Differences in evidence-based osteoporosis medication use between elderly men and women enrolled in Medicare Part D

Abstract

Objective

To compare osteoporosis treatment utilization rate between elderly men and women.

Methods

Cross-sectional and panel analyses of Medicare beneficiaries (1) aged 70 years and older, (2) enrolled in Part A, B and Part D stand-alone prescription drug plans from 1 January 2006 through 31 December 2008, or death, and (3) diagnosed of osteoporosis drawn from a random 5% sample of the Medicare population. Prescription drugs used to treat osteoporosis included bisphosphonates, calcitonin, parathyroid hormone analog and selective estrogen receptor modulator. Covariates included demographic characteristics, conditions and medications contributing to osteoporosis, bone fracture, or falls, contraindication and side effects of osteoporosis medications captured from Medicare claims.

Key findings

The study sample comprised 8465 men and 90 956 women with osteoporosis. Prevalence of osteoporosis medication use in men was substantially less than for women (25.2% versus 44.3% in 2006). With advancing age, men were more likely to be treated than women. Black men and women were much less likely (RR 0.76, 95% CI (0.68, 0.85) for men; RR 0.61, 95% CI (0.59, 0.63) for women) to be treated for osteoporosis compared to white men and women. Men who received BMD testing were much more likely to be treated compared to women. Bisphosphonates were the treatment of choice for both men and women, with nonbisphosphonate use increasing with advancing age.

Conclusion

Elderly men are significantly undertreated for osteoporosis compared to elderly women, and this is particularly problematic among blacks.

gender difference in osteoporosis treatment, Medicare Part D, osteoporosis medications, pharmaceutical HSR

Introduction

Among all patients diagnosed with osteoporosis, approximately 20% are men and 80% are women.^[1] Men are not only less likely to be diagnosed with osteoporosis than women, and men are also treated less often than women even when diagnosed with osteoporosis. The prevalence of osteoporosis medication use ranges from 7.1 to 27% in men compared to 4.6 to 50% in women^[2–4] despite the fact that osteoporosis treatment guidelines recommend initiation of pharmacologic therapy in both men and women who have bone mineral density (BMD) T-scores ≤−2.5 at the femoral neck or spine by dual-energy x-ray absorptiometry (DXA) and those who have experienced fragility fractures.^[5,6] The lower medication use rate in men may be owing to the lack of awareness of osteoporosis and its consequences in men.^[7] Osteoporosis is an asymptomatic disease and has long been seen as a women's disease; therefore, men may not be aware of osteoporosis and its related bone fractures and thus do not pursue care. Even if men pursue care, their physicians may lack awareness of osteoporosis in men and thus do not properly screen and diagnose the disease. Underdiagnosis of osteoporosis is an issue in men. Even when diagnosed with osteoporosis, there are fewer evidence-based medications available for men. Among all evidence-based osteoporosis medications, only three classes are indicated for men while four are indicated for women.

Men diagnosed with osteoporosis have a higher risk of bone fracture^[8] compared to women and suffer more severe outcomes after having a bone fracture.^[9,10] Twenty-eight per cent of hip fractures in the United States occur in men,^[8] and mortality after hip fractures is doubled in men as compared to in women.^[9,10] Therefore, osteoporosis care in men requires improvement.

Besides disparities in osteoporosis care between men and women, there is also a gap in research addressing osteoporosis management in men and women. The vast majority of osteoporosis research to date has been focused on women. Few studies have been conducted in men, and fewer still on males in minority populations. The 2004 Surgeon General's Report on Bone Health and Osteoporosis indicated that the prevalence of osteoporosis among Hispanic women was similar to that in Caucasians^[11]; however, no comparisons were made among men.

In this study, we examined how use of evidence-based osteoporosis medications differed by sex in the Medicare Part D enrollees from 2006 through 2008. Based on the few comparative research studies on osteoporosis,^[12,13] we hypothesized that men would be less likely to receive any osteoporosis treatment, and among men receiving osteoporosis treatment, more would receive bisphosphonates. The study results will provide baseline information of clinical differences in osteoporosis treatment use in men and women in the early years of the Medicare Part D drug benefit and to identify challenges and future direction for treatment of osteoporosis, especially in the male Medicare population.

