Osteoporosis International

, Volume 18, Issue 12, pp 1595–1600

Hip fractures in users of first- vs. second-generation bisphosphonates


  • M. Mamdani
    • Keenan Research CentreLi Ka Shing Knowledge Institute of St. Michael’s Hospital
    • Institute for Clinical Evaluative Sciences
    • University of Toronto Faculty of Pharmacy
    • University of Toronto Faculty of Medicine
    • Department of Health Policy, Management and EvaluationUniversity of Toronto
  • A. Kopp
    • Institute for Clinical Evaluative Sciences
    • Institute for Clinical Evaluative Sciences
    • University of Toronto Faculty of Medicine
    • Women’s College Hospital
    • Department of Health Policy, Management and EvaluationUniversity of Toronto
Original Article

DOI: 10.1007/s00198-007-0446-5

Cite this article as:
Mamdani, M., Kopp, A. & Hawker, G. Osteoporos Int (2007) 18: 1595. doi:10.1007/s00198-007-0446-5



This study compared population hip fracture rates for women with a prior fragility fracture who were treated with first-generation versus second-generation bisphosphonate therapies. The observational study found that, relative to women treated with etidronate, a first-generation bisphosphonate, women treated with the second-generation therapies ‘alendronate’ or ‘risedronate’ were equally likely to be admitted to hospital for hip fracture. Our findings must be confirmed in large randomized head-to-head controlled trials.


Few studies have examined hip fracture outcomes among users of first- versus second-generation bisphosphonates. We compared hip fracture rates among elderly women with a history of fracture dispensed first- and second-generation bisphosphonates, hypothesizing that hip fracture rates would be higher among users of first- versus second-generation bisphosphonates after adjusting for confounders.


Administrative data from Ontario, Canada from 01 April 1998 to 31 March 2002 was used to identify population-based bisphosphonate-naïve cohorts of subjects age 66 years and older initiated on first- (etidronate plus calcium; n = 19,127) or second-generation (alendronate or risedronate; n = 1,460) bisphosphonates. Multivariate Cox proportional hazard models were used for analysis.


During over 23,000 person-years of follow-up, we observed 293 hospital admissions for first hip fracture. The unadjusted event rates yielded approximately 12.5 hospital admissions for hip fracture per 1,000 person-years of follow-up in each study group. Relative to the etidronate plus calcium group, females in the alendronate or risedronate group were equally likely to be admitted for hip fracture (adjusted rate ratio [aRR] = 1.0; 95% CI 0.6–1.6).


The findings of this study suggest similar rates of hip fracture between the first- and second-generation bisphosphonates when used continuously among elderly females with a prior history of fracture.


BisphosphonatesHip fracturesObservational cohort studyPostmenopausal osteoporosis


Osteoporosis is a highly prevalent and costly condition among older adults. Postmenopausal women have a 40% lifetime risk of experiencing an osteoporotic fracture [1] and those who experience a fracture are at a 50–100% increased risk of experiencing a subsequent fracture [2, 3]. Moreover, 50% of women who experience a hip fracture do not return to their baseline level of functioning [4], 20% will require long-term care [4], and there is an associated 20% increase in mortality [5]. Unfortunately for many women, the first indication of osteoporosis is when they present with a fracture [6]. Costs associated with osteoporosis care have been estimated to exceed US $1 billion annually in Canada [7] and are expected to reach US $62 billion in the United States by the year 2020 [8].

Bisphosphonate drugs are widely used to treat osteoporosis and prevent fractures. They increase bone mineral density by inhibiting osteoclast-mediated bone resorption and significantly reduce the risk of fractures in postmenopausal women [915]. Three oral bisphosphonates are currently approved for use in Canada for the treatment of osteoporosis: etidronate, alendronate and risedronate. The available evidence suggests that the newer second-generation bisphosphonates alendronate [10] and risedronate [11] are associated with a significant reduction in the risk of hip fracture. The older first-generation bisphosphonate — etidronate — is often used in combination with calcium, although evidence from large well designed randomized controlled trials supporting its effectiveness for the prevention of hip fracture is lacking. Further, there currently exists no comparative data examining the risk of hip fracture among high-risk osteoporotic populations prescribed first- versus second-generation bisphosphonates. This issue is of considerable importance to policy-makers given the widespread use of bisphosphonates and the substantial cost difference between these two groups of drugs — second-generation bisphosphonates are approximately three times the cost of first-generation bisphosphonates [16] — and also to clinicians and patients who seek to provide the best possible care for their patients within the constraints of a healthcare system with limited resources. We therefore conducted an observational study comparing first and second-generation bisphosphonates for the prevention of hip fracture among elderly postmenopausal women with a fracture history.

