Introduction

Osteoporosis (OP) is a chronic condition associated with increased morbidity and mortality and decreased quality of life [1, 2]. Older adults who fracture have a 20 % risk of another fracture within 1 year [1–3] and treatment with a bisphosphonate reduces this risk by 50 % [4–6]. As such, the focus of the 2010 Osteoporosis Canada Guidelines (as well as guidelines endorsed by the National Osteoporosis Foundation and the International Osteoporosis Foundation) is secondary prevention. This entails identifying patients with fragility fractures of the upper extremity, spine, or hip, ordering bone mass density (BMD) tests and offering treatment to those at the highest risk of another fracture [7–9]. Regardless of this guidance, less than 20 % of people are treated for OP in the year post-fracture [1, 2, 6, 9]. Equally worrisome, of those written a prescription for OP treatment, 30 % will not fill their prescription (primary non-adherence) [10] and of those who do fill their first prescription, at least half will stop treatment within 1 year [11, 12]. Indeed, even in the setting of a randomized trial to improve testing and treatment of OP after a fragility fracture, 18 % of those who started treatment in the intervention arm had stopped treatment within 1 year [13]. Several models address how individual factors, such as knowledge and attitudes, and social processes, including identity and support, influence behaviors [14]. However, the patient-level factors related to starting and stopping OP treatments remain relatively under studied and yet critically important to understand [15–17].

To better understand long-term persistence (and non-persistence) with OP treatments, we capitalized on participants in the Comparing Strategies Targeting Osteoporosis to Prevent Recurrent Fractures (C-STOP) trial. This trial, based on a positive pilot study [18], is currently enrolling patients with upper extremity fragility fractures in an active-comparator randomized controlled trial of nurse case management vs a multifaceted intervention directed at patients (i.e., printed materials, education, telephonic counseling) and their physicians (i.e., reminders, opinion leader endorsed guidelines). The nurse case-manager identifies and interviews patients from clinic settings (Emergency Departments and Fracture Clinics), arranges BMD tests, and offers in-person education, counseling, and guideline-based prescription treatments as needed, and then follows patients up for 1 year.

Patients in the case-manager arm of this study represented the “best-case scenario.” For example, they are relatively healthy although they have all recently suffered a fragility fracture and are all at similar risk of re-fracture based on the most recent major osteoporotic fracture and calculated FRAX with BMD scores of moderate to high risk. In addition, they all received standardized print materials with in-person counseling and were all offered treatment based upon standardized treatment algorithms [18]. In this setting of uniform care, we hypothesized we might be able to gain important and nuanced insights into patient-specific reasons for continuing or quitting OP medications. Therefore, rather than use surveys or secondary analyses of various databases as has been done previously [3, 15, 16, 19–21], we undertook a qualitative grounded theory study to understand the processes and reasoning that adults 50 years and older with recent fragility fractures go through when deciding to persist with or stop bisphosphonates in the 1-year post-fracture.

Methods

We used a grounded theory approach as outlined by Corbin and Strauss [22] to understand the processes and reasoning people engage in when deciding to persist with or stop OP medications in general, and more specifically, bisphosphonate treatment after suffering a fragility fracture of the upper extremity. Grounded theory is an inductive approach where models or concepts are “grounded” in the data collected without an a priori hypothesis. We chose to use a grounded theory approach to understand how patients approach a diagnosis of OP and how they negotiate persisting with or stopping treatment. We considered all of these patients (50 years and older with fragility fracture of the distal radius or proximal humerus and low bone mass at one or more skeletal sites) to have a clinical diagnosis of OP [23] and refer to this as OP throughout. When referring to BMD test results alone, we refer to osteopenia (T-score ≤1.0 to −2.5) or OP (T-score ≤2.5) [18, 23] and always make this distinction explicit.

Data collection and measurements

The primary study outcomes of the larger C-STOP trial were ascertained at 6 months, and the research team remained blinded to these data. That said, during 1-year follow-up of this population, we identified nurse case-management patients who had started bisphosphonate treatment at any time up to their 1-year data collection. We enrolled 150 patients in the nurse case-management arm of the larger study. Thus, all patients seen in the clinic were potentially eligible, and the only patients seen in the clinic were trial patients. We did not sample patients “randomly” for the purposes of generalizability of the results as conceptualized in quantitative research. Instead, we sampled patients based on their ability to act as key informants (i.e., someone with direct experience with the phenomenon who is able to reflect on and articulate their experience) for the purposes of transferability of the results to similar populations and contexts. At this time, we purposefully sampled patients who had started bisphosphonate treatment (by study treatment algorithm, either generic alendronate or risedronate prescribed weekly [18]) and reported whether they were still persistent with treatment at 1 year (“persisters”) or they had stopped treatment (“stoppers”). We verified all self-reported treatment data independently by using local pharmacy dispensing records.

