Treatment for older men with fractures
Less than 10% of men receive osteoporosis treatment, even after a fracture. A study of 17,683 men revealed that older men, those with spinal fractures, and those taking steroids or antidepressants are more likely to receive treatment after a fracture. Seeing a primary care physician also increases osteoporosis treatment rates.
In 2000, the FDA approved bisphosphonates for the treatment of osteoporosis in men. The purpose of this study is to estimate the frequency of bisphosphonate therapy within 12 months following a fracture and describe patient/physician factors associated with treatment.
Health insurance claims for 17,683 men ≥65 years of age, who had a claim for an incident fracture from 2000 to 2005, were followed for at least 6 months post-fracture for the initiation of treatment with a bisphosphonate. Patient characteristics, diagnostic procedures, therapies, co-morbidities, and provider characteristics were compared for men who received treatment with those who did not.
Eight percent of men (n = 1,434) received bisphosphonate therapy. Overall treatment increased from 7% in 2001 to 9% in 2005 (p < 0.001). Treatment for hip fractures remained at 7% (p = 0.747). Treatment increased with age: 6% in men aged 65–69 compared to 11.6% in men aged 85–89 (p < 0.001). Factors associated with treatment included: diagnosis of osteoporosis (OR = 8.8; 95% CI, 7.7, 10.4), glucocorticoid therapy (OR = 3.2; 95% CI, 2.4, 4.3), bone mineral density measurement (OR = 3.4; 95% CI, 2.9, 4.0), and antidepressant therapy with tricyclics (OR = 2.0; 95% CI, 1.2, 3.5) or selective serotonin reuptake inhibitors (OR = 1.7; 95% CI, 1.3, 2.4). Men with vertebral fractures (OR = 2.2; 95% CI, 1.8, 2.6) and men seen by primary physicians (OR = 2.6; 95% CI, 2.3, 3.1) were more likely to receive treatment.
Less than 10% of men received bisphosphonate therapy following a low-impact fracture. Men with a primary physician were more likely to receive bisphosphonate therapy; however, <25% of men were seen by a primary physician.
Osteoporosis and osteoporosis-related fractures in men are becoming increasingly important health issues, especially because osteoporosis-related fractures are increasing at a greater rate in men than in women . This may be explained in part by the fact that the rate of men entering the eighth decade of life has increased more than the rate of women . Furthermore, men are almost twice as likely to die in the hospital or in the first year after a hip fracture than women [3, 4]. Despite effective therapy for osteoporosis in men [5–7] and the nationwide emphasis on prevention and treatment of osteoporosis by federal health agencies , reported treatment rates for men have remained low, even after a serious fracture [9–11]. In this paper, we provide important new information on trends in osteoporosis treatment following FDA approval of bisphosphonates for treatment of osteoporosis in men.
In the 1980s, the FDA approved calcitonin for treatment of vertebral fractures; however, the first bisphosphonate, alendronate, was not approved for treating osteoporosis in men until October 2000 . Kiebzak et al. , reporting on treatment rates between January 1996 and December 2000, found that only 4.5% of men received anti-resorptive or calcium and vitamin D therapy at discharge from the hospital following a hip fracture. Only one of 110 men received a bisphosphonate, and two received calcitonin. Three of 44 men (6.8%) reported treatment with a bisphosphonate within 5 years of the index fracture. The Kiebzak study had only a small sample of men (n = 110) from a single institution and did not report on factors that may have influenced treatment decisions. Subsequently, Feldstein and colleagues , reporting on treatment rates between January 1998 through June 2001, showed that following the approval of alendronate, treatment rates for men with osteoporosis-related fractures remained low. In a sample of 1,171 men from a not-for-profit HMO in the northwestern USA, only 5.4% initiated bisphosphonate and/or calcitonin therapy within 6 months of discharge from the hospital after sustaining a hip fracture.
