Effectiveness of risedronate and alendronate on nonvertebral fractures: an observational study through 2 years of therapy
- First Online:
- Cite this article as:
- Lindsay, R., Watts, N.B., Lange, J.L. et al. Osteoporos Int (2013) 24: 2345. doi:10.1007/s00198-013-2332-7
- 643 Views
This observational study showed that after 2 years, both risedronate and alendronate lowered the risk of hip and nonvertebral fractures compared with patients filling in a single bisphosphonate prescription.
Post hoc analyses of the placebo-controlled trials suggested earlier effects for risedronate (6–12 months) than for alendronate (18–24 months). The present study extends our 1-year observational data that confirmed an earlier fracture reduction with risedronate and evaluated the absolute and relative effectiveness of alendronate and risedronate in clinical practice over 2 years.
We observed three cohorts of women aged 65 years and older who initiated once-a-week dosing of bisphosphonate therapy; (1) patients adherent to alendronate (n = 21,615), (2) patients adherent to risedronate (n = 12,215), or (3) patients filling only a single bisphosphonate prescription (n = 5,390) as a referent population. Proportional hazard modeling compared the incidence of hip and nonvertebral fractures among the cohorts over 2 years after the initial prescription.
In this cohort, we previously showed at 12 months a significant reduction of hip and nonvertebral fractures with risedronate but not with alendronate. At the end of 2 years, the cumulative incidence of hip fractures in the referent cohort was 1.9 %, and incidence of nonvertebral fractures was 6.3 %. Relative to the referent, 6 months after initiating therapy and continuing through 2 years, both risedronate and alendronate cohorts had approximately a 45 % lower incidence of hip fractures and a 30 % lower incidence of nonvertebral fractures.
These observations suggest that both risedronate and alendronate are effective at reducing the risk of hip and nonvertebral fracture after 2 years of treatment and support the post hoc analyses of placebo-controlled trials indicating an earlier effect of risedronate.
International Classification of Diseases, 9th Revision, Clinical Modification
More than two million fractures related to osteoporosis occur annually in the US . To reduce the risk of such fractures, numerous drug therapies have shown in randomized controlled trials to reduce the incidence of morphometric vertebral fractures, a necessary criterion for regulatory approval . Osteoporotic fractures occurring at skeletal sites other than the vertebrae (i.e., nonvertebral fractures, which are often reported as a composite of sites including the hip, wrist, humerus, and others) represent three fourths of fractures coming to clinical attention . In the prospectively planned analyses of randomized controlled trials, the incidence of nonvertebral fractures has been lower with treatment than with placebo for alendronate [4–6], risedronate [7–9], estrogen , zoledronic acid , and denosumab .
The efficacy of a therapy demonstrated in a randomized controlled trial may not predict its actual effectiveness in clinical practice because of differences in the characteristics of patients and in the level of medical care . Consequently, there is a need to conduct observational studies to assess the effectiveness of therapies in actual clinical practice. In addition, when more than one therapy option is available, observational studies can also provide information in an assessment of comparative effectiveness. Two of the most commonly utilized treatments for osteoporosis are the once-a-week dosing regimens of alendronate and of risedronate . Since both of these dosing regimens have been available in the US since 2002, there is now an opportunity to evaluate their effectiveness for reducing nonvertebral fractures in large numbers of patients in actual clinical practice.
In this observational study, we sought to determine if the effectiveness of alendronate and risedronate in clinical practice has been consistent with efficacies demonstrated in clinical trials. We have previously reported the incidences of fractures among women aged 65 years and older who initiated bisphosphonate during the 2002 to 2004 period . In that report, based upon records of administrative billing through 2004, we observed during the initial year of therapy that patients using risedronate had a lower incidence of clinical fractures than patients using alendronate. To obtain more information on relative effectiveness, we have extended the study with now available data through 2 years of therapy. In addition, we included a group of patients filling only a single bisphosphonate prescription to serve as a comparison cohort to estimate the effectiveness of bisphosphonate therapy relative to no therapy.
