Osteoporosis International

, Volume 14, Issue 9, pp 722–727

Mortality and morbidity associated with osteoporosis drug treatment following hip fracture

Authors

  • Marilyn W. Cree
    • Department of Mathematical and Statistical SciencesUniversity of Alberta
  • Angela G. Juby
    • Division of Geriatrics, Department of MedicineUniversity of Alberta
    • Department of Mathematical and Statistical SciencesUniversity of Alberta
    • 632 Central Academic Building, Department of Mathematical and Statistical SciencesUniversity of Alberta
Original Article

DOI: 10.1007/s00198-003-1430-3

Cite this article as:
Cree, M.W., Juby, A.G. & Carriere, K.C. Osteoporos Int (2003) 14: 722. doi:10.1007/s00198-003-1430-3

Abstract

This study examined post-fracture osteoporosis drug treatment in hip fracture patients and the association of treatment with mortality and morbidity. Pre- and post-fracture demographic/health information was collected on a cohort of hip fracture patients aged 65+ years. Post-fracture administrative data on prescription drug use and health care utilization was linked to the cohort data. Five classes of osteoporosis drugs were available during the study period: hormone replacement therapy (HRT), bisphosphonates (BSP), calcitonin, selective estrogen receptor modulators (SERMs) and vitamin D3 (Rocaltrol). Pre-fracture, 38 of 449 patients (8%) were on osteoporosis medications. Post-fracture, 81 of 356 patients (23%) were treated; 63 of these patients were untreated prior to fracture. Both treated and untreated patients had similar rates of subsequent hip fracture (6% and 4%, respectively) and Colles fracture (2%). Regardless of treatment status, patients were also equally likely to be hospitalized, both in the short-term (28% in treated, 27% in untreated) and in the long-term (43% versus 37%). However, mortality was significantly lower in the treated group. The lower mortality in the treated group, combined with the knowledge that antiresorptive drugs reduce fractures and increase bone density, merit undertaking a randomized trial to confirm our findings that antiresorptive therapy should be considered in all patients post-hip fracture.

Keywords

ElderlyHip fractureMorbidityMortalityOsteoporosisTreatment

Introduction

Osteoporosis affects one in four women and one in eight men over the age of 50 years and is often undiagnosed until a fragility fracture occurs. Most hip fractures are associated with osteoporosis [1] and increased mortality [2] and morbidity [3]. Quality of life is also compromised: following a hip fracture, 80% of women over age 75 report they would rather die than be institutionalized [4] and 40–70% of patients are so fearful of falling again that it interferes with their mobility and leads to depression and poor life satisfaction [5]. Unfortunately, these patients have good reason to be fearful because it is likely they will sustain another osteoporosis-related fragility fracture in the near future [6].

Despite the high prevalence and potentially devastating impact of osteoporosis, many of these patients are not being treated [6,7]. A possible reason for this low treatment rate in seniors is a perceived lack of evidence that these patients benefit from treatment.

This 5-year follow-up study was done to determine if patients were receiving osteoporosis treatment following hip fracture and whether this treatment was beneficial in reducing mortality and morbidity. In addition, because continuity of care is associated with quality of care [8], this study also investigates the association between continuity of care and osteoporosis therapy in patients after a hip fracture.