The primary purpose of this study is to describe trends in osteoporosis treatment for men ≥65 years of age and older after experiencing a low-impact fracture from 2001 to 2005. This time frame is important because it allowed for market penetration of bisphosphonate therapy. We hypothesized that treatment rates would increase with time. Also, in order to better understand why patients were or were not treated following an osteoporosis-related fracture, we studied various patient-, fracture-, and physician-related factors, including medical/surgical specialty that may have influenced treatment rates. This study not only provides important new information on treatment rates following a low-impact fracture in the 5 years following approval of treatment with bisphosphonates for men, but also reports additional information on physician characteristics associated with treatment.
Study design and population
We used health insurance claims data from an employed, commercially insured population with dependents. The data used in this study contained member, provider, hospital, and pharmacy insurance claims for approximately 39 million enrollees throughout the USA from the third quarter of 2000 through to the first quarter of 2007. Individuals 65 years of age and older included men with primary health insurance coverage and those on Medicare who participated in supplementary plans. This clinical electronic database captures close to 100% of all medical care and pharmacy services received by members for which a claim was issued. The records are de-identified and contain claims that are coded with International Classification of Diseases, Ninth Revision (ICD-9) diagnosis and procedure codes, as well as Current Procedural Terminology and Healthcare Common Procedure Coding System procedure codes. This study was approved by the Human Subjects Institutional Review Board of our institution.
The cohort of patients was defined as men 65 years of age and older with an incident fracture consistent with an osteoporosis-related fracture (by convention defined as fractures likely to be caused from a standing height or less) . To be included in the cohort, men had to have at least one claim for a study-defined fracture recorded between the years 2001 and 2005. “Study-defined” fractures were limited to any closed fracture, except of the skull, face, finger, toe, or ankle, that appeared as a billed diagnosis. These types of fractures have been associated with decreased bone mass . We required study subjects to have at least 1 year of continuous enrollment for entry into the study, a minimum of 3 months of enrollment prior to a claim for a fracture of interest based on ICD-9 codes, and a minimum of 6 months enrollment after the date of fracture. Men were excluded from the study if they were treated solely in skilled nursing or rehabilitation facilities for more than 12 months continuously following the index fracture because medication data and the treating physician were not available for them.
The dependent variable was treatment with any oral bisphosphonate or teriparatide. We defined treatment, in response to a low-impact fracture, as the second dispensation (original prescription filling plus at least one refill) of an oral bisphosphonate (etidronate disodium, alendronate sodium, risedronate sodium, ibandronate sodium) or teriparatide dispensed within 1 year of the date of the index fracture. Independent variables included: patient-, fracture-, and physician-related characteristics.
Patient-related variables included age at date of the index fracture, pre-fracture co-morbidities including risks for osteoporosis or falls, and medications prescribed prior to the fracture. Pre-fracture co-morbidities were identified by occurrence of specific ICD-9 codes through the date of the index fracture. We included common co-morbidities such as congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic obstructive lung disease, connective tissue disease (arthropathies), peptic ulcer disease, chronic liver disease, diabetes mellitus, paraplegia, hemiplegia, hypertension, hyperlipidemia, chronic kidney disease, cancer, and AIDS to calculate the Charlson Comorbidity Index  as adapted for use with ICD-9 codes [15, 16]. Patient risk factors for an osteoporotic fracture were also identified via specific diagnoses associated with secondary osteoporosis (hyperthyroidism, hyperparathyroidism, Cushing's syndrome, cirrhosis, chronic renal failure, and testicular hypofunction) and with risk of fracture due to falls (Parkinson's disease, rheumatoid arthritis, osteoarthritis, dementia, alcoholism, and diabetes mellitus type I and stroke). A patient was considered “diagnosed with osteoporosis” if a submitted claim with an associated ICD-9 diagnostic code for osteoporosis, i.e., “733,” was recorded. We did not require a claim for DXA or other radiological exam to have been billed to consider a diagnosis of osteoporosis as valid in this context of a reported fracture.
