1. Top of page
  2. Introduction
  3. Patients and Methods
  4. Results
  5. Discussion

Osteoporotic-related insufficiency fractures are a common cause of morbidity and excess mortality, yet high-risk patients often do not receive treatment. Initiating osteoporosis treatments in the acute-care setting is problematic for a variety of reasons, yet even after discharge, estimates from larger studies suggest that only 3–42% of postfracture patients subsequently receive any prescription therapy for secondary prevention (1–4). Because up to half of US Medicare patients receive home health services following hospitalization for a hip fracture (5) and more than 2 million home health care visits are provided annually to fracture patients (3), the home health care period may provide a window of opportunity for osteoporosis intervention. Many of the patients seen in home health care are at high risk for future fracture on the basis of multiple comorbidities including prior fractures, falls, stroke, and current glucocorticoid use (6–9). Moreover, home health offers the opportunity to provide standardized care to patients under the supervision of a primary care physician with whom continuity of care is likely to be greater than in other settings.

As a prelude to a randomized controlled trial designed to improve quality of care for patients at high risk of fracture, we examined the epidemiology of osteoporosis and its associated treatment among high-risk patients receiving home health care services. We were particularly interested in how patients' burden of medical comorbidity affected their receipt of osteoporosis medications. We hypothesized that individuals with greater comorbidity would be less likely to receive medications for primary or secondary prevention of fractures.

Patients and Methods

  1. Top of page
  2. Introduction
  3. Patients and Methods
  4. Results
  5. Discussion

Role of the study sponsor.

Eli Lilly had no role in the study design, data collection, analysis, or preparation or review of the manuscript, or the decision to submit the manuscript for publication.

Patient population.

Following institutional review board approval, we used the electronic databases of a large home health care agency to identify patients receiving home health services during 2003–2004 who were at high risk for fracture. This agency provides care to ∼7,000 patients annually throughout a single southeastern US state. We defined high-risk patients as those with a history of fracture, a physician diagnosis of osteoporosis, current systemic glucocorticoid use (any dose), and a history of stroke. The electronic database used for the study was the primary record of the patient's clinical encounter with the trained nurses providing home health services. The database captured demographic data and a comprehensive list of medical diagnoses, prescription drugs, and over-the-counter medications and supplements relevant to each home health visit. The index developed by Charlson et al was used to quantify the burden of medical comorbidity for each patient (10). This index assigns an integer value to a number of medical conditions based on the conditions' previously validated association with adverse outcomes. Each patient in the cohort was assigned an individualized score based on the summed values of their specific medical conditions recorded in the electronic database. Patients receiving hospice services were excluded from analysis, although these individuals were included if they received home health services prior to hospice referral.

Descriptive statistics were used to characterize the patient cohort and examine rates of prescription and over-the-counter osteoporosis treatments. Multivariable logistic regression was used to examine factors associated with prescription osteoporosis medication use (i.e., alendronate, risedronate, raloxifene, calcitonin, and teriparatide). For all multivariable models, covariates significant with a bivariate P value less than 0.25 were entered, and using a backward elimination procedure, a P value less than 0.05 was required to remain in the model. Model building was conducted according to Hosmer and Lemeshow, and model discrimination was assessed using a c statistic (11). Data management and statistical analysis were performed using SAS (SAS Institute, Cary, NC).


  1. Top of page
  2. Introduction
  3. Patients and Methods
  4. Results
  5. Discussion

Characteristics of the patients at high risk of fracture who were receiving home health services during 2003–2004 are shown in Table 1. Almost 75% were women, and the mean age was 78 years. Almost all were Medicare enrollees, and ∼40% were referred to home health care more than once during the 2-year observation window. Nineteen percent were referred specifically for a recent fracture, and >25% had a history of fracture at any site. Current glucocorticoid use was the most common reason that patients were considered to be at high risk of fracture.