Materials and methods

Study cohort

This study involves a cohort of patients originally studied in 1996/1997 and followed until 2001. The methodology involved in the original cohort study is outlined elsewhere [9]. Briefly, the original cohort included all Edmonton-area residents aged 65+ years who were hospitalized with a hip fracture. Study participants were interviewed in the hospital during the post-fracture week to collect pre-fracture function, health perception, and social support information. Over the ensuing 3 months, mortality and institutionalization outcomes were recorded and a second interview was conducted by telephone to ascertain 3-month function, health perception, and social support. Medical records were reviewed to gather co-morbidity and discharge data, including discharge location. Forty-eight (8%) eligible adults chose not to participate. Administrative data collected by Alberta Health and Wellness was linked to 449 of the others (80%) by matching the dates of birth and hospital admission. This linkage made it possible to obtain various measures including prescription information. Administrative data could not be linked to the remaining members of the cohort because there was not an exact match between the birth and admission dates. The analysis was restricted to patients living in the community prior to their fracture, because nursing homes directly cover the drug costs for their residents, and so these patients do not appear in the Alberta Health and Wellness database. The 50 individuals residing in the community prior to their fracture but in a nursing home 3 months post-fracture remained in the analysis and were assumed to be untreated. All research was done with prior approval of the ethics committee.

Measures

Health and demographic information

Health and demographic information was collected on the original cohort [9], and a trained Alberta Health data specialist extracted the following administrative data for each of the 449 study participants. From the Alberta Physician Claims File, we calculated the referral-adjusted continuity of care, called SECON, for each patient during the fiscal year of the hip fracture and the fiscal year following the hip fracture. While there are various ways to measure continuity of care, this measure is recommended for use in patient populations with chronic illness [10]. SECON assumes values between 0 (low) and 1 (high) and indicates the patient's tendency to see the same physician. That is, a patient visiting the same physician or receiving a referral from this physician would be considered to have a high continuity of care (SECON of 1). A patient visiting different physicians at each sequential visit would have SECON of 0. For each study patient, we calculated the SECON for the fiscal year of the fracture and the fiscal year following the fracture.

Using the Alberta Population Registry Database, we were able to identify patients who were deceased by the end of the fiscal year following their fracture and those who had died by 31 March 2001, our study end date. The Alberta Morbidity database contained post-fracture hospitalization information. Data on emergency room visits were provided by the Alberta Ambulatory Care Classification System.

Prescription information

The Alberta Blue Cross database contains information on reimbursement of prescription drugs, with each claim containing information pertaining to one prescription. Aboriginal adults, comprising approximately 1% of Alberta's population aged 65+ years, are the only group of older Albertans not covered under the plan. We looked at all prescriptions filled by the 449 participants over 4 fiscal years (1995/1996 to 1998/1999). We were interested in two variables: the drug identification number to identify the drug prescribed and the date the prescription was filled. Based on the drug identification number, we selected drugs used to treat or prevent osteoporosis and classified them into one of five groups: hormones (premarin, C.E.S., ogen, estrace, estraderm, vivelle, provera, prometrium); bisphosphonates (didrocal, didronel, fosamax); calcitonin (calcimar, miacalcin); vitamin D3 (Rocaltrol, 1α-cholecalciferol); and selective estrogen receptor modulators (SERMs) (Raloxifene). Residronate (Actonel) was not available in Alberta during the time period of this study.

Outcomes data

The dependent variables of interest were treatment (treated or not), mortality (died or not), and morbidity as measured by hospitalization (hospitalized or not). Mortality data were available to 31 March 2001, and was assessed over two time intervals: at the end of 1 fiscal year post-fracture (short-term mortality) and by 31 March 2001 (long-term mortality). Hospitalization was available to 31 March 2000, and was evaluated over two time periods: at the end of 1 fiscal year post-fracture (short-term morbidity) and by 31 March 2000 (long-term morbidity).

Statistics

Chi-square tests examined the association between categorical variables and t-tests were used to compare means. Stepwise logistic regression was done to investigate the relationship between the independent variables and mortality (1-year and long-term), or post-fracture treatment. Age, sex, and any of the input variables significant at the 0.1 level were considered for inclusion in the treatment model. Age, sex, treatment, and any other significant independent variables were included in the mortality models. SAS (version 7, SAS Institute Inc., Cary, N.C., USA) was used for all the statistical analysis.

