Osteoporosis International

, Volume 16, Issue 3, pp 296–301

Follow-up treatment for osteoporosis after fracture

Authors

    • School of Public Health & Health SciencesUniversity of Massachusetts at Amherst
    • Center for Health Policy and ResearchUniversity of Massachusetts Medical School
  • Stephen H. Gehlbach
    • School of Public Health & Health SciencesUniversity of Massachusetts at Amherst
  • Penelope Pekow
    • School of Public Health & Health SciencesUniversity of Massachusetts at Amherst
  • Elizabeth Bertone
    • School of Public Health & Health SciencesUniversity of Massachusetts at Amherst
  • Evan Benjamin
    • Baystate Medical Center
Original Article

DOI: 10.1007/s00198-004-1676-4

Cite this article as:
Hooven, F., Gehlbach, S.H., Pekow, P. et al. Osteoporos Int (2005) 16: 296. doi:10.1007/s00198-004-1676-4

Abstract

Studies of the management of osteoporosis in older women who have had hip or wrist fractures have found underdiagnosis and undertreatment of the disease. Few such studies have been conducted in the United States, however, and most studies have been confined to a subset of the treatments currently available to treat osteoporosis. Mail surveys were sent to 381 women between 50 and 84 years of age who had been treated for a hip or wrist fracture at a large northeast US teaching hospital between October 1, 1998, and September 30, 2000. These surveys included questions about osteoporosis risk factors and physician treatment both before and after the index fracture. Of 381 surveys mailed, 70 were returned because of an invalid address or by a relative because a patient was deceased. Of the remaining 311 surveys, 147 completed responses were received. Fifty-two percent of respondents reported having received either a prescription or a recommendation for a nonprescription medication used to treat osteoporosis before the fracture. After fracture, 60% of subjects were advised to take any osteoporosis medication, and 42% of were advised to take a prescription medication. Of women reporting no treatment advice before fracture, 33% reported treatment after. Twenty-four percent of patients reported a change in treatment after fracture versus before. No significant differences in treatment were found according to fracture history, maternal history of fracture, or maternal history of osteoporosis. Both prescription and nonprescription treatment prevalence after fracture were lower than expected, and there was only a small change in reported treatment prevalence after fracture versus before. There was also little difference in treatment prevalence based on risk factors for osteoporosis or osteoporotic fractures. A sizeable opportunity exists for intervention to reduce the risk of osteoporotic fractures for patients who have a history of fracture.

Keywords

Follow-up studiesHip fracturesOsteoporosisOsteoporosis diagnosisOsteoporosis epidemiologyOsteoporosis prevention and controlPhysicians’ practice patternsWrist fractures

Introduction

Treatment prevalence after fractures

An estimated 30% of US women 65 years of age and older suffer from osteoporosis [1, 2]. The consequences of fractures resulting from osteoporosis are substantial in terms of increased mortality, reduced quality of life, and cost [2]. The approximately 300,000 hip fractures that occur annually in the United States may account for about 31,000 excess deaths within 6 months of the fracture [3].

Several medications have been approved for the treatment of osteoporosis in postmenopausal women with low bone density that have been shown to maintain or increase bone density, and/or to reduce the risk of hip and wrist fractures [4]. These treatments include hormone replacement therapy (HRT), bisphosphonates, calcitonin, and raloxifene. Treatment guidelines also recommend that calcium and vitamin D supplements be included with any of these treatments to insure maximum effectiveness, though calcium and vitamin D on their own have been less consistently shown to reduce the risk of fracture or increase bone mass in postmenopausal women [5, 6].

Osteoporosis itself increases the risk of fracture, but prior fractures of the wrist, hip, or spine increase the risk of fracture for a woman above the risk attributable to low bone mass alone [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21]. The evidence to date shows that only 20–40% of older women who have experienced one of these fractures are being treated for osteoporosis [22, 23, 24, 25, 26, 27, 28, 29, 30, 31]. Few studies, however, have assessed the full range of options available to a physician when managing osteoporosis, and only two explicitly evaluate whether treatment initiated before fracture changes after fracture [27, 31].