Materials and methods
Computerized records of administrative billing provide a convenient data source for studying drug use and outcomes in large populations. Such databases provide patient-level data of: (a) inpatient and outpatient services specified by diagnostic codes of the ICD-9-CM, (b) retail and mail order pharmacy dispensations specified by national drug codes, and (c) demographic information including sex, age, and dates of health plan coverage in data source. The data for the current study, inclusive of January 2000 to June 2006, originated from two sources: Ingenix Lab/Rx (Eden Prairie, MN) and Medstat MarketScan (Ann Arbor, MI). In 2005, the number of women aged 65 years and older in these mutually exclusive data sources were 1.4 million in Medstat and 0.4 million in Ingenix. Geographically, one half of this population was located in Michigan, California, Florida, Ohio, Georgia or Texas, and one half in the other 44 states.
Three cohorts of women aged 65 years and older, included those who: (a) initiated and continued use of once-a-week dosing of alendronate 35 (8 %) or 70 mg (92 %) (alendronate cohort), (b) initiated and continued use of once-a-week dosing of risedronate 35 mg (risedronate cohort), and (c) initiated and filled only a single 30-day prescription of either once-a-week dosing of alendronate or risedronate (referent cohort). The inclusion criteria into the alendronate and risedronate cohorts were described previously in our earlier report . Briefly, subjects were required to have filled a bisphosphonate prescription between July 2002 and June 2004, a period chosen to allow contemporaneous availability for both bisphosphonate types, after at least 6 months of prior eligibility in the data source with no other bisphosphonate use. In addition, the first bisphosphonate prescription (i.e., the index prescription) for each subject had to be refilled to provide at least 3 months of therapy. The data management and statistical analyses of these cohorts were independently replicated in the original report by Michael B. Nichol, PhD, of the department of clinical pharmacy and pharmaceutical economics and policy at the University of Southern California.
A referent cohort was added in this study to represent the fracture incidence of a comparable population not on therapy. This cohort met the same inclusion criterion (i.e., index bisphosphonate during the study period) but included subjects filling only a single 30-day prescription within 1 year after index prescription of either alendronate or risedronate. Subjects starting either bisphosphonate were combined for the referent group based upon the observation that both groups had similar fracture incidence during follow-up after index prescription.
In our earlier report , medical billing data were available through September 2004, and thus some subjects had only 3 months of follow-up (e.g., subjects with an index bisphosphonate in June 2004). In this current study, medical claims were available through June 2006, thus allowing all three cohorts to be followed for up to 2 years after the index prescription. Follow-up for each subject in either the alendronate or risedronate cohorts was censored by the earliest occurrence of: date of first fracture, end date of data source eligibility, or end date of therapy adherence. Adherence was measured as a function of the gaps between refills; once the gap between the completion of a 30-day supply and subsequent refill exceeded 15 days, the end date of therapy adherence was the prescription date before gap plus 45 days. Follow-up for each subject in the referent cohort was censored by date of first fracture, end date of data source eligibility, or date of a second bisphosphonate prescription. All products used were branded.
Each subject was followed to identify two outcomes: a new hip fracture, inclusive of an inpatient fracture diagnosis at the hip (ICD-9-CM code 820, 733.14), and a new nonvertebral fracture, inclusive of an inpatient fracture diagnosis at the hip and inpatient or outpatient fracture diagnosis at the wrist (813, 733.12), humerus (812, 733.11), clavicle (810), pelvis (808), and leg (821, 823, 733.15, 733.16). New fractures were defined as a fracture at each body site for which there was no fracture at that same site in 6 months before index prescription. Likely traumatic fractures were excluded by eliminating diagnoses of an open fracture or a cause of injury other than an accidental fall (Ecode of E880–E888), an assumption excluding less than 10 % of fracture outcomes. During follow-up, subjects were allowed to have each outcome only once.
Cox proportional hazard modeling (PROC PHREG, SAS Institute, Cary, NC) was used to compare the fracture incidence among cohorts, adjusting for baseline differences in risk factors for fracture. Available risk factors in the data source included age, history of prior fracture, glucocorticoid use, and diagnosis of rheumatoid arthritis. Age was calculated at the year of cohort entry. History of prior fracture was defined by any clinical fracture diagnosis at the hip, wrist, humerus, clavicle, pelvis, leg, or vertebrae in the 6 months prior to cohort entry. Glucocorticoid use was defined by receiving 450 mg prednisone-equivalent pills within ± 90 days of cohort entry—an approximation of the American College of Rheumatology guideline of 5 mg prednisone for at least 90 days . A diagnosis of rheumatoid arthritis was based on any inpatient or outpatient diagnosis (ICD-9-CM code 714.0) within 6 months prior and 3 months postcohort entry. Risk factors not available in the data source included bone mineral density, body mass index, smoking, alcohol consumption, and family history of fracture.