The following input variables were considered for inclusion because of their established influence on hip fracture outcomes. The socio-demographic variables included: age; gender; social support: patient's assessment of whether they have someone to rely on for help or not; years of education; and occupational prestige: "low" (unskilled manual), "medium" (semi-skill and skilled manual and clerical) or "high" (management, technicians, and professionals). Health-related variables included: functional decline: decline of more than ten points in Barthel score [11]; and number of co-morbid conditions.

Other independent variables of interest included: health perception: poor to fair or good to excellent; continuity of care: "low" (0–0.667) or "high" (>0.667 in either year); receipt of post-fracture homecare services (some or none), and discharge location. Discharge location was classified as "subacute", "rehabilitation hospital", "other rehabilitation facility", or "other non-rehabilitation setting" (home, nursing home, etc.). Post-fracture osteoporosis treatment was categorized in two ways: "some" or "none"; and according to the number of prescriptions filled: "none", "one to three", or "four or more".

Results

Pre-fracture prescriptions

Between 1995/1996 and 1998/1999, 690 prescriptions for hormones, bisphosphonates, calcitonin, and vitamin D3 (Rocaltrol) were filled by the cohort patients. Of the 449 hip fracture patients, 38 (34 women and four men) filled 283 prescriptions for hormones (189 prescriptions), bisphosphonates (82 prescriptions), calcitonin (two prescriptions), and vitamin D3 Rocaltrol (ten prescriptions) prior to their fracture. Almost all of the prescriptions filled by females were for hormones, while three of the males were receiving bisphonates, and the other calcitonin. Six of the treated patients (one male and five females) were admitted to nursing homes prior to their hip fracture. Twenty-nine patients, including the four males, had been taking these medications for more than 1 year prior to their fracture.

Post-fracture prescriptions

The post-fracture analyses focused on the 356 patients living in the community at the time of their fracture. Of 124 patients with a discharge diagnosis of osteoporosis, 34 received subsequent treatment and another 20 were in a nursing home at 3 months. Following their fracture, 81 patients (23% of the 356) filled 407 prescriptions for the drugs of interest (Table 1). Of these 81 patients, 74 were females, 63 had not received therapy prior to their fracture, and 30 filled their first prescription within 6 months of the fracture. Eleven female patients were receiving more than one of the three treatments, with seven receiving hormones and bisphosphonates, three getting bisphosphonates and calcitonin, and one obtaining all three drugs. Fifty of the 81 patients were still filling prescriptions more than 1.5 years post-fracture, the maximum time being about 2.5 years post-fracture.
Table 1.

Numbers of post-fracture prescriptions and patients on therapy. Ten patients (all females) received two different drugs at some point during the post-fracture period, one patient (also female) received three different drugs

Drug

Number of prescriptions (n=407)

Number of patients (n=356)

No. females

No. males

Total no.

No. females

No. males

Total no.

Hormones

110

0

110

12

0

12

Bisphosphonates

247

26

273

67

7

74

Calcitonin

16

0

16

6

0

6

Vitamin D3 (Rocaltrol)

8

0

8

1

0

1

Treatment changed post-fracture for four of the females on therapy both pre- and post-fracture (Table 2). Although all four females were on pre-fracture hormones, two discontinued hormones post-fracture and began taking bisphosphonates. The remaining two continued on hormones but added bisphosphonates (one patient) or calcitonin (one patient).
Table 2.

Pre- to post-fracture treatment

Pre-to post-fracture

Drug

Hormones

Bisphosphonates

Calcitonin

Vitamin D3 (Rocaltrol)

Same treatment pre-to-post (n=14)

Male

0

1

0

0

Female

6

9

0

1

Treatment change pre-to-post (n=4)

Male

0

0

0

0

Female

4*

3

1

0

Treatment post-fracture only (n=63)

Male

0

6

0

0

Female

2

55

5

0

aTwo of these females remained on hormones post-fracture

Among the 63 patients beginning treatment post-fracture, bisphosphonates were most commonly prescribed (Table 2), sometimes in combination with hormones (two females). Three of the patients taking bisphosphonates tried calcitonin at some point during the post-fracture period.

