The effects of low persistence on fracture risk have not been fully addressed. The objectives of this study were to describe the persistence and compliance with bisphosphonates and to evaluate the association with fracture risk. The General Practice Research database was used to identify patients ≥18 yr of age prescribed alendronate or risedronate. The follow-up was divided into periods of current and past use. Time-dependent Cox regression was used. The study population included 44,531 patients; 58.3% of the patients continued bisphosphonate treatment for >1 yr and 23.6% for >5 yr. The risk of hip/femur fracture (adjusted relative rate [RR], 0.78; 95% CI, 0.64–0.94) and osteoporotic fracture (RR, 0.85; 95% CI, 0.76–0.94) were lower with current compared with past bisphosphonate use. The largest reduction in hip/femur and osteoporotic fracture risk was observed in patients treated for at least 6 mo and no reduction in those treated for <6 mo. The risks of hip/femur and osteoporotic fractures followed the pattern of nonosteoporotic fractures in the first 6 mo but then started to reduce after 6–12 mo of treatment. Increased risks of osteoporotic and hip/femur fractures were found in patients with low compliance. Use of bisphosphonates was associated with fracture risk reductions after 6–12 mo of treatment, but only 58% of the patients were treated for at least 1 year. Improvement in long-term persistence to bisphosphonate treatment may be important to reduce the impact of osteoporosis-related fractures.
Osteoporotic fractures represent a huge public health burden: worldwide, there were an estimated 1.66 million hip fractures in 1990. In recent years, randomized clinical trials have shown that bisphosphonates reduce the risk of hip fracture. Hip fracture is the most serious consequence of osteoporosis, leading to considerable morbidity and mortality. As osteoporosis is a chronic condition, treatments need to be taken long term to be effective. Several studies have shown that persistence with oral osteoporosis treatment (i.e., continuation of treatment over time) is suboptimal. However, the effects of low persistence on the risk of fracture have not yet been fully addressed. The objectives of this study were to describe the persistence and compliance with oral bisphosphonates and to evaluate the association with the risk of fracture. Persistence was defined in this study as the probability of continuing treatment over time (i.e., receiving repeat prescriptions over time), and compliance was defined as the proportion of the total treatment time that is covered by the dispensed medication.
MATERIALS AND METHODS
The data in this study were obtained from the General Practice Research Database (GPRD). The GPRD comprises the computerized medical records of general practitioners (GPs). GPs play a key role in the UK health care system, because they are responsible for primary health care and specialist referrals. The data recorded in the GPRD include demographic information, prescription details, clinical events, preventive care provided, specialist referrals, hospital admissions, and their major outcomes. The data collection started in 1987 and ended, for this study, in 2006.
The study population consisted of women or men ≥18 yr of age who received a prescription for alendronate or risedronate. Daily alendronate was first available for routine use in the United Kindom in 1995, weekly alendronate in 2000, daily risedronate in 2000, and weekly risedronate in 2003. Patients with a history of Paget's disease or <1 yr of follow-up before the first alendronate or risedronate prescription were excluded. Patients were also excluded if they were given both alendronate and risedronate on the same date or both weekly and daily dose regimens on the same date. Ibandronate was only recently introduced onto the UK market and thus was not considered in this study. A restricted cohort was generated including only those patients with at least two bisphosphonate prescriptions. It was considered more likely that patients in this second cohort had been exposed to bisphosphonates.
Persistence and compliance to bisphosphonate treatment
The persistence to bisphosphonate treatment was measured by calculating the proportion of patients who continued treatment over time. This was determined by measuring repeat prescribing of bisphosphonates within 3 mo after the expected end of use of a prescription. Cox proportional hazard regression models were used to determine the relative rate (RR) for persistence. The compliance to bisphosphonate treatment was measured by dividing the expected duration of bisphosphonate use (based on the number of tablets prescribed and prescribed daily dose) by the time between the two prescriptions. This medication possession ratio was based on the average of all previous prescriptions. Ordinal logistic regression was used to estimate the ORs for compliance.
The following characteristics were evaluated for their association with treatment persistence and compliance: age, sex, fracture history, history of peptic ulcer or upper gastrointestinal symptoms in the year before start of bisphosphonate treatment, dosing regimen of bisphosphonates (weekly or daily), calendar year, recent use of oral glucocorticoids (prescription in the 6 mo before) or upper gastrointestinal medication (antacids or ulcer-healing drugs), recent hospitalization, number of nonbisphosphonate prescriptions, or prescribing by the GP of calcium and/or vitamin D in the 3 mo before.