Methods

Data and sample selection

Medicare enrolment and claims data for the study were obtained from 2006 to 2008 Chronic Condition Data Warehouse (CCW) for a 5% national random sample of beneficiaries.^[14] The study sample included Medicare beneficiaries aged 70 years and older with evidence of osteoporosis prior to 1 January 2006. Setting the minimum age at 70 enabled the construction of a 5-year look-back window to check beneficiaries’ history of osteoporosis. Evidence of osteoporosis was defined using the CCW algorithm of at least one inpatient, outpatient or physician claim with ICD-9 diagnosis codes 733.00, 733.01, 733.02, 733.03, 733.09. The study sample was also restricted to beneficiaries with continuous enrolment in Medicare Part A, B and a Part D stand-alone prescription drug plan (PDP) from 1 January 2006 through 31 December 2008, or death. The sample excluded (1) Medicare Advantage Prescription Drug (MAPD) plan enrollees from 2006 to 2008, (2) beneficiaries not enrolled in Part D from 2006 to 2008 and (3) long-term care (LTC) facility residents whose admission or discharge date could not be identified in the Minimum Data Set (MDS) together with inpatient and skilled nursing facility (SNF) claims. MAPD plan enrollees were excluded because they lacked Part A and B claims files necessary for identifying diagnosis and utilization of Medicare services. Non-Part D enrollees were excluded because they lacked prescription drug data. LTC residents with missing admission or discharge dates were excluded because they lacked data necessary for sample classification by residence.

Drug measures

Outcomes of interest were dichotomized variables indicating use of any osteoporosis medications among individuals diagnosed with osteoporosis and use of nonbisphosphonates among osteoporosis medication users. Classes of osteoporosis medications listed in the 2008 National Osteoporosis Foundation (NOF) guidelines included bisphosphonates, calcitonin, parathyroid hormone analog and selective estrogen receptor modulator (SERM). Use of these medications was identified by National Drug Codes (NDCs) in the Part D prescription drug event (PDE) file using the FirstDataBank drug dictionary and by the Healthcare Common Procedure Coding System (HCPCS) codes in the Part B claims file for Part B covered osteoporosis drugs. The 2008 NOF guidelines were used because they were the most relevant guidelines for the study period. It should be noted that estrogen is sometimes used as treatment of osteoporosis; however, the Food and Drug Administration (FDA)-approved indication is only for prevention of osteoporosis, so use of estrogens was not included in this study. In preliminary analyses, we found fewer than 1% of study subjects used Part B covered ibandronate sodium and zoledronic acid injections during the study period, so use of those two Part B covered drugs was not included in the final analysis. The unit of observation for osteoporosis medication use was person-year. One or more fills in a given year was considered as having any use in that specific year. Osteoporosis drugs used during Medicare-covered hospitalizations and SNF stays could not be considered as they are covered under Part A and are thus not observable in the PDE file.

Other measures

The primary independent variable is sex. A conceptual framework for treatment of osteoporosis originally developed by Krishnan et al.^[15] for tuberculosis treatment was adapted to guide inclusion of other measures. These measures included demographics (age, race/ethnicity and geographic region), residential status (categorized as community only, LTC only, and both using the MDS together with inpatient and SNF claims). Socioeconomic status was captured via indicators of Medicare-Medicaid dual eligibility and receipt of Part D low income subsidy (LIS). General health status was represented by counts of conditions from the prescription drug hierarchical condition categories (RxHCC) risk adjustment measure developed by the Center for Medicare and Medicaid Services (CMS). Indicators for conditions and medications that may cause or contribute to osteoporosis and bone fractures and risk factors for fall were selected based on the 2008 NOF guidelines.^[16] Disease factors included prior hip fracture, diabetes mellitus, rheumatoid arthritis/osteoarthritis, congestive heart failure (CHF), depression, end-stage kidney disease (ESKD), chronic obstructive lung disease (COPD) and breast cancer, all defined by CCW diagnostic algorithms.^[17] Commonly used medications that may cause or contribute to osteoporosis and bone fractures included anticonvulsants, cancer chemotherapy and glucocorticoids. Risk factors for fall included Alzheimer's disease/dementia, glaucoma (as proxy for poor vision) and stroke (as proxy for poor balance). FDA-approved package inserts were used to identify possible contraindications/medication side effects that may influence osteoporosis medication choices (list provided in Table 1). The most common contraindication for osteoporosis medications is chronic kidney disease. The main side effect is GI problems. Other contraindications and side effects from the FDA inserts were identified in Medicare claims and reported as summary dummy variables. We measured number of other chronic medications used per month as an indicator of pill burden, as a high number of other chronic medications used per month may cause individuals unable to manage multiple medications, burden individuals with the cost of multiple medications or cause physicians to prioritize other conditions over osteoporosis management. Finally, we captured receipt of BMD testing as an indicator of osteoporosis management. We hypothesized that factors that increase risk of osteoporosis and/or falls would be positively associated with use of any osteoporosis medication, while contraindications for osteoporosis medications would be negatively associated with use of any pharmacologic treatment for osteoporosis. Among those receiving any osteoporosis medication, the choice of bisphosphonates versus nonbisphosphonates would be affected by contraindications or side effects of bisphosphonates.