Baseline characteristics of fracture risk in study population
Age, mean, years
≥75 years of age (% cohort)
Glucocorticoid use (% cohort)a
Rheumatoid arthritis (% cohort)b
Clinical fracture in prior: (% cohort)c
Bisphosphonate use in past: (% cohort)d
Hormone replacement therapy use in past: (% cohort)e
There were no differences in the number of days hospitalized for fractures between the groups as all hospitalizations were confirmed having at least 24 h between admission and discharge.
Time to event analysis for hip fractures at 2 years of bisphosphonate therapy
2 years on therapy
Fracture incidence (%) numeratora
Fracture incidence (%) denominatora
Crude rate ratiob
Adjusted rate ratioc
95 % confidence interval
Time to event analysis for nonvertebral fractures at 2 years of bisphosphonate therapy
2 years on therapy
Fracture incidence (%) numeratora
Fracture incidence (%) denominatora
Crude rate ratiob
Adjusted rate ratioc
95 % confidence interval
In this observational study, patients observed through 2 years on once-a-week alendronate or risedronate had a lower incidence of hip fractures and nonvertebral fractures than patients filling only a single bisphosphonate prescription (i.e., referent).
The annualized incidence of hip fractures in the referent cohort was 1.0 %, and the relative risk of hip fractures in risedronate or alendronate cohorts ranged from 0.62 to 0.73 at 2 years. In the randomized controlled trials that reported on hip fractures and bisphosphonates including alendronate, risedronate, or zoledronic acid, the annualized incidence of hip fractures in the placebo ranged from 0.7 to 1.1 %, and the relative risk of fracture in the treated ranged from 0.49 to 0.60 % at 3 years [4, 9, 11]. These studies suggest that the effectiveness of alendronate and risedronate in clinical practice has been consistent with efficacies reported in randomized controlled trials.
Recent observational studies have also provided an estimate of alendronate and risedronate effectiveness. The most common approach has been a cohort study design that includes a comparison in fracture rates between patients with a high level of bisphosphonate use (i.e., adherence, compliance) versus patients with a low level of use. A benefit of this design is an expectation of similar characteristics between cohorts, since the entire study population is placed on therapy, thereby potentially limiting confounding by indication . A limitation of this design is the potential for differences in outcomes between cohorts being due solely to the healthy behavior of adhering to therapy . Like the current study, all 15 recent cohort studies have suggested that patients with a high level of alendronate and risedronate use have fewer fractures than patients with a low level of use over a time period, generally of two or more years .
In addition to the current study, studies by Cadarette et al. , Curtis et al. , and Harris et al.  have examined the relative effectiveness among different forms of bisphosphonates. The four observational studies have a similar cohort design that includes a comparison in fracture rates between patients initiating one form of bisphosphonate versus another form. While similar in design, differences in age of study population, the approach to follow-up, and duration of follow-up are noteworthy in consideration of results.
The age of the study population should relate to the statistical power of a study. An older study population, relative to a younger population, will likely have more fractures, and these fractures will more likely be related to osteoporosis . Among the four studies, the mean age of study populations on bisphosphonates ranged from 60 years in Harris et al.  to 78 years in Cadarette et al. .
The approach to follow-up may attenuate differences in fractures between cohorts. Of patients initiating bisphosphonate therapy, 20 to 30 % never fill a second prescription [24, 25]. There is no evidence that these patients would receive a fracture reduction benefit. In addition, it has been established repeatedly that a low level of bisphosphonate adherence corresponds to little fracture reduction benefit . The approach to follow-up for both the current study and Harris et al.  included only subjects on therapy for at least 3 months and continued follow-up as long as adherent to therapy. The approach for Cadarette et al.  included subjects filling a single prescription and follow-up was independent of therapy adherence. The approach for Curtis et al.  included subjects filling a single prescription and continued follow-up as long as subjects were adherent to therapy.