Determinants of the risk of fracture
The study population was followed for the occurrence of fracture. The study follow-up started with the first bisphosphonate prescription and ended in the case of a patient's death, transfer out of the practice, or end of GPRD data collection, whichever date came first. The following fracture types were considered: osteoporotic fracture (defined as a hip/femur, vertebral, radius/ulna, humerus, rib, or pelvis fracture), hip/femur fracture, radius/ulna fracture, clinically symptomatic vertebral fracture, humerus, rib, and pelvis fracture. The incidence rate was estimated by dividing the number of patients with a fracture by the total follow-up time.
The associations between fracture and onset and offset of bisphosphonate effect and compliance were also assessed. The follow-up of patients was divided into periods of current and past use (with patients moving between these exposure categories). Current use was defined as the time from the date of a prescription until its expected duration of use. Where there was a repeat prescription within 3 mo, patients continued to be currently exposed. Past use was defined as the period of time starting 3 mo after the expected end of a bisphosphonate prescription. Time-dependent Cox regression was used to estimate the RR of fracture.
A descriptive analysis of the pattern of changes in fracture risk was also conducted to evaluate whether the hazard rate (i.e., absolute risk) for osteoporotic fracture would diverge from that of non-osteoporotic fracture over the duration of bisphosphonate treatment. The hazard rates were estimated by dividing the follow-up time into 100 periods and by calculating the absolute rate within each small period. These estimates were smoothed using the methods proposed by Ramlau-Hansen. The ratio of the smoothed hazard rates of osteoporotic over non-osteoporotic fractures was standardized to that of the first period of time.
As shown in Table 1, the study population included 44,531 patients who were prescribed either alendronate (N = 32,980) or risedronate (N = 11,551). More than one half of the patients (57.0%) were ≥70 yr of age, and 81.2% were women. About one quarter of the bisphosphonate users had a fracture history, and 32.0% recently used an oral glucocorticoid. More than 80% of the study population started on a weekly bisphosphonate, although this percentage varied over calendar time (because of the later market introduction of weekly bisphosphonates). There were 38,669 patients who received at least two bisphosphonate prescriptions.
Table Table 1.. Baseline Patient Characteristics
Persistence and compliance to bisphosphonate treatment
It was found that 58.3% of the patients continued bisphosphonate treatment for >1 yr and 23.6% for >5 yr. Patients using weekly bisphosphonates were more likely to continue bisphosphonate treatment and be compliant compared with patients using daily bisphosphonates (Table 2).
Table Table 2.. Predictors for Persistence and Compliance With Bisphosphonates
Determinants of the risk of fracture
Table 3 shows the results of the comparison of fracture risk with current and past bisphosphonate use. There was a 22% lower risk of hip/femur fractures during current use compared with periods of past use. In patients without recent prescribing of oral glucocorticoids, the fully adjusted RR during current use was 0.83 (95% CI, 0.73–0.94) for osteoporotic fractures and 0.71 (0.57–0.88) for hip/femur fractures. In recent glucocorticoid users, these figures were 0.86 (95% CI, 0.69–1.07) and 1.02 (95% CI, 0.67–1.55), respectively.
Table Table 3.. Risk of Fracture With Current Bisphosphonate Use Compared With Past Use (in the Cohort of Patients With at Least Two Prescriptions)
When stratifying past use by time since discontinuation and current use by time since starting bisphosphonate treatment, no reduction in hip/femur fracture risk was found during current use where treatment was recently started, whereas current use of at least 24 mo showed a reduction in risk of 34%. The pattern of hazard rates showed that the hazard rate of hip/femur fractures diverged from that of non-osteoporotic fractures only after ∼1 yr of bisphosphonate treatment (Fig. 1). The fracture risks during treatment of patients who started alendronate remained proportional over time compared with those starting with risedronate, without any statistically significant divergence of risks (Fig. 2). As shown in Fig. 3, there was an inverse association between risk of fracture and compliance for osteoporotic and hip/femur fractures (linear trend p < 0.05).
This study found that only 58% of the patients continued treatment with bisphosphonates for 1 yr. The rates of osteoporotic and hip/femur fractures only started to diverge from that of non-osteoporotic fractures after 6–12 mo of bisphosphonate treatment. There was no suggestion of a residual effect after stopping bisphosphonates.