Table 1. Characteristics of home health care patients at high risk for fracture*
CharacteristicAll patientsOsteoporosis diagnosisReferred for recent fractureHistory of hip fractureHistory of any fractureCurrent glucocorticoid useStroke
  • *

    Values are the number (percentage) unless otherwise indicated. High fracture risk defined by physician diagnosis of osteoporosis, history of fracture, current glucocorticoid use, or history of stroke. HMO = health maintenance organization; PPO = preferred provider organization; OTC = over the counter.

  • All patients column represents unique patients; all other columns reflect individuals with these conditions and are not necessarily mutually exclusive.

  • Denominator is total episodes of home health care (n = 3,197).

Total2,647 (100)501 (19)499 (19)184 (7)730 (28)1,104 (42)714 (27)
 Female sex (male referent)1,901 (72)474 (95)392 (78)138 (75)558 (77)753 (69)460 (65)
 Age, mean ± SD years78 ± 1282 ± 978 ± 1280 ± 1078 ± 1275 ± 1278 ± 10
  White2,339 (88)479 (96)469 (94)172 (94)685 (94)981 (89)574 (80)
  African American284 (11)21 (4)24 (5)10 (5)38 (5)118 (10)129 (18)
  Other24 (1)1 (0)6 (1)2 (1)7 (1)5 (1)11 (2)
 Insurance coverage       
  Medicare2,519 (95)482 (96)465 (93)172 (94)681 (94)1,036 (94)677 (95)
  HMO/PPO/commercial43 (2)3 (1)7 (2)0 (0)10 (1)23 (2)9 (1)
  Self pay63 (2)13 (2)14 (3)8 (4)21 (3)29 (3)13 (2)
Health service utilization       
 Previous home health referrals       
  01,568 (59)273 (55)303 (61)120 (65)437 (60)570 (52)460 (65)
  1592 (22)123 (25)108 (21)39 (21)160 (22)274 (25)141 (20)
  >1487 (18)105 (20)88 (18)25 (14)133 (18)260 (23)113 (15)
 Referred specifically for fracture care499 (19)100 (20)499 (100)145 (29)433 (59)75 (15)30 (6)
 Days of service per home health care episode, mean ± SD117 ± 178142 ± 28078 ± 11071 ± 10383 ± 138150 ± 21690 ± 124
 Referred for hospice services or died during episode of care263 (8)27 (5)22 (4)8 (4)27 (4)119 (11)60 (8)
 Location of home health services provided       
  Patient's home1,994 (62)350 (70)366 (73)132 (72)529 (72)866 (78)521 (73)
  Family member's home484 (15)82 (16)89 (18)31 (17)129 (18)172 (15)143 (20)
  Assisted living facility196 (6)62 (12)36 (7)16 (9)59 (8)55 (5)37 (5)
  Other/not known523 (16)7 (1)8 (2)5 (2)13 (2)11 (1)13 (2)
 Impaired decision-making capacity at the time of referral670 (25)130 (26)92 (18)36 (20)153 (21)239 (22)230 (32)
 Charlson comorbidity index, mean ± SD1.5 ± 1.41.1 ± 1.20.9 ± 1.20.7 ± 1.00.8 ± 1.11.8 ± 1.52.1 ± 1.3
 Number of unique prescription and OTC medications, mean ± SD19 ± 1419 ± 1416 ± 1313 ± 715 ± 1125 ± 1616 ± 12

Rates of osteoporosis-specific medications, hormone therapy, and over-the-counter calcium and vitamin D use are shown in Table 2. Almost three-quarters of the patients who had a physician-identified diagnosis of osteoporosis received either prescription medications or over-the-counter calcium and vitamin D supplements. In contrast, fewer than half of the patients specifically referred for fracture received any therapy, and patients with a history of stroke had the lowest rates of treatment in this high-risk cohort. The amount of overlap between the various high-risk categories was relatively low (data not shown). For example, only 100 of the 499 patients referred for fracture had physician-diagnosed osteoporosis.