Materials and methods

Study design and patient population

A retrospective cohort study was conducted by linking administrative healthcare databases covering over 1.4 million individuals 65 years of age and older in Ontario, Canada from 01 April 1993 through 31 March 2003.

Data sources

The administrative healthcare databases in Ontario allowed for cohort identification, comorbidity assessment, and endpoint ascertainment. The linked databases included highly accurate computerized pharmacy records of the Ontario Drug Benefit Program (ODB) [17], which includes records of prescription drugs dispensed to all Ontario residents 65 years of age and older. Hospitalization records were obtained from the Canadian Institute for Heath Information (CIHI) Discharge Abstract Database, which contains a detailed record of all hospital admissions. The Ontario Health Insurance Plan provided physician billing information for inpatient and outpatient services, and the Ontario Registered Persons Database contained basic demographic and vital statistics information, including death date, for each Ontario resident. These databases were linked anonymously using encrypted individual health card numbers.

Cohort construction

Using these databases, we constructed two separate cohorts consisting of new users of first-generation bisphosphonates (i.e., etidronate plus calcium) or second-generation bisphosphonates (i.e. alendronate or risedronate). Patient accrual began April 1st, 1998 and ended March 31st, 2002. Females 66 years of age and older who were dispensed a prescription for etidronate plus calcium, alendronate, or risedronate during the accrual period were included. The index study date was defined as the date of initial study drug prescription during this timeframe. To create a cohort of osteoporosis treatment-naïve subjects within the two study drug groups, we excluded individuals who were dispensed any osteoporosis related drug, except estrogen (i.e., etidronate combinations, alendronate, risedronate, clodronate, pamidronate, calcitonin, raloxifene) in the year preceding the index date, as well as those dispensed study drugs from more than one group on the same day. Over the time period of the study, only etidronate was listed on the Ontario provincial Drug Benefit formulary under general drug benefits. Alendronate and risedronate were listed on the Ontario formulary as ‘limited use’ products from November 2000 and March 2001, respectively. As ‘limited use’ products, these bisphosphonates were covered as a benefit if the prescribing physician indicated on the prescription using a pre-specified code that prior etidronate use had failed (i.e., continued loss of bone mineral density of more than 3% after 2 years of therapy or new fracture after one year of therapy) or was not tolerated by the patient (allergy or side effects). All individuals with a history of hip fracture within 5 years prior to the index date were also excluded, as were those residing in a long-term care facility at the time of inclusion into the study. Further, as conditions that may independently be associated directly or indirectly with hip fracture and therefore confound results, we excluded individuals with a documented history of Paget’s disease, hypercalcemia, epilepsy, trauma related hospitalizations, admissions for pathological fracture, and malignancies of the breast, bone, colon, rectum, or lung, and multiple myeloma or metastatic cancer within 5 years prior to cohort entry. To minimize confounding and create groups of homogenous patients for analysis, the analysis was limited to patients with a documented history of one or more non-hip fracture, not including trauma- or cancer-related fractures, within 5 years prior to the index date. Sub-analyses among patients only with prior wrist or humerus fractures and those with prior BMD testing were also conducted.

This was an on-treatment analysis. For each of the study drug cohorts, we defined the duration of exposure as the period of continuous, exclusive enrolment in the study medication group after the index date. The second-generation bisphosphonate cohort was allowed to switch among alendronate and risedronate during the observation period. The ‘days supply’ variable of the pharmacy claims database allowed us to estimate the intended duration of each prescription. If subjects were dispensed a drug prior to the end of this period, the excess drug supply was carried over to the next prescription’s day supply estimation. Subjects were allowed a 50% grace period on the previous day supply to refill the next prescription. If they did not refill their prescription for the study drug within these successive time windows, they were deemed to have discontinued the study drug.


The primary outcome of the study was admission to hospital for hip fracture. Admission to hospital for non-trauma-related hip fracture was defined using admission with a primary diagnosis of hip fracture. Follow-up for each subject was censored upon admission to hospital for hip fracture, exposure to a medication from another study group, discontinuation of the study medication, death, or the end of the observation period (31 March 2003). A maximum follow-up of 2 years was allowed for each patient.