This study assesses the patterns of screening, diagnosis, and treatment for osteoporosis in a northeast US metropolitan area, among women age 50 to 84 years who have a history of wrist/forearm or hip fractures. We evaluate the full range of prescription and nonprescription treatments available, as well as change in treatment after the fracture.

Methodology

Population

Patients eligible for this study were women between the ages of 50 and 84 years who had been treated for a fracture of the hip or wrist at Baystate Medical Center in Springfield, Massachusetts, between October 1, 1998, and September 30, 2000. We identified women with primary diagnoses of fracture of the hip or wrist using diagnostic codes recorded in the hospital’s administrative database (International Classification of Diseases, 9th Revision, codes 813, 733.12, 733.14, 820.0, 820.2, 820.8). We excluded any patient from the study group if her questionnaire response indicated that her fracture had been the result of a major trauma, such as a car accident or a fall from greater than standing height.

Data collection

We mailed a survey to these subjects that consisted of questions pertaining to treatment of osteoporosis, history of fractures, maternal osteoporosis and fracture history, menopausal status, education, race, and smoking status. The treatment section included questions on the four primary forms of prescription treatment (bisphosphonates, estrogens, calcitonin, and raloxifene), as well as nonprescription treatment with calcium and vitamin D. The treatment questions followed the format: “Has a physician or other healthcare provider ever prescribed [name of medication] for you?” Respondents were given the following response options to the treatment questions: “Yes, before my fracture only,” “Yes, after my fracture only,” “Yes, both before and after my fracture,” “No,” and “Can’t recall.”

We mailed the surveys between July and September 2001. If we received no response to the first mailing after approximately 6 weeks, we mailed a second survey. After another 6 weeks had passed, we mailed a third survey for those who had not responded to the first two.

Data analysis

We obtained age and fracture type from the administrative data. We assessed the significance of differences in frequency distributions of respondents by age group and fracture type using the Pearson χ-square test.

We described the respondent population using frequencies and proportions for the following variables obtained from questionnaire responses and administrative data: age group, fracture type, race, education, marital status, maternal history of osteoporosis, maternal history of fracture, current smoking status, ever smoked, living alone, fracture prior to the index fracture, and fracture since the index fracture.

By combining responses to each of the treatment questions we created two treatment outcome variables: “any treatment after fracture” and “prescription treatment after fracture.” We classified the outcomes as “yes,” “no,” or “can’t recall.” In a similar fashion we created independent variables for prescription treatment and any treatment before fracture.

Other independent variables derived from the questionnaire, which we used to assess potential associations with each of the outcome variables, were education, race, maternal history of fracture, maternal history of osteoporosis, ever smoked (ever having smoked at least 100 cigarettes), or currently smoking. We categorized age as 50–64, 65–74, and 75 −84 years after an examination of the age distribution, to assure an adequate number of observations in each category.

We calculated frequencies and proportions for prescription treatment after fracture and any treatment after fracture, and stratified these outcomes by each of the categorical independent variables, including age category, smoking status, prior fracture history, and education. We used χ-square tests for independence to assess the significance of deviations of observed from expected values in the resulting contingency tables. To test for the significance of change in prevalence of treatment before and after fracture we used the McNemar test of agreement.

We created an additional outcome variable: “change in treatment after fracture,” to reflect not only the prevalence of treatment after the fracture, but also whether treatment had been changed after fracture in comparison with treatment status before. This variable describes only positive changes in treatment, meaning that a change from any treatment before fracture to no treatment after was coded as “no change in treatment.” Negative changes were excluded because we were interested in evaluating only the addition of a new treatment, or a decision to change a previously prescribed medication.

Results

Three hundred eighty-one fracture patients met the criteria for inclusion in the study sample. Of 381 surveys mailed, 70 surveys were returned due to an incorrect address, either by the postal service or after having been opened, usually with a note indicating that the subject was deceased or institutionalized. Of the remaining 311 surveys, we received 147 responses, for a response rate of 47%. This rate was lower for 75–84 year olds (44%), than for the 50–64 year olds (49%), and 65–74 year olds (51%). The proportion of respondents within fracture type was lower for hip fracture (46%) than for wrist fracture patients (51%).