Table 1

Contraindications and side effects of osteoporosis medications included in the study

Drug class	Contraindications	Side effects
Bisphosphonates	Chronic kidney disease Hypocalcemia Esophagitis Malabsorption syndrome	GI problems Heartburn Nausea Dysphagia Abdominal pain Constipation Diarrhoea Gastric ulcer Acid reflux Dyspepsia Osteonecrosis of the jaw Osteomalacia Subtrochanteric and diaphyseal femoral fracture Bone, joint, muscle pain Back pain
Nonbisphosphonates
Calcitonin		Nasal mucosa ulcer Rhinitis Epistaxis Back pain Joint pain Headache
Parathyroid hormone	Paget's disease Kidney stone Hypercalcemia Bone metastases Skeletal malignancy	Osteosarcoma Positional hypotension Hypercalcemia Nausea Leg cramps Dizziness
Selective estrogen receptor modulator	Breast cancer Liver problem Kidney problem Hyperglyceridemia	Myocardial infarction Pulmonary embolism Deep vein thrombosis Swelling in legs Sudden change in vision Hot flashes Leg cramps

Statistical analysis

The descriptive analyses included, first, an examination of differences in characteristics between men and women among Part D enrollees with osteoporosis. For categorical variables, both chi-square tests and t-tests were conducted but only the results of t-test were reported to show where the difference between men and women occurred. For numerical variables, t-tests were conducted. Next, we analysed prevalence of any evidence-based osteoporosis medication use by sex and then by drug class.

The multivariate analysis was driven by the following considerations. In the clinical decision-making process, a physician would normally first decide whether or not to prescribe any osteoporosis medication to a patient, then decide if the patient was a good candidate for bisphosphonates since bisphosphonates are the first-line medication recommended by clinical guidelines. The final clinical decision would be choice of specific bisphosphonate for good candidates for bisphosphonates or selection of an alternative medication (calcitonin, teriparatide or raloxifene) based on considerations of contraindications and/or side effects. However, due to the relatively few nonbisphosphonate users in our sample, we were only able to model the first two steps in this clinical decision-making process.

We used modified Poisson regressions^[18] with robust errors variance to estimate factors associated with any osteoporosis medication use, and among users, use of bisphosphonates versus nonbisphosphonates. We used modified Poisson rather than logistic regression in order to derive accurate measures of relative risk because when the probability of an outcome is extremely low or high, the odds ratio approximates relative risk; however, when the probability of the outcome is common, conversion of an adjusted odds ratio to a relative risk produces inconsistent estimates of the true risk. All variables described in the measures section above were included as covariates. To understand the difference each factor had on drug use for men and women, we first conducted a full analysis with interaction terms between sex and age, race, and residential status. Significant coefficients on the interaction terms were justification for stratified analyses by sex, which were subsequently conducted. The analyses were conducted using SAS 9.3 with the REPEATED command to adjust the standard errors for multiple observations of the same individuals.

The study protocol was approved by the University of Maryland Baltimore IRB.

Results

The final study sample included 99 421 Medicare beneficiaries aged 70 or older with a diagnosis of osteoporosis: 8465 men and 90 956 women (Figure 1). The overall prevalence of osteoporosis was 11% among men meeting all the study inclusion criteria besides evidence of osteoporosis compared to 44% among women. Compared to women, men diagnosed with osteoporosis were younger, more likely to be Hispanic and Asian, in poorer health (i.e. higher RxHCC counts and higher rates of diabetes, CHF, end-stage renal disease, COPD, Alzheimer's disease/dementia, glaucoma, stroke and chronic kidney disease), and at higher risk of death (Table 2). Men were also more likely to use anticonvulsants, cancer chemotherapy and glucocorticoids, but were less likely to have BMD tests.

Figure 1

Sample selection flowchart.

Table 2

Baseline (2006) characteristics of Medicare Part D enrollees aged 70+ diagnosed of osteoporosis, by sex (N = 99 421)