Large randomized trials have shown that bisphosphonates reduce the risk of hip and vertebral fracture after several months of treatment. The randomized trials have shown considerably smaller effects of bisphosphonates on the risk of non-osteoporotic fractures compared with that on hip or vertebral fractures. Therefore, one would expect the hazard rates of osteoporotic and non-osteoporotic fractures to diverge over time. The findings of this study are broadly consistent with these clinical trial results, with a divergence in the hazard rates of hip/femur and vertebral fracture from that of non-osteoporotic fractures after 6–12 mo of treatment. This study raises the possibility that the considerable proportion of patients who did not persist with bisphosphonate treatment long term may not have experienced any benefit from treatment. An improvement in long-term persistence is, therefore, important in reducing the incidence of osteoporosis-related fractures and their subsequent impact.
We found that the fracture risks remained proportional over time between alendronate and risedronate users, without any divergence of risks over duration of treatment. These findings contradict those reported in a 1-yr study in a U.S. claims database. The study reported a 46% reduction within 6 mo of treatment with risedronate compared with alendronate. This contrasts to a finding in a large clinical trial that reported that the risks of hip fracture decreased with risedronate only after 12 mo of treatment. In actual practice, one would generally expect lesser effects because of lower treatment compliance.
This study found that patients who recently discontinued bisphosphonate treatment had similar fracture risks as patients who discontinued more distantly and as patients who just started treatment. If this finding is correct, this would suggest that there is little residual effect on fracture risk reduction after stopping bisphosphonates. Pharmacokinetic studies have shown that bisphosphonates remain in bone matrix for many years and remains inactive until the bone containing it is reabsorbed. However, the magnitude and duration of any effect after this release of bisphosphonates is unknown. Unfortunately, there are only few data from randomized trials that addressed this duration of residual effect and no data from observational studies. An extension of the Fracture Intervention Trial found that women who discontinued alendronate after 5 yr had a similar risk of nonvertebral fractures compared with women who continued treatment, although the risk of vertebral increased statistically significantly. In this study, there were very few patients who used bisphosphonates for 5 yr or longer. The duration of any residual effects of bisphosphonates is important. Various cost-effectiveness models have assumed that there was 100% persistence and compliance to treatment and that the effects of bisphosphonates would last for several years after discontinuation. However, the results of these cost-effectiveness analyses may change with different assumptions for treatment persistence and compliance and shorter residual effects.
Three previous studies have also evaluated the association between utilization characteristics and risk of fracture, but there are important methodological differences between these published studies and this study. In the published studies, no distinction was made between made between a patient who discontinued bisphosphonate treatment some time ago but used it regularly (a nonpersistent patient who was previously compliant) and a patient who was currently using it but infrequently (a persistent patient with low compliance). In this study, the effects of persistence (i.e., the duration of treatment) and compliance (i.e., the extent the dosage instruction were followed while being treated) were considered separately. As a consequence of these methodological differences, less than one half of the patients were considered highly compliant in the study by Caro et al. compared with a figure of ∼70% in this study. The relatively good level of compliance compared with the relatively low level of persistence in this study suggests that the emphasis should be on improving the average duration of bisphosphonate treatment rather than improving the compliance to the bisphosphonate regimen.
There are several limitations of this study. This was an observational study, where the patient groups were not randomized. However, it may be difficult to evaluate the effects on fracture risk of persistence and compliance in a randomized trial, because one cannot randomize patients to being compliant or noncompliant. Unobserved confounders may have biased results in this study. To minimize the potential for bias, the analyses were conducted only within a cohort of bisphosphonate users. Another limitation is that complete information on all risk factors for fracture, such as on the BMD, were not available. Also, the results of this may not be generalizable to other populations of bisphosphonate users. About 25% of the female bisphosphonate users in this study were recent users of oral glucocorticoids, similar to a Dutch study, but considerably higher than a U.S. study.
In conclusion, only 58% of the patients remained treated with bisphosphonates for 1 yr, whereas fracture risks reduced only after 6–12 mo of treatment. There was no suggestion of a residual effect after stopping bisphosphonates. There was a 22% lower risk of hip/femur fractures during bisphosphonate current use. Improvement in long-term persistence with bisphosphonate treatment may be important to reduce the impact of osteoporosis-related fractures.
The study was funded by Novartis Pharmaceuticals. The views expressed in this paper are those of the authors and do not reflect the official policy or position of the Medicines and Healthcare products Regulatory Agency, UK.