As the study progressed, we used theoretical sampling, one of the core strategies of grounded theory [22, 24]. Theoretical sampling is responsive to the data in that the researchers sample concepts and not persons [22]. Since the nurse case-manager had met and remained in contact with all trial patients, she helped identify potential patients who could talk directly to our various emerging concepts. For example, we recruited patients diagnosed with either osteopenia or OP as defined strictly by BMD test results to elicit a deeper understanding of perceived severity, when this specific concept emerged from analysis. As well, we re-interviewed several patients as themes and concepts emerged.

A trained researcher conducted qualitative in-person interviews using a semi-structured interview guide (Appendix), which was refined as data analysis progressed. Interviews were conducted in the same location as case-management appointments at a single center and lasted up to 45 min. Interviews were digitally recorded for subsequent analysis and verified for accuracy. The University of Alberta Health Ethics Research Board approved the parent trial and this qualitative sub-study, and all subjects provided written informed consent.

Analysis

We used constant comparison to analyze the data, using open coding and memos [22, 25, 26]. One analyst (LW) identified emerging codes and concepts. The research team held regular meetings to review code definitions, emerging concepts, and memos and to refine the interview guide. We conducted data collection and analysis concurrently until theoretical data saturation [27, 28] was achieved, that is, the major concepts were well defined and explained and no new concepts or themes emerged [22]. We implemented several strategies throughout the research process to ensure rigor including methodological coherence [29], peer debriefing [24], reflective writing [22], and maintaining an audit trail [24]. Data were managed and analyzed using Atlas.ti version 7 (Berlin, Germany, Scientific Software Development GmbH) [30].

Results

From April 2014 through March 2015, we conducted 21 interviews with 12 different patients, including a second follow-up interview with the initial 9 patients to address questions that emerged through concurrent data collection and analysis. Of the 12 patients recruited, 7 persisted with bisphosphonates and 5 stopped. Treatment status was determined at the last point of contact (e.g., a patient who was a persister at her first interview but a stopper at her second interview was classified as a stopper). Of note, three patients changed their treatment status during the study. Two thirds of patients were 60 years or older, 75 % were women, 67 % had osteopenia (T-score ≤1.0) at one or more skeletal sites according to BMD testing, and 92 % had better-than-average OP-related knowledge compared with the average score (57 %) reported previously in a similar trial population [31] (Table 1). By trial design, no patient had been taking OP medications at the time of fracture.

Table 1 Patient characteristics according to treatment status
Full size table

Three major themes emerged. We provide a list of codes, sub-codes, and supplemental illustrative quotes that comprise each theme in Table 2, and we detail each theme below.

Table 2 Codes, sub-codes, and selected supporting quotes by three main themes
Full size table

Theme 1: negligible appreciation of risk regarding severity and impact of OP

Regardless of treatment status, patients identified similar reasons for deciding to start bisphosphonates. These reasons included: (1) being advised to do so by a healthcare professional, (2) to prevent the progression of their condition or to maintain current bone health, or (3) to stay healthy.

Nonetheless, and despite the fact that all but one of these patients had better-than-average OP-related knowledge, they did not consider OP to be a serious health condition, particularly compared with other diseases like cancer. Furthermore, based on their understanding of their own BMD test results, patients (all of whom had a fragility fracture) perceived a test result of osteopenia as less serious than OP and defined it as pre- or borderline-OP, at the beginning stages of OP, or within acceptable range. In addition, patients were not concerned about their bone health or OP prior to their fracture. Many patients believed OP was a natural part of aging, especially for women, and some commented that they did not experience symptoms or feel differently because they now had an OP diagnosis.

Of particular note, patients generally perceived minimal susceptibility to the potential negative consequences of OP in the future. They believed that it was in their control to prevent future fractures by being more careful or avoiding falls. From their perspective, the fracture was their fault. However, some patients said the fracture was unavoidable or would have happened to anyone, indicating that it was not a result of compromised bone health. Generally, patients explained that an upper extremity fracture did not affect their lives substantially in the way that a hip fracture would.