Beneficiary characteristics	Women (N = 90 956)	Men (N = 8465)
Demographic characteristics
Age
70–74	22.2	24.8*
75–79	25.1	28.1*
80–84	23.4	23.2
85–89	17.0	15.3*
90+	12.3	8.5*
Race/ethnicity
White	84.9	77.3*
Black	6.2	5.0*
Hispanic	3.8	5.5*
Asian	3.5	9.5*
Other	1.6	2.6*
Geographic region
Northeast	20.8	19.7*
Midwest	23.5	20.3*
South	39.9	34.5*
West	15.8	25.5*
Residential status
Community only	79.3	80.2
LTCF only	9.9	8.2*
Both	10.8	11.7*
Socioeconomic status
Non-low income subsidy (LIS)	43.6	43.3
LIS recipient but nondual	4.2	2.6*
Medicare/Medicaid dual eligibility	52.1	54.1*
Health status
RxHCC count (SD)	7.4 (3.6)	8.2* (3.9)
Comorbidities causing or contributing to osteoporosis and fractures
Prior hip fracture	9.1	8.2*
Diabetes mellitus	31.1	41.3*
Rheumatoid arthritis/osteoarthritis	57.4	56.5
Congestive heart failure	38.2	49.2*
Depression	35.7	32.5*
End-stage renal disease	0.6	1.3*
COPD	26.9	43.9*
Breast cancer	6.6	–
Prostate cancer	–	17.9
Medications causing or contributing to osteoporosis and fractures
Anticonvulsants	14.2	17.5*
Cancer chemotherapy	6.2	11.7*
Glucocorticoids	20.7	24.7*
Risk factors for fall
Alzheimer's/dementia	27.7	30.9*
Glaucoma	23.2	24.2*
Stroke	19.9	25.0*
Contraindications/side effects of all osteoporosis medications
Chronic kidney disease	14.2	24.4*
Other contraindications of all osteoporosis medications	7.1	7.5
Side effects of all osteoporosis medications
GI problems	23.8	22.9
Other side effects of all osteoporosis medications	45.7	38.2*
Number of chronic medications used per month (SD)	3.5 (2.5)	3.4* (2.6)
Receipt of BMD testing	15.2	9.9*
Died in 2006	10.1	14.2*
Died in 2007	10.0	12.9*
Died in 2008	9.8	11.9*

BMD, bone mineral density; COPD, chronic obstructive lung disease.

*

Difference between men and women significant at α = 0.05 level.

About three quarters of men and more than half the women in our study sample received no osteoporosis drug treatment between 2006 and 2008 (Table 3). Treatment rates among women fell sharply with age from 47.7% (70–74) to 36.5% (90+), whereas the overall treatment rate for men differed little by age. There were marked differences in treatment rates by race/ethnicity. Among women, 64.4% of Asians were treated compared to 30.0% for blacks. Among men, the rates were 37.9% treated for Asians and 15.5% among blacks. Women residing in nursing homes were less likely to be treated than those living in the community (34.9% versus 44.0%), but this relationship did not hold for men. For both sexes, the likelihood of receiving any osteoporosis treatment and bisphosphonates in particular was much higher among Medicare beneficiaries with higher socioeconomic status (i.e. not receiving either Medicaid or LIS). As expected, given possible contraindications, Medicare beneficiaries with ESKD were least likely to be treated with bisphosphonates irrespective of sex, but there was no evidence that this led to a greater likelihood of treatment with nonbisphosphonates. Also as expected, osteoporosis medication rates were higher than average in both men and women receiving BMD testing, and bisphosphonate use was especially high in this group.

Table 3

Prevalence of evidence-based osteoporosis medication use among Medicare Part D enrollees aged 70+ diagnosed of osteoporosis, by sex and drug class used in 2006 (N = 99 421)