Univariate associations

Selected socio-demographic characteristics and post-fracture information on treated patients and all patients are shown in Table 3. The socio-demographic characteristics presented in Table 3 include age, sex, and any variables significantly associated (at the 0.05 level) with treatment. Following their hip fracture, 25 (7%) patients were readmitted to the emergency room within the study period with another fracture: 17 with a hip fracture, seven with a Colles fracture, and one patient with both fractures. Eighteen (72%) patients, including the patient with both fractures, were untreated. Treatment in one of the patients may have started as a result of the second fracture because both the fracture and the initiation of treatment occurred in the same week.
Table 3.

Frequency of patients with selected characteristics by post-fracture osteoporosis treatment

Characteristic

All patients

Patients receiving treatment (% of total)

Number

356

81 (23)

Gender

Female

263

74 (28)

Male

93

7 (8)

Age

<75

82

23 (28)

75–84

155

45 (29)

>84

119

13 (11)

Education

Less than grade 10

128

24 (19)

Grade 10 or more

170

53 (31)

Medical conditions

Two or less

196

59 (30)

Three or more

160

22 (14)

Functional decline

None

185

59 (32)

Some

99

18 (18)

Continuity of care

Low

288

49 (22)

High

128

32 (25)

Subsequent fracture

Hip

17

5 (29)

Colles

7

2 (29)

Both

1

0

None

331

74 (22)

Non-rehabilitation setting

72

10 (14)

Rehabilitation facility

284

71 (25)

Rehabilitation hospital

123

18 (15)

Subacute

133

48 (36)

Other

28

5 (18)

Homecare within 3 months

None

63

5 (8)

Some

293

76 (26)

Three-month mortality

Died

21

0

Survived

335

81 (24)

One-year mortality

Died

82

3 (4)

Survived

274

78 (28)

Long term mortality

Died

158

14 (9)

Survived

198

67 (34)

One-year hospitalization

Hospitalized

98

23 (23)

Not hospitalized

238

58 (25)

Long-term hospitalization

Hospitalized

137

35 (26)

Not hospitalized

219

46 (21)

Over the years following the fracture, treated and untreated patients were equally likely to be hospitalized, both in the short-term (28% in treated, 27% in untreated) and in the long-term (43% versus 37%), and consequently we did not explore this further in a logistic regression analysis. However, the association between treatment and mortality was explored in more detail in the logistic regression analysis because of the reduced mortality in treated patients.

Adjusting for covariates

Although all the covariates were considered for inclusion in both the treatment and mortality models, only the significant variables are presented in Tables 4 and 5.
Table 4.

Logistic regression results to predict post-fracture osteoporosis treatment

Variable

Post-fracture osteoporosis treatment

OR

95% CI

Female gender

5.79*

2.49, 13.49

At most two comorbidities

2.38*

1.34, 4.24

Age (compared to 85+ years)

65–74

4.10*

1.84, 9.13

75–84

3.50*

1.74, 7.06

Some post-fracture homecare

3.82*

1.43, 10.21

*Significant at 0.05 level

Table 5.

Logistic regression results to predict long-term mortality

Variable

One-year mortality

Long-term mortality

OR

95% CI

OR

95% CI

Male gender

3.56*

1.59, 7.98

2.58**

1.35, 4.95

Age (compared to 85+ years)

65–74

0.56

0.17, 1.84

0.58

0.26, 1.28

75–84

0.90

0.39, 2.07

1.19

0.64, 2.20

Three or more comorbidities

2.72*

1.24, 5.96

2.62*

1.52, 4.50

Some functional decline

2.53*

1.15, 5.57

1.93*

1.10, 3.39

Some post-fracture homecare

NS

2.61*

1.09, 6.27

Some osteoporosis treatment

0.25**

0.06, 1.12

0.34*

0.17, 0.70

*Significant at the 0.05 level

**P-value is 0.071

Treatment results are outlined in Table 4. Therapy was much more likely for females (OR=5.79) than males and for patients aged 65–74 or 75–84 (OR=4.1 and OR=3.5, respectively) compared to older patients. Having few medical conditions and receiving some post-fracture home care services also increased chances of treatment approximately 3-fold.