We examined use of medications that have been associated with the development of osteoporosis as well as medications associated with increased risk of falls and fractures. Anticonvulsant use  was defined as any single prescription dispensed for any anticonvulsant (full list available) prior to the index fracture. Chronic glucocorticoid use was defined as usage in excess of 5 mg of prednisone or its equivalent per day for >90 days prior to the index fractures [18, 19]. Lastly, long-acting benzodiazepines, tertiary tricyclic antidepressants, and selective serotonin reuptake inhibitors were also examined [20–25]. We considered a patient to be a user of these medications at the time of the index fracture if the date the prescription was dispensed plus the days of the medication supplied included the index fracture date.
Fracture-related variables consisted of sites of fracture. Fractures were defined and identified by ICD-9 diagnostic codes and then reclassified as hip, vertebral, and other. The specific diagnostic codes used are listed in Appendix 1.
Physician-related variables were defined by physician specialty. Primary care physicians included family physicians, general internists, and geriatricians; specialty physicians included all other physicians. Visits by advanced practice nurses or physician assistants were excluded as were visits without an identified provider type, i.e., visits reported only by setting (acute care hospital, ambulance, or emergency center). The “prescribing provider” was defined by the specialty of the most recent provider seen by the patient within 30 days prior to the “fill date” of an osteoporosis medication.
To describe the study population, we report descriptive statistics, including means and standard deviations for continuous data and frequencies for nominal and ordinal data, for the cohort, for patient-, fracture-, and provider-related variables. Bivariate comparisons of the characteristics for treated versus untreated men (all of whom had sustained fracture) were used: Student's t test for continuous variables, Fischer's exact test for dichotomous variables, and Pearson's chi-square test for multinomial variables. Unadjusted odds ratios are reported for post-fracture therapy with bisphosphonate or teriparatide. To assess changes in treatment over time, we evaluated the presence or absence of treatment within 12 months of the index fracture for three fracture sites (hip, spine, and any other site) stratified by year (2001–2005) and used the Cochran–Armitage chi-square test for trend [26, 27]. In a sub-analysis of 10,108 men, matched by date of index fracture, we used a multivariable logistic regression model to determine which patient-, fracture-, and provider-related variables were significantly associated with post-fracture treatment with bisphosphonates or teriparatide. Variables were included in the logistic regression model based on the odds ratios observed in bivariate analyses (i.e., OR > 1.0). We matched “treated” patients to those “not treated” by date of fracture. For both groups, the treating or prescribing provider was defined by the specialty of the most recent provider seen by the patient within 30 days prior to the “fill date” of the treatment drugs of interest. For “untreated” patients, the provider of record was defined by the specialty of the provider closest to the “fill date” of the treated individual with whom the untreated subject was matched. Unmatched patients (treated and untreated) were dropped from this phase of the analysis. We matched 1,288 treated individuals with 8,820 non-treated individuals with this approach. One hundred forty six of the 1,434 treated patients (10.2%) were unable to be matched. We used SAS version 9.1.3  to perform all statistical analyses.
