An Opportunity for Improving Osteoporosis Treatment in Home Health Care

Abstract

Purpose: To examine osteoporosis prevention and treatment among home health care (HHC) patients at risk of fragility fracture in a large, Midwestern integrated HHC system. Methods: All patients who received HHC services in 2006 were identified. International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes and pharmaceutical data were examined between January 1, 2004 and December 31, 2005 to determine risk status (high vs average) for fragility fracture. Patients with a documented diagnosis of osteoporosis, osteopenia, previous fragility fracture, stroke, or those taking a glucocorticoid were categorized as high risk. Pharmaceutical data (eg, estrogen, bisphosphonates) were obtained during the same 2-year period to determine treatment status. Descriptive statistics documented the proportion at high risk and treatment status. Inferential statistics tested differences in characteristics (age, gender, race, number of comorbidities) among high-risk patients with and without treatment. Results: 2798 patients were seen in HHC during 2006 and had utilization data available in 2004 and 2005. Of these, 754 were categorized as high risk and 2044 as average risk. Approximately one third (34%) of high-risk patients received osteoporosis medication compared to 4% of average risk (P < .0001). We found no treatment differences based on age. Those with higher comorbidity profiles were less likely to receive treatment (P < .0001). Conclusion: Only 34% of HHC patients at high risk for fracture received adequate treatment. Patients with more comorbidities were least likely to receive treatment. Since these individuals are receiving medical and nursing care, an opportunity exists to increase treatment rates for those at greatest risk.

Introduction

Osteoporosis, characterized by low bone mass and structural deterioration of bone tissue, leading to bone fragility and increased susceptibility to fractures, is a significant and common health problem facing the aging population of the United States.^1–3 It is the leading underlying cause of fracture in elderly individuals, resulting in significant morbidity and mortality.^2,4,5 Mortality from osteoporosis-related fractures is greater in women than ovarian and breast cancers combined.⁶ The health burden and impact on independence from this condition is considerable. During the first year after a hip fracture, 20% of patients die and one third are admitted to nursing homes.⁶

Even though osteoporosis is common, it is often undertreated, even in patients with a history of fracture.^7–11 In 2005, Solomon et al reported that only 22% of patients admitted for hip and forearm fractures filled a prescription for osteoporosis medication.¹¹ Harrington et al also noted considerable undertreatment in patients with hip fracture across 4 health systems; only 12% to 24% of patient received dual energy X-ray absorptiometry (DEXA) scans and 5% to 37% were prescribed osteoporosis medications.¹² In a more recent study of 240 patients who sustained wrist fractures, only 21.3% underwent DEXA scans following their fracture.¹³

While most individuals at risk for osteoporosis live within the community, those individuals who receive home health care (HHC) are in a good position to be assessed and treated. However, even this environment may represent an overlooked opportunity for education about and management of osteoporosis. Curtis et al have documented undertreatment in this population and found that those with more medical comorbidities were found least likely to receive treatment.¹⁴

We sought to determine whether findings from the literature corresponded to the care of patients receiving HHC within our integrated health system where records are comprehensive. We categorized patients by risk factor, compared the proportion of high-risk patients versus average risk patients receiving treatment, and compared characteristics of at-risk patients who did and did not receive treatment. This article provides the results of our investigation.

Methods

Setting

This project was conducted through a partnership among the HealthPartners Research Foundation (HPRF), HealthPartners Integrated Home Care, and the Geriatric Orthopaedic Trauma Program at the health plan’s owned hospital. HealthPartners (HP) is an integrated health care network of owned and contracted clinics providing health care services and coverage to more than 700 000 members overall and 300 000 members in its owned facilities. Coverage is comprehensive for most members, so there is an incentive to seek all care within the system, and an integrated medical record follows the patient throughout the system. Electronic medical records (EMRs) were fully implemented in 2004 and contain information on medication orders, diagnoses, and demographic information. The HPRF supports the research activities of HP. Integrated Home Care is the Medicare-certified home health agency of HP, which provides around 52 000 intermittent skilled nursing and therapy visits annually to over 3000 patients. The average age of patients receiving HHC services is 74 years. All home care services are under the supervision of an attending physician, who signs the home care plan of care. The Geriatric Orthopaedic Trauma Program is affiliated with the owned hospital, the health system, and the University of Minnesota. The study was approved by the health plan’s Institutional Review Board.