Beneficiary characteristics	Women (N = 90 956)			Men (N = 8465)
	Any treatment	Bisphosphonate users	Nonbisphosphonate users	Any treatment	Bisphosphonate users	Nonbisphosphonate users
Overall prevalence	44.3	35.6	8.7	25.1*	22.9*	2.2*
Demographic characteristics
Age
70–74	46.7	38.7	7.9	26.3*	24.6*	1.8*
75–79	46.8	38.4	8.4	23.9*	22.5*	1.4*
80–84	45.1	36.6	8.5	25.1*	22.5*	2.7*
85–89	42.2	33.0	9.2	25.7*	23.1*	2.6*
90+	36.5	26.3	10.2	24.6*	20.2*	4.5*
Race/ethnicity
White	44.4	35.4	9.0	24.5*	21.9*	2.6*
Black	30.0	25.2	4.8	15.5*	14.1*	1.4*
Hispanic	46.5	41.2	5.3	19.3*	18.4*	0.9*
Asian	64.4	53.1	11.3	37.9*	37.1*	0.9*
Other	52.0	43.8	8.2	27.9*	26.9	1.0*
Geographic region
Northeast	44.0	35.6	8.4	25.5*	22.5*	3.1*
Midwest	45.6	36.6	9.1	26.6*	24.3*	2.3*
South	42.2	33.4	8.8	21.8*	19.5*	2.4*
West	48.2	40.1	8.1	28.1*	26.8*	1.3*
Residential status
Community only	46.4	38.4	8.0	25.1*	23.6*	1.6*
LTCF only	34.9	23.0	11.9	24.1*	19.4*	4.6*
Both	37.5	27.1	10.4	25.7*	20.4*	5.3*
Socioeconomic status
Non-low income subsidy (LIS)	48.3	39.7	8.6	25.6*	24.0*	1.6*
LIS recipient but nondual	39.3	30.9	8.5	19.5*	18.1*	1.4*
Medicare/Medicaid dual eligibility	41.4	32.6	8.8	25.0*	22.2*	2.8*
Comorbidities causing or contributing to osteoporosis and fractures
Prior hip fracture	42.8	31.9	10.8	18.8*	22.6*	6.2*
Diabetes mellitus	37.8	30.3	7.5	21.9*	19.8*	2.1*
Rheumatoid arthritis/osteoarthritis	42.2	33.5	8.7	23.8*	21.7*	2.2*
Congestive heart failure	37.1	28.8	8.4	22.2*	19.8*	2.4*
Depression	39.3	30.2	9.1	24.1*	20.7*	3.5*
End-stage renal disease	22.2	13.9	8.3	20.0*	17.3*	2.7*
COPD	40.6	31.9	8.7	24.6*	22.1*	2.6*
Breast cancer	45.5	38.0	7.5	–	–	–
Prostate cancer	–	–	–	26.1	24.4	1.7
Medications causing or contributing to osteoporosis and fractures
Anticonvulsants	42.5	32.8	9.8	26.6*	22.5*	4.2*
Cancer chemotherapy	47.1	37.6	9.5	29.1*	27.3*	1.8*
Glucocorticoids	47.5	38.1	9.4	29.7*	26.7*	3.1*
Risk factors for fall
Alzheimer's/dementia	37.2	27.9	9.4	23.0*	19.7*	3.3*
Glaucoma	44.5	36.2	8.4	24.9*	22.9*	2.0*
Stroke	36.8	28.5	8.3	22.5*	19.9*	2.7*
Contraindications of bisphosphonates
Chronic kidney disease	35.4	27.2	8.2	22.3*	20.0*	2.3*
Other contraindications of bisphosphonates	45.5	36.6	8.9	28.4*	21.6*	6.8*
Side effects of bisphosphonates
GI problems	43.5	32.7	10.9	26.0*	22.7*	3.3*
Other side effects of bisphosphonates	51.8	41.6	10.1	31.8*	29.1*	2.7*
Receipt of BMD testing	67.1	56.6	10.5	51.3*	48.8*	2.5*
Died in 2006	37.5	28.2	9.3	29.6*	25.1*	4.5*
Died in 2007	36.0	26.0	10.0	22.3*	19.3*	3.0*
Died in 2008	38.0	28.1	9.9	23.3*	19.9*	3.4*

BMD, bone mineral density; COPD, chronic obstructive lung disease.

*

Difference between men and women significant at α = 0.05 level.

We also tracked changes in use of osteoporosis medications from 2006 to 2008. Overall, treatment rates ranged from 44.3% in 2006 to 42% in 2008 among women diagnosed with osteoporosis and from 25.2% in 2006 to 24.5% in 2008 among men. When stratified by drug class, bisphosphonates were the most commonly prescribed medication for osteoporosis for both men and women in all 3 years (Figure 2). Bisphosphonate use went up slightly (76.4% in 2006–78.3% in 2008 for women and from 91.3% in 2006 to 93.3% in 2008 for men). Calcitonin use went down slightly (11.6% in 2006–8.9% in 2008 for women and 10.4% in 2006–7.2% in 2008 for men). Use of SERM in women and parathyroid hormone in both men and women remained stable for all 3 years.

Figure 2

Choice of evidence-based osteoporosis medication among drug users, by drug class, sex and year.

In the initial adjusted analyses including both men and women, men had significantly lower probability of receiving osteoporosis medications (RR 0.20, 95% CI (0.19, 0.22)) (results not shown). Interaction terms between sex and age, race, and residential status were all statistically significant; therefore justifying stratified analysis. Table 4 presents adjusted results of the stratified analysis for any osteoporosis medication use among women and men. The probability of receiving any evidence-based osteoporosis medication increased with age among men, but remained relatively constant among women except for those aged 90 or older when utilization rates dropped. Black men and women were much less likely (RR 0.76, 95% CI (0.68, 0.85) for men; RR 0.61, 95% CI (0.59, 0.63) for women) to be treated for osteoporosis compared to white men and women. Hispanic men were equally likely (RR 0.95, 95% CI (0.83, 1.09)) to receive osteoporosis medications compared to white men, while Hispanic women were less likely (RR 1.36, 95% CI (1.32, 1.41)) to receive osteoporosis medication compared to white women. In contrast to the unadjusted results, nursing home residence actually associated with increased probability of treatment once all other factors were controlled (RR 1.65, 95% CI (1.44, 1.90) for men and RR 1.17, 95% CI (1.15, 1.20) for women). Having diabetes or chronic kidney disease was associated with reduced likelihood of treatment in both sexes, whereas using anticonvulsants, cancer chemotherapy and glucocorticoids had the opposite association. Having BMD testing was associated with significantly increased probability of receiving any evidence-based osteoporosis medication among both men and women, although the association was much stronger in men.