Study population of treated and non-treated men after low-impact fracture
Unadjusted OR (95% CI)
Mean agea (standard deviation)
Mean Charlson Comorbidity Indexa (standard deviation)
Charlson comorbidity index distribution
1.7 (1.4, 2.1)
1.5 (1.2, 1.8)
Pre-fracture Rx treatment
20.4 (16.8, 24.7)
Had BMD measurement (anytime)
11.4 (10.1, 12.8)
Osteoporotic fracture risk:
9.9 (8.8, 11.2)
Risk of secondary osteoporosis:
1.3 (1.1, 1.4)
Chronic glucocorticoid use
4.5 (3.7, 5.4)
1.6 (1.4, 1.8)
Any secondary osteoporosis risk (see note)
1.8 (1.6, 2.0)
Risk from falls:
1.1 (1.0, 1.3)
Tertiary tricyclic antidepressant
1.8 (1.2, 2.6)
1.5 (1.2, 1.8)
Selective serotonin reuptake inhibitor
1.6 (1.3, 2.0)
Any risk from falls (see note)
1.2 (1.0, 1.1)
Chronic medical conditions:
Diabetes mellitus II
0.8 (0.7, 0.9)
1.0 (0.9, 1.2)
1.0 (0.8, 1.1)
0.9 (0.8, 1.0)
1.1 (1.0, 1.3)
1.4 (1.2, 1.5)
0.9 (0.8, 1.0)
1.3 (1.2, 1.5)
1.3 (1.1, 1.5)
Bisphosphonates accounted for the vast majority (95%) of the prescription medications prescribed. Overall, 8.1% of men received treatment with a bisphosphonate after experiencing a study-defined fracture (See Appendix 1). Approximately 80.0% of those treated received bisphosphonate therapy for the first time following the index fracture, resulting in an absolute rate of 6.5% for new therapy. Overall fracture treatment with bisphosphonates trended upward during 2001–2005 from 7.0% to 9.0% (p value < 0.0001). Calcitonin accounted for <3% of treated individuals. Evaluation of treatment trends revealed that calcitonin treatment rates declined during 2000–2005 from 4.1% to 1.4%. As measured by the Cochran–Armitage test for trend analysis, calcitonin prescriptions declined significantly as a treatment for hip, vertebral, and other fractures (p value < 0.0001). Teriparatide accounted for 0.24% of prescriptions. Therefore, subsequent analyses focused primarily on bisphosphonates and factors associated with their use.
The average age of men treated following a fracture was 77.4 ± 6.0 years compared to 76.5 ± 6.3 years for men who were not treated. Treatment rates increased significantly with increasing age: 6.0% of men 65–69 years of age, 7.6% of men 70–74 years of age, 8.7% of men 75–79 years of age, 10.1% of men 80–84 years of age, and 11.6% of men 85–89 years of age received treatment following a study-defined fracture (Cochran–Armitage Trend Test, two-sided; p < 0.0001). Factors strongly associated with receiving post-fracture treatment included: pre-fracture treatment, BMD measurement, diagnosis of osteoporosis, and various conditions associated with secondary osteoporosis as listed in Table 1. Co-morbidity affected post-fracture treatment. For each additional point of the Charlson Comorbidity Index, treatment rates increased approximately 3%. Individual medical conditions associated with increased rates of treatment included: COPD, osteoarthritis, and depressive disorders. Use of medications associated specifically with an increased risk for falls was associated with slightly increased rates of treatment: tertiary tricyclic antidepressants, long-acting benzodiazepines, or selective serotonin reuptake inhibitors. Overall, an increased risk of falls was not associated with increased rates of treatment (Table 1).
In the multivariate logistic regression analysis of baseline characteristics of 10,108 patients (1,288 = treated and 8,820 = untreated patients matched by fracture date ± 30 days), osteoporosis diagnosis was strongly associated with bisphosphonate therapy (OR = 8.8; 95% CI, 7.7, 10.4), followed by evidence of bone mineral density test (OR = 3.4, 95% CI, 2.9, 4.0), and history of glucocorticoid therapy (OR = 3.2; 95% CI, 2.4, 4.3). The dispensings of medications known to increase the risk of falls were associated with an increase in treatment, but medical conditions associated with increased fall risk (COPD, depression, neoplasms) did not affect treatment rates (Table 3).
Osteoporosis treatment within 12 months of fracture occurrence
Type of fracture
p value (2-sided)a
Treatment of men after a hip fracture frequently required admission to skilled care units. Men with a hip fracture were five times more likely to have been admitted to a skilled nursing facility compared to men with other types of fractures. Of men admitted to a skilled nursing facility, 5.2% received treatment compared to 7.5% of those who were not admitted to a skilled nursing facility.