Participant Risk Classification

All health plan patients who received HHC services in 2006 were identified through the EMRs. From this population, we classified each patient as being at high risk or average risk for future fracture. International Classification of Diseases, Ninth Revision (ICD-9) diagnoses codes and pharmacy generic product identifier (GPI) codes documented between January 1, 2004 and December 31, 2005 (the 2 years prior to when participants were identified) were used for classification. High-risk conditions were obtained from the literature included a diagnosis of osteoporosis or osteopenia, history of previous fragility fracture or stroke, or glucocorticoid use (at least 2 orders). This study limited glucocorticoid use to prednisone. Those without these conditions were considered “average” risk. Although vitamin D deficiency is included the literature, we found very few recorded instances of this condition among our participants and therefore did not use it as a criterion.

Determining Treatment

Pharmacy data were also obtained during the same time period. Treatments included alendronate, risedronate, raloxifene, calcitonin, estrogen, teriparatide, and testosterone. We also attempted to identify over-the-counter use of calcium, vitamin D, and multivitamins through the use of physician notes within the EMRs that required a text search of keywords within the note for visits that occurred during the study time frame. Over-the-counter medications (OTCs) were poorly documented, however, and not robust enough to be used as a treatment category.

Treatment was defined as a dichotomous (yes/no) variable. To be considered “treated,” the patient needed to be taking at least one of the osteoporosis treatment medications mentioned above. In addition to ICD-9 and GPI codes, we also obtained demographic data (age, gender, race, comorbidity status). Age categories included 60 years old and younger, 61 to 70, 71 to 80, 81 to 90, and 91 years and older. Comorbidity was determined using the Charlson score and assessed using 0, 1, or 2+ as categories for analysis.¹⁵ Charslon score was constructed by examining outpatient and inpatient diagnoses between January 1, 2004 and December 31, 2005.

Analysis

Descriptive statistics were used to document the proportion of patients at high risk overall and by each risk factor to summarize the proportion of high risk and average risk patients receiving treatment and to summarize demographic characteristics for each of the risk groups. Pearson chi-square statistics were calculated to assess demographic differences between the high-risk group, with and without treatment, by gender and race. Because age and comorbidity status were categorized into more than 2 groups, Cochran-Armitage trend tests were used to examine differences between the high-risk group with and without treatment. Findings were considered significant at the .05 level.

Results

There were 2798 patients seen in HHC in 2006. Of these, 754 patients were classified as high risk for fragility fracture and 2044 as average risk. The primary reasons for high-risk status were a documented diagnosis of osteoporosis (44.3%), glucocorticoid use (33.8%), history of stroke (23.2%), and osteopenia (22.6%). About a fifth (19.6%) of the high-risk patients had a previous fracture. Just over a fifth (21.13%) of the high-risk group had more than one risk factor. We also examined treatment by risk factor to assess whether any indication correlated with an increased likelihood of treatment. We found that those with a documented diagnosis of osteoporosis were the most likely to be treated (58%; see Table 1 ).

Table 1.

Treatment by Risk Factor in High-Risk Group (n = 754)

Risk Factor	At Least 1 Treatment, n (%)	No Treatment, n (%)
Osteoporosis	150 (57.9)	109 (42.1)
Prior fracture	49 (33.3)	98 (66.7)
Osteopenia	23 (28.8)	108 (81.2)
Use of glucocorticoid	25 (18.8)	57 (71.2)
History of stroke	12 (08.9)	123 (91.1)

By comparing the high-risk and average risk groups for treatment, it was found that 259 or about 34% of high-risk patients versus 4.2% of the average risk group received at least 1 treatment (P < .0001). Bisphosphonates were the most prescribed therapy in the high-risk group (27.1%). All other types of osteoporosis medication prescribed in this group were used by fewer than 10% of patients. Of the 4% on treatment in the average risk group, estrogens were the most prescribed medication category (2.3%).

The majority (75.5%) of patients in the high-risk group were female. Although race was missing for 24% of the high-risk group patients, 88% of those whose race was characterized were Caucasian (Table 2 ). Approximately 74.4% of the patients were 71 years or older. Charlson scores were computed for each patient in the high-risk group (0, 1, or 2+). Fifty-eight percent of high-risk patients had a Charlson score of 2+, indicating increased comorbidity. A total of 259 (34.4%) patients from the 754 high-risk group had at least 1 treatment.