Table 2. Rates of receipt of prescription osteoporosis medications, oral bisphosphonates, calcium/vitamin D, and hormone therapy*
 No.Alendronate or risedronateRaloxifene, calcitonin, or teriparatideAny prescription osteoporosis medication§Hormone therapyCalcium and/or vitamin D supplementsAny of these therapies
  • *

    Values are the number (percentage) unless otherwise indicated.

  • Columns do not sum to total because patients could be classified in more than 1 category.

  • More than half of medication use in this column was for calcitonin.

  • §

    Includes alendronate, risedronate, raloxifene, calcitonin, or teriparatide.

  • Includes estrogen and estrogen/progestin preparations.

Diagnosis of osteoporosis501182 (36)126 (25)270 (54)55 (11)213 (43)357 (71)
Referred to home health for recent fracture49986 (17)62 (12)132 (26)41 (8)119 (24)204 (41)
History of hip fracture18423 (13)10 (5)31 (17)9 (5)33 (18)56 (30)
History of any fracture730100 (14)60 (8)146 (20)51 (7)151 (21)252 (35)
Current systemic glucocorticoid use (any dose)1,104143 (13)101 (9)211 (19)100 (9)197 (18)374 (34)
History of stroke71440 (6)26 (4)63 (9)41 (6)78 (11)144 (20)
Total2,647341 (13)227 (9)513 (19)202 (8)488 (18)874 (33)

The bivariate and multivariable-adjusted factors associated with receipt of osteoporosis-specific medications are shown in Table 3. A referral specifically for fracture and a physician diagnosis of osteoporosis were strongly associated with receipt of osteoporosis medications. Conversely, higher degrees of medical comorbidity were associated with a decreased likelihood of receiving prescription therapy in the multivariable model.

Table 3. Factors associated with receipt of prescription osteoporosis therapies (n = 2,647)*
 Crude OR95% CIAdjusted OR95% CI
  • *

    Includes alendronate, risedronate, raloxifene, teriparatide, and calcitonin. OR = odds ratio; 95% CI = 95% confidence interval.

  • Only factors independently significant at P < 0.05 were retained in the multivariable model; c statistic for multivariable model = 0.81; model satisfies Hosmer-Lemeshow goodness-of-fit.

 Female sex (male referent)4.273.16–5.772.251.62–3.12
 Age, years1.021.01–1.03  
Health service utilization    
 Referred for fracture during this home health care episode1.671.33–2.101.871.43–2.45
 Previous home health referrals (0 referent)
 Days of service (30-day increment)1.031.02–1.05  
 Medicare enrollee1.470.93–2.33  
 Referred for hospice services0.730.54–0.98  
 Services provided in patient's home0.840.68–1.05  
Osteoporosis indicators and selected comorbidities   
 Physician diagnosis of osteoporosis9.157.34–11.418.266.50–10.49
 History of hip fracture0.830.56–1.24  
 History of nonhip fracture1.140.91–1.44  
 Current systemic glucocorticoid use (any dose)0.970.80–1.18  
 History of stroke0.320.24–0.42  
 Impaired decision-making capacity1.020.81–1.27  
 Charlson comorbidity index0.860.80–0.930.860.78–0.94
 Number of unique medications1.031.02––1.05

Absolute unadjusted rates of prescription osteoporosis treatment decreased as comorbidity increased (data not shown). A total of 26% of patients with a Charlson comorbidity score of 0 received prescription osteoporosis medications, which declined to a low of 17% for patients with a Charlson score ≥3.


  1. Top of page
  2. Introduction
  3. Patients and Methods
  4. Results
  5. Discussion

Because the window of opportunity for osteoporosis intervention may not be in the inpatient setting for hospitalized patients, we examined rates and factors associated with osteoporosis treatment among patients with a history of fracture or one of several other high-risk conditions who were receiving home health care services. Only one-third of the cohort received any prescription or over-the-counter medications used to prevent or treat osteoporosis, which is consistent with national data showing widespread undertreatment (1–4, 12–15). Some studies suggest that hospitalized fracture patients may have better outcomes if they receive home health services after discharge (16), and home health care may be an ideal setting in which to intervene and focus fracture prevention efforts.