Multivariate logistic regression predicting post-fracture osteoporosis drug treatment (n = 10,108)
Adjusted odds ratio
Had BMD measurement
Selective serotonin receptor uptake inhibitors
Chronic Obstructive Lung Disease
Fall Risk by diagnosisb
Any primary care
This study evaluated trends in osteoporosis treatment in men following low-impact fracture allowing ample time for market penetration after approval for bisphosphonates therapy for osteoporosis in US men in 2000. We also report prescriptive practices based on physician specialty. Despite the fact that osteoporosis in men is recognized increasingly as a major health problem, our study found that treatment rates following any low-impact fracture increased by a mere 2% from 2001 to 2005 and that treatment rates for hip fracture, which are associated with the highest mortality and long-term morbidity , did not improve at all. In 2005, 5 years after approval of bisphosphonates for treatment of men with osteoporosis, only 9% of men receive such treatment after a low-impact fracture. The treatment rates observed are far from optimal given that any low-impact fracture foretells a greater risk of a subsequent fracture [29–36]. Another important finding of this study is that when primary care physicians, compared to specialists only, are involved in post-fracture care, men were more than twice as likely to receive treatment for osteoporosis following a fracture; however, treatment rates were still unacceptably low.
Consideration or recognition of osteoporosis appears to be the most significant factor influencing treatment decisions following a low-impact fracture. In our study, we found that treatment of osteoporosis prior to the fracture or diagnosis of osteoporosis or measurement of bone mineral density before or after a fracture was strongly associated with increased odds of treatment ranging from 10- to 20-fold. Juby et al. also reported that in patients undergoing rehabilitation for hip fracture, a diagnosis of osteoporosis influenced rates of treatment. In patients diagnosed with osteoporosis, 48% received treatment compared to only 5.5% in patients without a diagnosis of osteoporosis . Feldstein and colleagues also reported lower treatment rates in patients without a diagnosis of osteoporosis .
We found differences in treatment rates based on site of fracture. The highest treatment rates (17%) were observed in men who sustained vertebral fractures. This finding may reflect influence by the 2000 article by Orwoll et al. that reported a 50% reduction in osteoporosis-related vertebral fractures in men treated with alendronate for 2 years . Higher treatment rates in men with vertebral fractures may also be related to the lack of surgical approaches for these fractures. All fractures are painful, but many are alleviated by stabilization (external via splint/cast or internal via reduction/fixation). Vertebral fractures are not amenable to this type of treatment, so physicians may consider prescriptive therapies sooner when surgical options are not available. It should be noted that most vertebral fractures are not symptomatic and therefore may not be identified in health insurance claims. These asymptomatic men represent a potentially large untreated population who may benefit from treatment to decrease risk of future fractures.
Unfortunately, treatment rates for hip fracture, the fracture with the highest mortality and long-term morbidity , were low (7%) and did not appear to improve from 2001 to 2005. One potential reason for this finding is that the literature published during this time frame regarding bisphosphonate treatment of men after hip fracture is not as clear as that reported for vertebral fracture . In studies published subsequent to this time frame, Sato and colleagues did report significant improvements in relative risk of hip fracture in elderly males treated with risedronate compared to placebo after a stroke  and in men with Parkinson's disease . However, these studies included small numbers of men, who had a very high risk of hip fractures due to increased risk for falls. Another reason that might contribute to observed low rates of treatment following a hip fracture is the dispensings of oral medication in inpatient rehabilitation centers or long-term care nursing homes are not available in the health insurance claims dataset used for this study. While outpatient pharmacy claims are captured in the dataset, inpatient prescriptions (including those given while in extended care facilities) are not. Therefore, data on men who sustained a hip fracture and spent an entire year in various inpatient facilities post-fracture would not include prescriptions (i.e., bisphosphonates) even if they were given. Given the morbidity of hip fractures in men and their consequent need for rehabilitative or long-term care, this limitation in the dataset could have resulted in under-reporting of men treated after hip fracture.