The duration of follow-up may attenuate differences in fracture rates between cohorts. The randomized controlled trials were generally 3 years in duration. Fracture reductions have been noted in the range of 6 to 24 months after the start of therapy within post hoc pooled analysis of clinical trials [26, 27]. The percentage of subjects followed through 1 year was approximately 50 % in the current study, 80 % in the study of Cadarette et al. , 22 % in Curtis et al. , and a mean of 7 months in Harris et al. . The duration of follow-up also influences the outcomes from the current study. In our previous report following 1 year of therapy , patients receiving risedronate had lower rates of hip and nonvertebral fractures compared with patients receiving alendronate. The rates of hip fractures were 0.4 vs. 0.6 % for risedronate and alendronate (95 % CI, 13–63), while the rates of nonvertebral fractures were 2.0 vs. 2.3 % (95 % CI, 2–32), respectively . The findings of the 1-year study (i.e., that patients using risedronate had a lower incidence of clinical fractures than patients using alendronate)  were supported in the current analysis by the observation of the early separation and lower rate of both hip and nonvertebral fracture with risedronate compared with alendronate during the first 6 to 12 months of therapy. In the current analysis after 2 years, the rates of nonvertebral and hip fractures were similar between the risedronate and alendronate groups (approximately 30 % reduction in nonvertebral fractures and 45 % reduction in hip fractures in both groups). This suggests that the therapeutic effect of risedronate on fracture risk starts earlier than that of alendronate. Considering that up to 80 % of patients do not persist with their bisphosphonate prescription beyond 2 years  and the fact that effectiveness of therapy is strongly dependent on adherence , the early effect of risedronate on fracture risk may be an advantage for those patients who do not persist with their prescribed therapy.
The significance of absolute effectiveness of alendronate and risedronate in clinical practice is magnified by the approximately five million patients currently on these bisphosphonates in the US . If one half of these patients were adherent to therapy for 2 years, then approximately 1 and 2 % absolute differences between therapy and no therapy in hip fractures and in nonvertebral fractures, respectively, should translate to 25,000 and 50,000 fewer hip and other nonvertebral fractures over 2 years. This lower fracture rate among subjects treated may explain in part the declining rate of fractures occurring in the general population [29–32]. With the direct cost of hospitalization for hip fracture of approximately $30,000 , the financial significance of 25,000 hip fractures is $750 million.
A limitation of this study is that the baseline differences in the groups that were followed and those that were lost to follow-up could not be examined because the data was unavailable. This limitation is one of the many limitations for observational studies in general. Another limitation to this study is the difficulty to see whether the cohorts are balanced, which is a general limitation for any observational studies. The cohorts in this study seem similar on measured variables. However, this is not known for unmeasured variables, such as equal vitamin D and calcium intake. Violation of the proportional hazards assumption may lead to confidence intervals around point estimates that are overly narrow. Hence, there may be more inherent variability in results than currently reflected by the statistics presented.
In conclusion, these observations suggest that the clinical effectiveness of risedronate and alendronate after 2 years of treatment are similar and consistent with efficacies reported in randomized controlled trials. That the effects of risedronate were detected earlier than the effects of alendronate may be an advantage for those patients who do not persist with their prescribed therapy. Hence by extension, a significant number of osteoporotic fractures are being prevented among the approximately five million patients in the US on either of these bisphosphonate therapies.
Sources of funding
The data source for this article was purchased by Warner Chilcott (US), LLC, and Sanofi.
Conflicts of interest
Dr. R. Lindsay has served as a consultant for Amgen, Eli Lilly, Procter & Gamble, NPS, Sanofi, Roche-GlaxoSmithKline, Novartis, and Wyeth. Dr. P.D. Delmas (deceased) served as a consultant and speaker for Procter & Gamble and Sanofi. Dr N.B. Watts has served as a consultant and speaker for Amgen, Novartis, and Warner Chilcott; he has served as a consultant for Amgen, Arena, Baxter, Eli Lilly, InteKrin, Johnson & Johnson, Medpace, Merck, NPS, Orexigen, Pfizer/Wyeth, Sanofi, Takeda, and Vivus; he has received research grants from Amgen, Merck, and NPS. Dr J.L. Lange is an employee of Procter & Gamble. Dr S.L. Silverman has served a consultant and speaker for Amgen, Eli Lilly, Roche, and Procter & Gamble; he has served as a consultant for Wyeth; he has received research grants from Eli Lilly and Procter & Gamble.