Table 4

Estimated predictors for any use of evidence-based osteoporosis medications

	Women		Men
	Prevalence ratio	95% confidence interval	Prevalence ratio	95% confidence interval
Age
70–74 (ref)
75–79	1.03	1.02–1.04*	1.06	1.00–1.12*
80–84	1.05	1.04–1.07*	1.25	1.17–1.34*
85–89	1.02	1.01–1.04*	1.29	1.20–1.40*
90+	0.96	0.94–0.98*	1.39	1.27–1.52*
Race/ethnicity
White (ref)
Black	0.61	0.59–0.63*	0.76	0.68–0.85*
Hispanic	1.36	1.32–1.41*	0.95	0.83–1.09
Asian	1.96	1.90–2.02*	1.89	1.70–2.10*
Other	1.25	1.19–1.31*	1.25	1.07–1.46*
Geographic region
Northeast (ref)
Midwest	0.94	0.92–0.96*	0.97	0.89–1.05
South	0.95	0.93–0.96*	0.93	0.86–1.00*
West	1.04	1.01–1.06*	1.18	1.09–1.28*
Residential status
Community only (ref)
LTCF only	1.17	1.15–1.20*	1.65	1.44–1.90*
Both	1.12	1.10–1.14*	0.97	0.89–1.06
Socioeconomic status
Non-LIS recipient (ref)
Low income subsidy (LIS) recipient but nondual	0.88	0.86–0.91*	1.01	0.88–1.15
Medicare/Medicaid dual eligibility	0.84	0.82–0.85*	1.07	1.00–1.13*
RxHCC count	1.02	1.02–1.02*	1.01	1.01–1.02*
History of hip fracture	1.23	1.20–1.26*	1.14	1.03–1.25*
Diabetes mellitus	0.80	0.79–0.81*	0.89	0.85–0.94*
Rheumatoid arthritis/osteoarthritis	1.11	1.10–1.12*	1.11	1.05–1.16*
Congestive heart failure	0.85	0.84–0.86*	0.98	0.92–1.03
Depression	0.92	0.91–0.93*	1.03	0.97–1.09
End-stage renal disease	0.74	0.68–0.8*	0.96	0.82–1.13
COPD	0.97	0.95–0.98*	1.03	0.98–1.08
Breast cancer	1.04	1.02–1.07*	–	–
Prostate cancer	–	–	1.09	1.02–1.16*
Alzheimer's/dementia	0.90	0.89–0.91*	1.01	0.95–1.07
Glaucoma	1.04	1.03–1.05*	1.02	0.97–1.09
Stroke	0.90	0.89–0.91*	0.99	0.93–1.04
Anticonvulsants	1.17	1.15–1.19*	1.63	1.51–1.77*
Cancer chemotherapy	1.15	1.13–1.18*	1.36	1.22–1.52*
Glucocorticoids	1.22	1.21–1.24*	1.82	1.69–1.96*
Chronic kidney disease	0.90	0.89–0.91*	0.94	0.89–0.99*
Other contraindications of all osteoporosis medications	1.00	0.98–1.01	1.00	0.93–1.08
GI problems	0.98	0.97–0.99*	1.00	0.95–1.04
Other side effects of all osteoporosis medications	1.03	1.02–1.04*	1.08	1.04–1.12*
Number of chronic medications used per month	1.07	1.06–1.07*	1.04	1.03–1.05*
Receipt of BMD testing	1.33	1.31–1.34*	2.34	2.11–2.58*
Year
2006 (ref)
2007	0.99	0.99–1.00*	1.01	0.98–1.04
2008	0.98	0.97–0.99*	1.00	0.96–1.04
Death	1.52	1.50–1.54*	4.20	3.98–4.44*

BMD, bone mineral density; COPD, chronic obstructive lung disease.

*

P < 0.05.

When treated, men had significantly higher probability of receiving nonbisphosphonates in the combined analyses of men and women (RR 1.16, 95% CI (1.16, 1.17)) (results not shown). Interaction terms between sex and age, race, and residential status in this model were all statistically significant, thus justifying stratified analysis by sex. Table 5 presents adjusted results from the stratified analysis by type of medication use (nonbisphosphonates versus bisphosphonates) among the treated. For both men and women, the probability of receiving nonbisphosphonates increased with age. Asian men were much less likely to receive nonbisphosphonates (RR 0.31, 95% CI (0.18, 0.52)) compared to white men; however, Asian women were equally likely (RR 0.97, 95% CI (0.89, 1.06)) to receive nonbisphosphonates compared to white women. Nonbisphosphonate use was much higher among men and women residing in nursing facilities (RR 1.27, 95% CI (1.21, 1.32) for women, and RR 1.43, 95% CI (1.14, 1.78) for men). Among disease factors, having ESKD was a strong predictor of receiving nonbisphosphonates in women (RR 1.52, 95% CI (1.32, 1.76)), but not in men.