Our findings regarding treatment differences between primary care physicians and specialist physicians differ from those reported by Solomon and colleagues who reviewed published studies from 1992 to 2003. Their study, which included 76 published studies, was composed of similarly aged, mostly women, patients. They reported that patients seen by generalists were less likely than those seen by specialists to receive treatment, but they did not identify whether they considered obstetrician/gynecologists as generalists or specialists . Because our study focused on men, we avoided the classification question associated with women's health providers. Our data suggest that men seen by primary care physicians are more than twice as likely to receive osteoporosis treatment compared to those seen by specialists only.
Osteoporosis is a multidisciplinary problem involving primary care physicians, endocrinologists, and orthopedists in diagnosis and treatment, but no single group is assuming primary responsibility. Orthopedists may hesitate to initiate a long-term medication in patients they are unlikely to follow long term, and primary care physicians and endocrinologists may not be involved in the patient's post-fracture care. However, from the patient's perspective after a fracture, osteoporosis treatment is vitally important because the risk of a second fracture increases greatly if treatment is not initiated [29–36].
Our study had several limitations, which are inherent in analyzing claims data. The database used in this study did not include certain patient-related variables such as weight, BMI, and race/ethnicity. As noted previously, the database does not include complete treatment data for men who spent most of the first year following the index fracture in extended care facilities. Misclassification can be an issue with claims data. For example, ICD-9 diagnostic codes for fractures were not validated, leaving a potential for misclassification of cases. Additionally, patients were considered “diagnosed with osteoporosis” if a submitted claim with an associated ICD-9 diagnostic code for osteoporosis, i.e., “733,” was recorded. We did not require a claim for DXA or other radiological exam to have been billed to consider a diagnosis of osteoporosis as valid in this context of a reported fracture. An “osteoporosis” diagnosis could have resulted from the fracture, from DXA or heel ultrasound, or professional opinion of the billing doctor.
In addition, the database includes information of dispensings received by patients. Therefore, the possibility exists that a patient filled a prescription but did not comply by taking the drug. However, we report the more important clinical observation—filling of more than one prescription—which is a reasonable surrogate for actually taking the medication. Another potential limitation is that we could not assess treatment with calcium and vitamin D supplements because over-the-counter supplements are not included in claims databases. While calcium and vitamin D supplements may increase bone mineral density and may decrease the risk of osteoporotic fracture [40, 41], these supplements alone are not considered adequate treatment for established osteoporosis [42–44]. Finally, linking prescriptions to visits and prescribing provider presented challenges because prescription data are stored separately from diagnostic and visit data. Therefore, we established a rule to assign the type of treating physician, primary care or specialist, based on the type of physician claim associated with the visit immediately preceding the initial fill date for a study medication. This indirect approach may have led to inaccurate attribution of prescribing physician, especially if there were multiple visit dates in close proximity.
This study included treatment data during a 1-year period following an index fracture from settings throughout the USA. Despite approval of bisphosphonates for the treatment of osteoporosis in men in the year 2000, our study found that physicians initiated bisphosphonate therapy in less than 10% of the men who incurred a low-impact fracture in the ensuing 5 years. Recognition of osteoporosis at the time of the fracture and involvement of a primary care physician in post-fracture care were strongly associated with bisphosphonate therapy. Both of these factors are “modifiable.” For example, Kuo and colleagues  reported that discharge planning, with an automatically scheduled visit that provided for a DXA scan and other tests, improved treatment rates. Furthermore, a multidisciplinary approach that involves primary care physicians soon after a fracture may also enhance treatment rates. Osteoporosis is a chronic condition requiring several years of treatment and surveillance; primary care physicians need to take ownership of its management, both for primary prevention of the initial fracture and for secondary prevention after the fracture.
This study was supported by a grant from the AAFP Foundation Joint Grant Awards Program (grant number: G0811).
Conflicts of interest