Table 2.

Characteristics by Treatment Status of High-Risk Patients (n = 754)

Characteristic	At Least 1 Treatment		P Value
	No	Yes
Total, n (%)	495 (65.6)	259 (34.4)
Gender			<.0001
Male	165 (89.2)	20 (10.8)
Female	330 (58.0)	239 (42.0)
Age, y			.10
≤60	70 (73.7)	25 (26.3)
61-70	57 (58.2)	41 (41.8)
71-80	150 (66.7)	75 (33.3)
81-90	183 (64.2)	102 (35.8)
91+	35 (68.6)	16 (31.4)
Race			.017
Caucasian	323 (64.1)	181 (35.9)
Black or other	34 (49.3)	35 (50.7)
Missing	138 (76.2)	43 (23.8)
Charlson score			<.0001
0	77 (53.1)	68 (46.9)
1	104 (59.8)	70 (40.2)
2+	314 (72.2)	121 (27.8)

Pearson chi-square tests were conducted to examine associations by gender, race, and treatment status. The percentage of high-risk women receiving treatment was significantly higher than the percentage of high-risk men receiving treatment (42% vs 11%; P < .0001). There were also differences in treatment found by race, among patients in whom race was known. The percentage of non-Caucasians receiving treatment (50.7%) was higher than the percentage of Caucasians (35.9%), P = .017. We also found an association between treatment group and Charlson score; those with higher comorbidity scores were found to be less likely to receive treatment (P < .0001). No significant treatment differences were found by age group. Rates for treatment ranged between 26% and 42%.

Physician Notes and Over-the-Counter Medications

Within the physician notes from visits that took place within the study time frame, we looked specifically for mention of calcium carbonate, calcium, multivitamins, and vitamin D. Of the nearly 2800 individuals in the study population, only 100 had notes containing any of the above-mentioned keywords. Of these, 56 were in the high-risk group and 44 from the average risk group. “Tums” was the term most often found. Of the 100 individuals identified in the keyword search, 33 were also on one of the prescribed bone health medications (25 from the high-risk group and 8 from the average risk).

Discussion

Currently, in the United States, approximately 10 million people have osteoporosis, and 34 million have osteopenia or low bone mass.¹⁶ One in 2 women and 1 in 4 men will have an osteoporosis-related fracture in their lifetime.¹⁷ In 2005, osteoporotic fractures cost more than $19 billion in direct costs of hospitalizations, nursing home admissions, and outpatient services, and these costs are projected to increase 50% by 2025.¹⁶ Thus, the problem is prevalent and costly.

There are well-established methods of reducing the risk of sustaining a fragility fracture. Treating a patient with osteoporosis with medication designed to decrease fracture incidence is one. Several studies have demonstrated their benefits. In women with osteoporosis, bisphosphonates reduce the incidence of hip fracture, distal radius fracture, and vertebral compression fracture.^18–22 Bisphosphonates have also been shown to decrease fracture incidence in individuals who have glucocorticoid-induced osteoporosis.^23,24 Selective estrogen receptor modulators have been shown to decrease vertebral compression fractures by 30% to 50%.^25,26 While calcitonin has not been shown to decrease the incidence of hip fracture, it has been shown to decrease pain caused as a result of vertebral compression fractures and to prevent future vertebral compression fractures.²⁷ In postmenopausal women, teriparatide has been shown to decrease vertebral fracture and nonvertebral fracture incidence in patients who have sustained a previous vertebral compression fracture.²⁸ Thus, in populations at risk of fragility fracture, treatment with osteoporosis medication can significantly decrease the risk of subsequent fracture.

Because we were interested in assessing care for osteoporosis in patients seen within our system, we characterized osteoporosis risk in this patient population. We used criteria for fragility fracture risk well documented in the literature.^3,29,30 Incorporating these criteria, we found approximately 27% of the home health population to be at high risk; and similar to the findings of others found only about one third of these individuals to be receiving treatment.^11,14 Our finding that high-risk women were more likely to be receiving treatment than high-risk men (42% vs 11%) was also consistent with the literature.