The results of the mortality analyses (both 1-year and long-term) yielded similar results for most of the covariates (Table 5). That is, risk was approximately tripled for males and patients with three or more comorbidities; and roughly doubled for those experiencing functional decline. Risk of long-term mortality was also approximately double for individuals needing some post-fracture homecare. Although short- and long-term mortality was reduced for treated patients, the number of osteoporosis medication prescriptions filled was unimportant.

Discussion

This is the first long- term study to evaluate osteoporosis treatment post-hip fracture as a major outcome in a "real world" setting (i.e. a non placebo-controlled trial or care map setting). It shows the extremely low usage of anti-resorptive therapy (23%), which is disappointing in view of the number of choices of drugs now available, from multiple drug classes, and with multiple modes of administration. It seems that lack of compatibility/tolerance to treatment does not seem to account for the under-treatment as we were able to track all prescriptions filled, and in only one patient were prescriptions written for HRT, bisphosphonates and calcitonin. Continued prescription renewals occurred in 68% of the treated group up to 2.5 years post-fracture. In spite of the fact that not all treated patients continued therapy long term, the treatment was still associated with reduced risk of mortality.

Previous studies have shown the very low prevalence of osteoporosis treatment following a hip or other fragility fracture [6,7,12]. Treatment was not initiated in either the acute care or the rehabilitation setting [7]. Because some experts postulate the lack of a consistent primary care physician to be one barrier to treatment, we explored whether treatment was more likely in patients under the consistent care of the same family practitioner (high continuity of care as measured by SECON). We found no difference in treatment rates of our patients associated with continuity of care: 22% with "low" care and 25% with "high" care were treated. We also found that the treated patients received more homecare support than their untreated peers (94% versus 79%), leading us to postulate that perhaps homecare represents a different level of overall care from that of the primary care physician and that homecare may enhance the chance of receiving osteoporosis therapy.

Overall, the rate of subsequent fracture was only 7%: 6% (treated group) and 4% (untreated group) for hip fractures and 2% for Colles fractures. Although this re-fracture rate is too small to permit an analysis adjusting for other determinants of treatment, our data suggest a possible lack of efficacy of treatment on subsequent fractures. The apparent lack of benefit from treatment may be due to several factors: the multifactorial nature of hip fracture etiology; the short duration of therapy in some of the patients in the treated group; the higher use of homecare which may reflect less familial/spousal caregiver support increasing the chance/need for potentially risky behavior; the "uncontrolled" nature of medication use in the community; and the use of calcium and vitamin D supplementation which could not be measured in this study.

The debate has long raged about the utility of treating hip fracture patients in terms of preventing subsequent hip fracture and reducing morbidity and mortality. Studies investigating the utility of treatment with respect to preventing the first hip fracture are now published. For example, McClung and colleagues evaluated the impact of residronate treatment on high-risk older women with and without osteoporosis as measured by bone mineral density. Although treatment had no effect on women without osteoporosis, primary hip fractures were reduced by 28% in osteoporotic women [13]. However, no investigation of the association between residronate treatment and morbidity or mortality was undertaken. Similarly, the FIT trial with alendronate [14] revealed a 51% reduction in new hip fractures among women with osteoporosis, but not necessarily a previous hip fracture. Our study differed from the residronate and FIT trials by evaluating treatment benefits to those previously sustaining a hip fracture.