Statistical analysis

Time-to-event analyses were conducted for hospitalization for hip fracture using Cox proportional hazards models with the etidronate plus calcium group as the reference, controlling for all covariates outlined in Table 1. Patient entry year and quarter was also included as a covariate in the multivariate analysis. As an overall measure of comorbidity, we controlled for the number of distinct drugs dispensed in the year prior to the index date, a measure comparable to the Charlson comorbidity index [18]. All analyses were performed using SAS for UNIX, Version 8·2 (SAS Institute, Cary, NC). All statistical tests were performed at the 5% level of significance and were two-sided.
Table 1

Potential confounders included in the analysis

Documented diagnoses (within 5 years prior to index date)

Drug utilization (within 1 year prior to index date)

Conditions with either a documented diagnosis or drug utilization


Prior fracture history:


Asthma / COPD / related drug use



Oral corticosteroids

Dementia / use of acetylcholinesterase inhibitors

Socioeconomic status (via neighbourhood income quintile)


Sedative hypnotics

Parkinson’s disease / anti-Parkinson’s drugs

Bone mineral density testing in past 2 years

 Clavicle, scapula


Thyroid disorders / related drugs

Dialysis in past 2 years



Rheumatoid arthritis / related drugs

Number of distinct drug products dispensed in past 1 year

 Wrist, hand


Diabetes (ODD) / related drugs

Year and quarter of index date

 Ankle, foot

Cardiovascular drugs:



Fall-related hospitalizations




Alcohol abuse


Chronic renal failure

 Beta blockers


 Thiazide diuretics

Visual impairment

 Loop diuretics

Cushing’s syndrome

 K-sparing diuretics

 ACE inhibitors / ARBs



COPD = chronic obstructive pulmonary disease; ODD = Ontario Diabetes Database


We identified 19,127 new female users of etidronate plus calcium and 1,460 new users alendronate or risedronate age 66 years and older meeting our inclusion criteria (Table 2). Females in the alendronate or risedronate group were observed to be generally similar to their etidronate + calcium-using counterparts with a few notable exceptions. Females in the alendronate or risedronate group appeared to be of higher socioeconomic status and were more likely to have had a BMD test conducted in the past 2 years relative to females in the etidronate + calcium group. Overall, the females in the alendronate or risedronate group may be slightly healthier than those in the etidronate + calcium group as evidenced by the comorbidity index used in this analysis.
Table 2

Baseline characteristics

Sample size, n

Study group

Alendronate or rsedronate

Etidronate plus calcium



Age, mean ± SD

75.5 ± 6.4

76.0 ± 6.6

Socioeconomic status: neighbourhood income quintile

1 (lowest)

241 (17%)

4,323 (23%)


268 (18%)

4,171 (22%)


264 (18%)

3,614 (19%)


266 (18%)

3,015 (16%)

5 (highest)

352 (24%)

3,216 (17%)


69 (5%)

788 (4%)

Number of drugs in past 1 year, mean ± SD

9.7 ± 6.3

10.3 ± 6.8

BMD testing in past 2 years

1,286 (88%)

13,587 (71%)


2 (<1%)

17 (<1%)

Fracture history

Vertebral fracture

186 (13%)

1,926 (10%)

Pelvis fracture

98 (7%)

1,080 (6%)

Clavicle/Scapula fracture

28 (2%)

408 (2%)


197 (14%)

2,563 (13%)

Wrist/Hand fracture

590 (40%)

8,516 (45%)

Ankle/Foot fracture

149 (10%)

2,123 (11%)

‘Other’ fracture

703 (48%)

8,710 (46%)

Conditions characterized using diagnostics data

Chronic renal failure

156 (11%)

2,213 (12%)

Fall-related hospitalizations

281 (19%)

3,860 (20%)


26 (2%)

369 (2%)

Alcohol abuse

16 (1%)

333 (2%)

Stroke / TIA

178 (12%)

2,304 (12%)

Visual impairment

30 (2%)

423 (2%)

Cushing’s syndrome

6 (<1%)

64 (<1%)

Conditions characterized uing diagnostics or drug utilization data

Asthma / COPD

377 (26%)

4,959 (26%)


106 (7%)

1,425 (8%)

Parkinsons disease

50 (3%)

561 (3%)

Thyroid disorders

331 (23%)

3,823 (20%)

Rheumatoid arthritis

178 (12%)

2,196 (12%)


136 (9%)