Theme 2: ongoing evaluation of risks vs benefits of treatments

The main difference between patients who persisted with or stopped taking bisphosphonates was in their evaluation of its risks and benefits. Patients who persisted perceived greater benefit than risk from taking OP medications. They believed that medication was required or the best option to treat their condition. In addition, they perceived their risk of future fracture without OP treatment as high and said they would continue taking prescription treatment even if they had another fracture. Lastly, they did not report experiencing any side effects.

In contrast, patients who stopped treatment within 1 year believed that bisphosphonate treatment was optional and no more beneficial than exercising and taking vitamin D and calcium. They perceived their risk of future fracture without treatment as low. In addition, they reported concerns about side effects and this was enough to tip the balance towards non-persistence.

Theme 3: re-evaluation of severity and impact of OP vs risks and benefits of treatment over time

Patients reported changing (or having the potential to change) treatment status even 1-year post-fracture because they reassessed the severity and impact of OP as well as the risks and benefits of bisphosphonate treatment over time. Two patients who restarted OP medications after initially quitting explained that they have OP (rather than just osteopenia according to a BMD test) and that prescription treatment was now required rather than optional. Conversely, the patient who persisted with bisphosphonates but then stopped 1-year post-fracture explained that she had osteopenia, and not OP, according to her BMD tests and that there had been no further decline in her bone health as far as she could tell. In addition, she explained prescription treatment was but one option and not necessary to treat OP. Lastly, she reported experiencing what she perceived as side effects of bisphosphonates, including pain in her hip when laying down.

Even patients who did not change treatment status reported the potential to change status at almost any time post-fracture. For example, patients who persisted with treatment described “lapses” in adherence, primarily when traveling. One patient commented that she forgot to take her medication while traveling because OP is not life threatening. In addition, patients who persisted reported the potential to stop taking OP medications in the future if, for example, they started experiencing side effects or saw no improvement in their BMD test results. On the other hand, several patients who stopped explained that they would consider re-starting treatment in the future if, for example, they were diagnosed with OP (rather than osteopenia) based on their BMD results, believed that OP could affect their life negatively, or were informed that bisphosphonates were their best treatment option. For example, the patient who stopped prescription treatment 1-year post-fragility fracture said that she would re-start if her family physician diagnosed her with OP and told her that OP medications were required to treat it.

Discussion

In this grounded theory study examining how patients’ understand the clinical problem of OP and how they decided to persist with or stop treatment 1-year post-fracture, three broad themes emerged. First, patients perceived that OP was not a serious health condition and that its impact was negligible regardless of their decision to start treatment. Second, the main difference between persisters and stoppers was in their ongoing evaluation of the risks vs benefits of bisphosphonate treatment, where patients who persisted perceived more benefit and less risk than patients who stopped. Third, we found that patients had the potential to change treatment status even 1-year post-fracture because they re-evaluated the severity and impact of OP vs the risks and benefits of treatments over time. We have attempted to represent and conceptualize this decision-making process in Fig. 1.

Fig. 1
figure 1

Conceptualization of the decision-making process to persist with, stop, or re-start prescription treatment for clinical OP over time

Full size image

According to the Information, Motivation and Behavioral Skills model [32], people are more likely to initiate and maintain health-promoting behaviors, including medication adherence [33, 34], which produce positive outcomes when they are well-informed, motivated to act, and possess the skills and confidence to take action. Perhaps the most important insight gained from our study is that even in a healthy, knowledgeable, and well-informed high-risk population with recent fragility fracture, the decision to start and then persist with long-term treatment is not a binary “yes-or-no” process situated at one point in time (i.e., the diagnostic moment). Rather, patients appear to re-evaluate and re-frame their willingness to adhere to OP medications over time based on the perceived severity and impact of OP on the one hand vs the risks and benefits of bisphosphonate treatment on the other.

The results also suggest that these risk-benefit re-assessments are driven by multiple factors related to time. First, time from the clinical fracture likely plays some role: as the memory of the clinical fracture event fades, the perceived severity diminishes. Second, the potential negative impact of OP also diminishes over time while the risks of bisphosphonate treatment become more prominent over time, particularly since the absence of another fracture is the only tangible benefit of ongoing treatment. This has implications for facilitating patient-centered decision-making [35, 36] beyond the diagnostic moment where ongoing clinical dialog is going to be needed to support persistence and to continuously re-engage patients who stop treatment. This finding is particularly relevant as demonstrated by a recent front-page article in The New York Times reporting that people tend to avoid effective osteoporosis treatment based on the fear of extremely rare side effects and calling for a more balanced perspective in weighing the favorable benefit-to-risk profile of treatment [37].