Table 5

Estimated predictors for nonbisphosphonate use, stratified by sex

	Women		Men
Prevalence ratio	95% confidence interval	Prevalence ratio	95% confidence interval
Age
70–74 (ref)
75–79	1.03	1.00–1.06*	0.97	0.80–1.18
80–84	1.05	1.02–1.09*	1.29	1.05–1.59*
85–89	1.15	1.10–1.19*	1.43	1.14–1.80*
90+	1.22	1.17–1.28*	1.44	1.12–1.86*
Race/ethnicity
White (ref)
Black	0.75	0.69–0.81*	0.71	0.50–1.00
Hispanic	0.62	0.56–0.69*	0.46	0.28–0.76*
Asian	0.97	0.89–1.06	0.31	0.18–0.52*
Other	0.82	0.71–0.93*	0.55	0.30–1.02
Geographic region
Northeast (ref)
Midwest	1.02	0.97–1.07	0.77	0.62–0.97*
South	1.14	1.10–1.19*	1.11	0.91–1.35
West	0.89	0.84–0.94*	0.66	0.51–0.86*
Residential status
Community only (ref)
LTCF only	1.27	1.21–1.32*	1.43	1.14–1.78*
Both	1.15	1.11–1.19*	1.31	1.10–1.55*
Socioeconomic status
Non-LIS recipient (ref)
Low income subsidy (LIS) recipient but nondual	1.04	0.98–1.11	1.24	0.85–1.81
Medicare/Medicaid dual eligibility	1.04	1.01–1.08*	1.28	1.07–1.54*
RxHCC count	1.00	0.99–1.00*	1.01	0.99–1.03
History of hip fracture	1.04	1.00–1.09*	1.25	1.03–1.53*
Diabetes mellitus	1.00	0.97–1.03	0.92	0.80–1.07
Rheumatoid arthritis/osteoarthritis	1.02	1.00–1.05	0.94	0.81–1.09
Congestive heart failure	1.03	1.00–1.06*	1.16	0.99–1.36
Depression	1.07	1.04–1.10*	1.07	0.91–1.24
End-stage renal disease	1.52	1.32–1.76*	1.38	0.85–2.22
COPD	1.05	1.02–1.09*	0.96	0.82–1.11
Breast cancer	0.89	0.84–0.94*	–	–
Prostate cancer	–	–	0.96	0.80–1.16
Alzheimer's/dementia	1.06	1.03–1.09*	0.97	0.83–1.13
Glaucoma	0.99	0.96–1.02	0.94	0.79–1.11
Stroke	1.02	0.99–1.05	0.98	0.83–1.14
Anticonvulsants	0.99	0.96–1.02	1.20	1.04–1.40*
Cancer chemotherapy	0.96	0.92–1.00	0.95	0.76–1.19
Glucocorticoids	0.98	0.96–1.00*	0.95	0.83–1.09
Chronic kidney disease	1.02	0.99–1.05	1.05	0.91–1.22
Other contraindications of bisphosphonates	0.95	0.88–1.02	1.23	0.81–1.86
GI problems	1.03	1.01–1.05*	1.14	1.01–1.28*
Other side effects of bisphosphonates	1.01	0.99–1.02	1.05	0.93–1.17
Number of chronic medications used per month
Receipt of BMD testing	0.90	0.89–0.91*	0.83	0.75–0.93*
Year
2006 (ref)
2007	0.97	0.96–0.98*	0.89	0.82–0.96*
2008	0.93	0.92–0.95*	0.75	0.67–0.84*
Death	0.97	0.94–1.00	1.05	0.90–1.22

BMD, bone mineral density; COPD, chronic obstructive lung disease.

*

P < 0.05.

