews and meta-analyses which reported inconsistent associations between adherence and most demographic and clinical characteristics [
9,
34,
36], including quality-of-life [
45], depression [
31], and medication switch, which facilitated adherence in some studies, and nonadherence in others [
34].
The influence of side effects on adherence was also implied by patients’ self-reports, given that five out of 15 patients named “side effects” as their reason for discontinuation. Notably, whereas side-effect severity predicted adherence in the logistic regression model, numbers of side effects did not, suggesting that the link may differ depending on how side effects are operationalized. This has also been reflected by reviews [
9,
34], which reported the relationship between side effects and adherence to be inconsistent across studies. We propose that the link is unequivocal since it might be influenced by side-effect appraisal. For example, a recent qualitative study found that women were more willing to accept side effects when they were aware that the treatment was necessary to prevent recurrence [
35]. Moreover, our findings indicated that side-effect management may play a role considering adherence, giving that, especially among the nonadherent group, expected coping with side effects became less optimistic over time (37.5% vs. 5.4% for nonadherent and adherent patients). As side-effect severity did not change significantly from 3 to 24 months—neither in the overall sample nor in the nonadherent or adherent patient group—these results suggest that expected low self-efficacy to cope with future side effects may influence adherence behavior which goes beyond side effects per se. Taken together, we suggest that burden of side effects may affect adherence behavior, yet should be examined in the context of further related factors, e.g., necessity–concern beliefs and side-effect management.
In the literature, side-effect management is acknowledged as important to maintain or increase adherence to AET [
21,
48], and a variety of recommendations are suggested for different side-effect categories [
17]. However, systematic research examining the degree of its implementation is lacking, whereas trials which investigate the efficacy of those strategies are few (e.g. for hot flushes, see [
44] for arthralgia, see [
3,
13]). Although due to the correlative nature of the data, we cannot answer the question whether patients discontinue treatment because they expect their management to be ineffective or whether they discontinue first and rationalize their behavior thereafter, our findings nonetheless reinforce the importance of side-effect coping in AET. Its pivotal role considering adherence optimization has also been pointed out by both patients [
46,
48] and practitioners [
49] in prior qualitative studies. Furthermore, in a trial of cognitive-behavioral therapy for vasomotor symptoms of breast cancer survivors, beliefs about coping and control of hot flushes were found to be the main mediator of how burdensome symptoms were perceived after the therapy [
10]. Overall, more studies are needed to investigate whether poor coping abilities lead to nonadherence, and to which degree patients wish for further support.
Limitations of this study include a selection bias. The enhanced information about AET which was provided as part of the study might have appealed to women who were more open to the treatment or perhaps even increased patients’ willingness to be adherent. We believe the latter bias to be minor since interventions which aimed at improving adherence by providing information were found to be not effective [
26]. Nonetheless, our nonadherence rate was lower than the rates found in large health plan studies (15% vs. 27%–45%) [
20,
29], which indicates a bias which might have weakened the external validity of this study. By means of a subjective measure to assess adherence, we obviously cannot exclude a reporting bias. However, a single-item, self-report adherence measure has been found to be associated with estrogen serum levels [
7]. Also, we assessed necessity–concern beliefs at 3 months, whereas side-effect and coping expectations were assessed at treatment start. Since expectations might be affected by treatment experiences, comparability of these factors’ trajectories may be limited. Finally, the numbers of nonadherers who were included in stability analyses (
n = 8) and the numbers of patients who have specified the reason for discontinuation were very small (
n = 9). Thus, interpretations must be viewed in light of limited representativeness.
Conclusions
In the context of the present body of research, our findings show a coherent picture indicating the importance of patients’ understanding of the individual necessity of AET. More specifically, it seems important that necessity beliefs outweigh individual concerns, which are an inherent part of patients’ treatment evaluations. Overall, patients reporting more necessity beliefs than concerns and experiencing fewer initial side effects were more likely to be adherent in the long term. Also, positive coping expectations with side effects of the AET decreased over time in nonadherent compared with adherent patients.
To reduce the perceived burden of side effects, practitioners could support patients’ side-effect management [
47]. Follow-up visitations could be used to screen patients with poor coping expectations, who are then provided with individual management strategies. In summary, by addressing benefits of the AET, by exploring potential concerns of a patient, and by offering coping strategies during the course of the treatment, adherence might be optimized in the long term.