Similar to our findings, others have found that patients misunderstood their diagnosis in the context of having suffered a fragility fracture [38, 39]. As such, as suggested by an expert consensus [23], our findings indicate we should use the term “clinical OP” and refrain from using or overly emphasizing BMD test result labels such as “osteopenia” when designing patient-directed interventions or educational materials. Others have also found that patients did not consider OP serious [17, 40–42]. For example, Sale et al. found that patients did not consider compromised bone health as serious, although this perception was not related to use of OP medications [17] possibly explaining why we found this opinion common to both persisters and stoppers. In addition, other studies have commented on the challenges of appropriately treating a clinically silent disease [39, 43, 44]. Challenges related to persistence in taking long-term OP medications are not necessarily unique and seem similar to the problems with adherence and persistence demonstrated for other chronic conditions that are also “asymptomatic” such as hypertension or dyslipidemia [45, 46].

Patients may also under-estimate their fracture risk or perceive themselves as immune to the effects of OP [43]. Indeed, other studies have shown that patients believed that the fracture was unavoidable and caused by non-preventable circumstances [40, 41] and, therefore, within the realm of normal everyday occurrence. Conversely, patients may also believe the fracture was related to their own unsafe behaviors and, therefore, within their control to modify to prevent the next fracture [41, 43]. However, both of these rationalizations neglect the fact that healthy bones would not have fractured under the same low trauma circumstances.

In addition, patients appear to re-evaluate risks vs benefits of treatments for OP over time, and it is already known that patients are very concerned about possible or perceived (vs actual) side effects of bisphosphonates, and this strongly influences their decisions regarding treatment persistence [47–49]. Our work suggests that a major component of patients’ risk-benefit analysis relates to their belief about whether treatment with bisphosphonates is required (rather than “optional”) to decrease their risk of another fracture. This finding has implications for the design and content of interventions intended to promote persistence among patients who have started treatment regardless of their concerns related to potential or actual side effects. Clearly, OP medication use varies over time with patients starting, stopping, and re-initiating treatment [15, 20, 47, 50, 51]. This process of re-evaluation might be considered promising by some (because patients who stopped bisphosphonates may re-start) although others believe that it is very concerning [15, 47]. For example, Klop et al. found that persistence after re-starting OP medications remained low and was substantially lower than initial rates of persistence [15].

Despite its strengths (grounded theory approach, theoretical sampling, uniform patient population), our work has limitations. First, even by the standards of qualitative inquiry, some might consider our sample small; however, we sampled until data saturation was reached, employed multiple checks for rigor and validity, and selected participants from both ends of the persistence spectrum. Second, it was somewhat difficult to use the technique of theoretical sampling within the constraints of an ongoing trial. For example, we were unable to sample for patients who changed treatment status more than 1 year after their fracture and we had no access to patients who dropped out of the study altogether. Third, we did not set out to specifically examine sex and gender differences in health behaviors. While this is an important area of inquiry, by our grounded theory design, no sex or gender differences in pill-taking behavior were evident in our study, so we did not pursue increased sampling of constructs related to sex and gender. Last, all subjects were relatively healthy, of good socioeconomic status, had good knowledge related to osteoporosis, had no contraindications to bisphosphonate treatment, and all had an upper extremity fragility fracture, and our findings were based on the experiences of trial patients with universal healthcare coverage from one Canadian province. Thus, some might not consider our results transferrable to other populations, including patients with other types of fractures including hip or spine or settings.

The processes and reasoning related to persisting with or stopping bisphosphonate treatment for OP after a fragility fracture are complex and dynamic over the long term. Our findings suggest two potential areas to leverage for providers and those designing interventions. First, healthcare providers should reinforce the severity and risks and potential harms related to untreated clinical OP and refrain from using BMD test-based labels such as osteopenia, which appears to diminish perceived disease severity among patients. Second, healthcare providers need to review regularly the very favorable benefit-to-risk profile of treatment for this high-risk group of patients. Perhaps what has become most clear through this research is that starting treatment for OP is the beginning of a long-term negotiation with patients in an effort to prevent the next fracture and its potential consequences.