Unlike prior reports where treatment was more likely in younger patients, we found no difference in treatment by age.¹¹ Although we are uncertain as to the reason for this, it could be a reflection of a focus in our health system on geriatric care overall. Also, contrary to other studies, non-Caucasians were more likely to be receiving treatment.^11,31,32 The total number of non-Caucasians was not large, and we had missing racial data for 24% of the high-risk group. Nonetheless, even if all those with unknown racial data were Caucasian, the result would have shown equal treatment by race. This finding differs from what others have reported.^31–34

We also examined treatment by risk factor to determine whether differences in treatment rates would be found. A greater percentage of patients were treated if they had osteoporosis (57.9%) and prior fracture (33.3%) compared to those with other risk factors. Nonetheless, even for these diagnoses far more treatment is warranted.

We found that patients with higher comorbidity scores were less likely to receive treatment for fragility fracture. Patients within HHC require attention for a variety of conditions and treatment for osteoporosis may not be as compelling as treatment required for other conditions these individuals have. In our system, patients receiving HHC services often have multiple comorbid conditions including diabetes, cardiac disease, pulmonary disease, neuro diagnoses, and open wounds; approximately 44% of patients have orthopedic diagnoses. The fact that the greater the number of comorbidities, the less likely one is to receive treatment in our study and those of others^11,14 indicates that other issues may take precedence.

Limitations

This study was conducted in a single home health provider group, so these results may not be generalizable to multiple settings. However, the population studied was nearly 2900 individuals and our findings were generally consistent with the work of others. Another limitation was the paucity of information on over-the-counter medications taken by this population. It is quite likely that far more than 100 individuals were using various over-the-counter medications related to bone health. Still, most benefit will be from the treatment from prescribed products, and our data on these medications was very comprehensive. We presumed that patients not currently on treatment have never been treated. Patients may have been treated with osteoporosis medications prior to our study period but discontinued due to intolerance or treatment may have never been initiated due to the risk of side effects. Additionally, as the diagnosis of osteopenia or osteoporosis could be considered a proxy for having undergone a DEXA scan and, as all patients over the age of 65 should have had a DEXA scan, the lack of data regarding testing prevalence could also be considered a study limitation. It is also possible that we had some misclassification and that some considered “average risk” were not diagnosed. However, because the health system has comprehensive records on multiple diagnoses as well as comprehensive pharmacy information, our classifications should be fairly robust.

A final limitation is that we have analyzed patients' risk factors for fracture as though each diagnosis represented the same fracture risk. We are aware that risk for fragility fracture is not the same in a patient with osteopenia compared to risk in an individual who has been diagnosed with osteoporosis. By including diagnoses that arguably carry with them less risk of fragility fracture than does osteoporosis or prior fracture in our treatment assessment, we have magnified the undertreatment in this population. Nonetheless, only 58% of individuals with a known diagnosis of osteoporosis and 33% of those with a previous fracture were treated, conditions highly associated with re-fracture. While we did have the above-mentioned limitations, the comprehensive nature of our data and the health system’s opportunity for future intervention is a major strength.

Future Steps and Conclusions

This study was conducted in an environment ideal for the comprehensive collection of data. Relational databases allowed the team to link diagnostic codes to pharmacy records. Furthermore, focusing on the HHC environment allowed us to limit our attention to an arena where efforts to improve the quality of care for those at risk of fracture could be readily implemented. As we have demonstrated, there is great room for improvement. Understanding care provider attitudes toward bone densitometry and barriers to treatment would be worthy. Future measures could include electronic reminders to primary care physicians about a given patient’s risk status. Another possibility involves developing a multidisciplinary “fragility fracture risk assessment team” that would review all HHC patients over the age of 65, and for those at risk institute a further evaluation and treatment protocol (eg, prescription for a osteoporosis medication, home safety evaluation, fall prevention, and exercise instruction).

Consideration of reminders to the patient’s primary care provider who orders home health care could increase awareness of the patient’s vulnerability to future fracture. Where electronic medical records exist, it may be worthy to consider working with providers to build a relevant prompt to emphasize fragility fracture risk. Furthermore, examining ways to maximize nurse management of patients at risk of fragility fracture may enhance the benefits of pharmacotherapy. The role of the nurse in coordination of care and tracking of adherence to medication and behavior recommendations is an advantageous key component of HHC, which is often lacking in the traditional outpatient setting. Understanding care provider attitudes toward bone densitometry and barriers to treatment would also be worthy. This study has documented the need to do more in this system. It provided evidence that consideration of an intervention to improve care for those at risk of fracture is warranted.