It is well known that sustaining an osteoporotic fracture increases mortality [2], but we are unaware of any long-term studies assessing the association of osteoporosis treatment on mortality. Our long-term mortality rate of 44% is similar to the 6-year rate of 47% reported by Dirschl and colleagues [12], although they did not study the impact of treatment on mortality, perhaps because their treatment rate was so low (two patients on HRT, two on vitamin D supplements, and 11 on calcium supplements) and their 6-year follow-up group of 21 patients was too small. Magaziner and colleagues [15] followed patients for 2 years after hip fracture and found a mortality rate of 16.8% in the first year, and 25.3% in the second year, but did not comment on the percentage of patients on osteoporosis treatment or the impact of this on mortality. March et al. [16] found no effect of a care pathway on hip fracture mortality in a four-month study, but their pathway did not include any specific osteoporosis therapy.

The present study has shown a statistically significant decrease in mortality in the treated versus the untreated group in spite of the small numbers of treated patients. It could be argued that patients were untreated because they died, but 131 surviving patients were also untreated. It could also be that the treated group was more compliant and health-oriented, although this was not reflected in reduced morbidity. Our treatment group comprised predominantly women, those aged 75–84 years, and the majority had at least 10 years of formal education. These factors, and not osteoporosis treatment, could account for the mortality rate differences. However we adjusted for these characteristics in our analysis. Despite our efforts, we acknowledge that other unmeasured variables, related to both treatment and mortality, may be responsible for the observed association. Perhaps the association is simply attributable to the fact that the sickest patients, or those with dementia, are more likely to die and less likely to be treated. Further research is needed to increase our understanding of the association between mortality and treatment.

Our study had some limitations. We had a high record linkage failure rate (20%), which we attribute to errors in recording and entering dates; this should, however, be unrelated to the patient characteristics and unlikely to bias our results. The lack of data on calcium and vitamin D supplementation reflects the problems with monitoring the use of non-prescription medications. However, no studies have proved calcium and vitamin D will prevent subsequent fractures in a cohort of hip fracture patients that is similar to ours, although the benefits of antiresorptive therapy have been shown [13,14]. We also expect calcium and vitamin D usage would have been low, based on a previous study looking at a similar patient group [7]: calcium use was less than 10% and vitamin D supplementation less than 5% at the time of discharge from rehabilitation. There are concerns regarding the validity of the Blue Cross Drug data, although our low treatment rates concur with findings of other studies [6]. Finally, because nursing homes pay for the drugs consumed by their residents, treatment information for our patients in long-term care was unavailable. We know 14% (50 patients) of our patients were in long-term care 3 months post-fracture and expect this proportion would approach roughly 35% by 6 months post-fracture [17]. We assume only a small proportion of our patients in nursing homes would be treated. Only 30% of nursing home physicians would treat patients with a recent fracture and only 7–8% of patients with confirmed osteoporosis would be prescribed bisphosphonates or HRT [18]. Based on this, we included the 50 long-term care patients in our analysis and assumed they would not be treated after admission.

In conclusion, our study highlights the continuing care gap in osteoporosis treatment post hip fracture, despite the numerous proven therapeutic options and guidelines on osteoporosis definition and treatment [19,20]. We know that practice patterns may not change even when guidelines indicate hip fracture patients should be treated [21]. The overall treatment rate was low in our study, but males were even less likely to be treated than females (9% versus 91%, respectively), although they have higher mortality from hip fracture [7,8]. All hip fracture patients are hospitalized, placing them in contact with many health care professionals and providing ample opportunity for practitioners to diagnose and treat osteoporosis to prevent subsequent fragility fractures and decline in health. Now that we have shown a potential relationship between osteoporosis treatment post-hip fracture and reduced mortality, combined with the knowledge that antiresorptive therapy increases bone density and reduces fractures, we recommend undertaking a randomized trial for future study. Such a randomized trial will strengthen our findings in favor of treating all patients, regardless of age, gender or health status.

Acknowledgements

This research was funded by the Alberta Heritage Foundation for Medical Research.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2003