2,671 (14%)

Drug utilization

Estrogen related therapy

293 (20%)

3,149 (17%)

Oral corticosteroids

138 (10%)

1,859 (10%)

Sedative hypnotics

454 (31%)

6,226 (33%)


281 (19%)

3,872 (20%)


41 (3%)

479 (3%)


519 (36%)

6,684 (35%)

Cardiovascular drugs:

ACE inhibitors / ARBs

399 (27%)

4,899 (26%)


95 (7%)

1,290 (7%)


28 (2%)

232 (1%)

Beta blockers

230 (16%)

3,308 (17%)

Calcium channel blockers

304 (21%)

4,518 (24%)

K-Sparing diuretics

110 (8%)

1,745 (9%)

Loop diuretics

138 (10%)

2,242 (12%)


201 (14%)

2,698 (14%)


155 (11%)

2,458 (13%)

‘Other’ antihypertensives

12 (<1%)

146 (<1%)

During over 23,000 person-years of follow-up, we observed 293 hospital admissions for hip fracture (Table 3). The unadjusted event rates yielded approximately 12.5 hospital admissions for hip fracture per 1,000 person-years of follow-up in each study group. Relative to the etidronate plus calcium group, females in the alendronate or risedronate group were equally likely to be admitted to hospital for hip fracture (adjusted rate ratio [aRR] = 1.0; 95% CI 0.6–1.6). These relative risk estimates were adjusted for all covariates outlined in Table 1, including diabetes, estrogen use, and renal disease. Sub-analyses that included only patients with prior wrist or humerus fractures and those with prior BMD testing revealed similar findings (data not shown).
Table 3

Primary outcome analysis


Alendronate or risedronate

Etidronate + calcium

N = 1,460

N = 19,127

Hip fractures (n)



Mean ± SD days of follow-up

380 ± 256

418 ± 253

Total follow-up period (patient-yrs)



Incidence rate per 1,000 person-years



Unadjusted HR (95% CI)

1.0 (0.6–1.6)

1.0 (reference)

Adjusted HR (95% CI)

1.0 (0.6–1.6)

1.0 (reference)


The findings of this study suggest similar rates of hip fracture between the first- and second-generation bisphosphonates when used continuously among elderly females with a prior history of fracture. To the best of our knowledge, this is the largest study to date comparing first- and second-generation bisphosphonates in this population.

Several limitations of this study deserve mention. First, the major source of bias in this study likely relates to selection bias. It may be possible that individuals with more severe osteoporosis may have been first-generation bisphosphonate users in the distant past (i.e., greater than one year prior to cohort entry) and then were subsequently initiated on second-generation bisphosphonates given the relatively more restrictive policies for prescribing second-generation bisphosphonates. It is possible that this population represents a relatively more fragile population in ways that were not measured by this study. However, potential known confounding factors were generally well distributed between groups and the unadjusted rate ratio was virtually identical to the adjusted rate ratio, implying little effect of potential confounding due to the variables outlined in the analysis. Further, it has been well documented that physicians may often ‘game the system’ to prescribe their drug of choice [19]. Although we attempted to control for many important confounders, we were unable to account for some potentially important factors such as bone mineral density, over-the-counter calcium or vitamin D use, body mass index, diet, and exercise. The distribution of such factors among the different groups studied and the consequent influences on our findings are unknown. Second, we used administrative databases to identify and define exposure to study drugs and clinical outcomes. We have no direct measure of adherence or appropriateness of use. Third, we were unable to examine different bisphosphonate formulations. Recent evidence suggests that once-weekly formulations may be associated with significantly greater adherence rates than once-daily formulations [20], which in turn may be associated with better patient outcomes [21, 22]. Fourth, the low absolute number of events in the study groups precluded reliable subgroup analyses, such as comparisons among individuals with chronic renal failure. However, we had at least 90% power to detect a 10% relative difference between groups in hip fracture rates between treatment groups. Fifth, the generalizability of our findings to younger patients or settings with different drug policies over longer durations of follow-up is uncertain.

The findings of this study support a class effect between first- and second-generation bisphosphonates for the prevention of hip fracture among elderly females with a prior history of fracture. However, given the aforementioned limitations of this observational study, the findings support but do not prove the hypothesis that second-generation bisphosphonates do not have significantly greater effectiveness than first-generation bisphosphonates. As such, large randomized head-to-head controlled trials are urgently needed to better examine this important clinical question.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2007