Health care providers who recognized that a patient had osteoporosis were much more likely to provide treatment. These data are concordant with other studies demonstrating improved osteoporosis management at the time of discharge in hospitalized patients with hip fracture whose doctor diagnosed them with osteoporosis (17). Although perhaps intuitive, this finding underscores the opportunity for quality improvement interventions and allied health providers to assist physicians in identifying that an insufficiency fracture confers the diagnosis of osteoporosis, a systemic skeletal disease that may benefit from treatment.

Concordant with our initial hypothesis, we demonstrated that an increased burden of medical comorbidities decreased the likelihood that patients received osteoporosis treatment. This observation is similar to results from a study of 21,192 Medicare beneficiaries in Pennsylvania who recently experienced a hip or wrist fracture, which demonstrated that the odds of receiving osteoporosis medication were 0.65 among patients with ≥2 medical comorbidities compared with those with no comorbidities (2). In contrast, another study examining 3,492 postmenopausal women enrolled in managed care found that osteoporosis treatment rates increased with medical comorbidity (quantified using a different metric, the Chronic Disease Score) (4). That study also demonstrated that older age was associated with a decreased likelihood of receiving osteoporosis treatment, in contrast to our results, which did not show that age was an independent factor. Differences between patient populations in mean age, insurance status, and fracture history likely account for this discordance. Similar to our study, rates of osteoporosis treatment for high-risk patients are low nationally despite the recognition that prior fracture, low-dose chronic glucocorticoid use (i.e., prednisolone 2.5–7.5 mg daily), and stroke increase fracture risk (1–4, 18, 19).

The strengths of our study include a well-characterized cohort of patients receiving home health services with data available from a comprehensive electronic record. To our knowledge, osteoporosis treatment for patients receiving home health care has not been previously examined within a US population. The limitations to our work include the somewhat-limited period that patients received home health care services (∼4 months per episode of care); patients may have received osteoporosis treatment after these services were discontinued. They may also have received bone mineral density testing, which was not captured in the home health care database. Although all patients in our cohort were selected on the basis of 1 or more risk factors for fracture, gradations in patient-specific absolute fracture risk and the expected benefit of pharmacologic therapy may have led physicians to decide that prescription osteoporosis medications were not appropriate for some individuals. However, we contend that calcium and vitamin D (which we assessed) should be provided as a minimum standard of care for patients who have even a moderate fracture risk. Finally, our data were obtained from patients receiving home health services in a single southeastern US state and therefore may have limited generalizability to other regions.

In conclusion, we demonstrated that rates of osteoporosis treatment are low for patients receiving home health care services who are at high risk for fracture. These data are similar to many national studies that also show undertreatment of osteoporosis for high-risk patients (1–4, 12–15). However, home health care has received only minimal attention as a setting that might be fruitful for osteoporosis quality improvement interventions. Because the recognition of a diagnosis of osteoporosis was associated with significantly better management, interventions that identify individual patients with osteoporosis or at risk for osteoporosis based on clinical factors, including those that we studied (20, 21), may be effective in promoting improved care. Additionally, higher degrees of medical comorbidity appear to reduce the likelihood that patients receive osteoporosis care, perhaps due to competing pressures on physicians' time to manage these illnesses. Rather than simply asking individual physicians to do better, physician-led changes in system-wide disease management programs are needed to treat these complex patients in a timely and resource-efficient fashion (22). We await results from osteoporosis quality improvement interventions and proffer the suggestion that home health care may be an important but underrecognized setting in which to focus these efforts.


  1. Top of page
  2. Introduction
  3. Patients and Methods
  4. Results
  5. Discussion