Discussion

This is the first large national study to compare patterns of evidence-based osteoporosis medication use among elderly men and women in the Medicare programme. We found low rates of osteoporosis medication use in both men and women diagnosed with osteoporosis, but it was especially low among men. Osteoporosis medication use rates were found to be 44.3–42% in women and 25.2–24.5% in men in this study, which were on the higher end compared to what was found in the literature.^[2–4] Even after adjusting for age and other characteristics, the probability of an elderly man with diagnosed osteoporosis being treated with evidence-based pharmacotherapy was just 20% of that in women. Although we found some factors associated with sex-related differences in treatment (e.g. higher prevalence of chronic kidney disease in men), most of the difference remains unexplained by the factors included in our model. This huge difference may be due to lack of awareness of the disease and its consequences in elderly men and their physicians.^[7] Osteoporosis has traditionally been seen as a women's disease, so elderly men and their physicians may fail to manage it appropriately.^[19,20] This lack of awareness was however unobserved in our results. Why that is the case requires further study. It may be due to other provider/system level factors, such as number/type of providers seen before diagnosed with osteoporosis, provider's adherence to national osteoporosis guidelines or provider-patient interaction that were shown in our conceptual framework but not observable in our data.

When treated, elderly men with osteoporosis were much more likely to receive bisphosphonates compared to elderly women. Whether this is true today remains to be seen. In 2010, a new class of osteoporosis drug – RANKL inhibitors was approved by the FDA. The first RANKL inhibitor, denosumab, is indicated for treatment of osteoporosis in men and postmenopausal women at higher risk for fracture and/or with nonmetastatic prostate cancer or breast cancer. Denosumab is a once every 6 months subcutaneous injection covered by Part B and is recommended as first-line therapy along with bisphosphonates by the 2010 American Association of Clinical Endocrinologist Medical guidelines^[21] and the 2014 NOF Clinicians’ Guide.^[22] Availability of RANKL inhibitors may increase men's likelihood of using evidence-based osteoporosis medications.

Besides low utilization rates, we also discovered important sex-related racial distinctions in osteoporosis treatment. Blacks had by far the lowest treatment rates (30% for women and 15.5% for men). Whites were in the middle (44.4% for women and 24.5% for men) with the highest rates for Asians (64.4% for women and 37.9% for men). Treatment rates among Hispanic women (46.5%) exceeded that of whites, but the rate for Hispanic men (19.3%) was significantly below that for white men. Such wide racial and ethnic differences raise important questions about osteoporosis education and access to care that deserve further research.

In particular, our finding that receipt of BMD testing significantly increases the probability of osteoporosis treatment use for both sexes, but more so for men is telling. This is consistent with the view that men receive BMD tests later in the course of their disease and thus are more likely to be treated as a result. This also suggests that BMD testing should be expanded particularly for men. Since Medicare Part B covers BMD tests once every 24 months for beneficiaries who are at risk of developing osteoporosis, physicians should make Medicare beneficiaries aware of and take better advantage of this coverage to help identify osteoporosis and initiate treatment early.

Limitations

This study has several limitations. First, we only tracked osteoporosis treatment patterns through 2008. The study provides important information about how osteoporosis was treated during the formative years of the Medicare Part D programme, but clearly, analysis of more recent trends is warranted. Second, the CCW provides data on drug use only for Medicare Part D enrollees. Therefore, the study findings may not generalize to Medicare beneficiaries not enrolled in Part D. Also, MAPD enrollees were excluded because they did not generate Part A and B claims necessary for identifying evidence of osteoporosis. Use and choice of osteoporosis medication may differ between MAPD and PDP enrollees, and therefore, the findings of this study may not be generalizable to MAPD enrollees, either. Fourth, drug use during beneficiaries’ Part A covered hospitalization and post-acute SNF stays was not observed due to CMS reimbursement policy. Fifth, use of chronic kidney disease as a contraindication of bisphosphonates may have introduced bias. The exact contraindication of bisphosphonates is creatinine clearance <30 ml/min. However, the actual creatinine clearance value is not available in CCW data, thus we can only use diagnosis of chronic kidney disease as a proxy and the result of chronic kidney disease's impact on treatment use are likely biased towards the null. Perhaps most significant of all limitations is the fact that we were unable to capture use of over-the-counter calcium and vitamin D, both of which may be effective in reducing bone fractures. Part D does not cover over-the-counter medications and thus we have systematically underestimated the degree to which Medicare beneficiaries received any osteoporosis treatment.

Conclusion

Elderly men are undertreated for osteoporosis compared to elderly women, and this is particularly problematic among blacks. Policy makers should expand the focus of improving osteoporosis treatment to include men, for example men should be included in the measure of osteoporosis management in addition to women in the Medicare Part C start rating system, and clinicians should treat their male patients with osteoporosis more appropriately in order to reduce the sexual difference in osteoporosis treatment.

Declarations

Conflict of interest

The Author(s) declare(s) that they have no conflicts of interest to disclose.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Authors’ contributions

FL was responsible for drafting the article. BS was responsible for acquisition of the data for the manuscript. FL, BS, DS AD, EO and MH were responsible for the conception and design of the manuscript, analysis of interpretation of the data, critically revising the article, and provided final approval of the article.