Of the final study sample, almost half of the study subjects were 75 years old and above (48%) (Table 1). The majority were white (87%), and graduates of high school or college (70%). Thirty-eight percent of subjects were married. Few reported a maternal history of osteoporosis (7%), and slightly more reported a maternal fracture (12%). Current smokers accounted for only 10% of respondents, though 48% reported having ever smoked at least 100 cigarettes.
Table 1

Characteristics of survey respondents (N=147; missing data in some categories account for totals of less than 147)

Population characteristic

Frequency

Percentage

Age, years

  50–64

35

24

  65–74

31

21

  75–84

71

48

Fracture type

  Hip

73

50

  Wrist

74

50

Race

  White/non-Hispanic

127

87

  Other

9

6

Education

  Primary school or some high school

33

22

  High school graduate

79

54

  College graduate

23

16

Marital status

  Married

55

38

  Divorced, separated, single

26

18

  Widowed

53

36

Maternal history of osteoporosis

10

7

Maternal history of fracture

17

12

Currently smokes

14

10

Ever smoked

71

48

Lives alone

48

33

Previous fracture

39

27

Fracture since index fracture

15

10

Slightly more than half (52%) of respondents reported being advised by a physician to take either a prescription or nonprescription medication used to treat osteoporosis at some point prior to the fracture (Table 2). Of these, 28% reported being advised to take a prescription medication, 21% to use HRT, 9% to take bisphosphonates, and 18% to take calcium/vitamin D at some point before the fracture.
Table 2

Advised to take medication before fracture or after fracture, and advised to add a new medication after fracture

Before fracture

After fracture

P Valuea

New medication

N

%

N

%

N

%b

Any medication

77

52

88

60

0.06

Any prescription medication

41

28

62

42

<0.01

35c

24

Medication type

  HRT

31

21

21

14

0.04

7

7

  Bisphosphonates

13

9

33

22

<0.01

24

18

  Raloxifene

3

2

9

6

0.02

6

5

  Calcitonin

6

4

17

12

0.02

15

12

  Calcium/Vit. D

27

18

65

44

<0.01

  Other medication for osteoporosis

7

5

17

12

0.05

Diagnosed with osteoporosis

33

22

43

29

<0.01

Had BMD testing

46

31

56

38

<0.01

aMcNemar test of agreement

bProportion of population who had not been advised to take the medication before the fracture

cThe total for the numbers for specific medications exceeds 35 due to cases in which more than one prescription was added

Sixty percent of subjects were advised to take either a prescription medication or calcium/vitamin D to treat osteoporosis after fracture (Table 2), an 8% increase compared with before fracture. Only 42% of patients received a prescription after fracture, an increase of 14% compared with before fracture (28%). Prescriptions for HRT were lower after fracture than before (14% after, versus 21% before), while bisphosphonate prescriptions increased (22% after, versus 9% before).

Twenty-nine percent of participants were diagnosed with osteoporosis after their fracture, compared with 22% before. Thirty-eight percent of respondents reported having a BMD exam after the fracture, compared with 31% before.

Twenty-four percent of subjects reported a prescription to add a new medication after fracture (Table 2). Bisphosphonates were the most common (18%), compared with calcitonin (12%), HRT (7%), and raloxifene (5%). Of 70 women who reported no treatment of any kind before fracture, only 23 (33%) reported treatment after (p=0.063). Among the 106 women who had no prescription medication before fracture, the proportion reporting a prescription after was also 33% (p=0.003).

There was some indication that age might influence change in medication, as a significantly higher percentage of those 75 and over had a change in prescription after fracture (37%), compared with subjects under age 75 (21%, p=0.048).

No significant differences were found in prevalence of treatment according to prior fracture, fracture after the index event, maternal history of fracture or maternal history of osteoporosis, for either prescription or for any treatment (Table 3). However, the number of women not reporting this information was high (n=57) and the number reporting maternal fracture or maternal osteoporosis was low (n=17, n=10, respectively).
Table 3

Treatment after fracture according to age, education, and risk factors

N

Any medication after

Any prescription after

N

%

N

%

All

147

88

60

62

42

Age group

  50–64

35

22

63

14

40

  65–74

31

17

55

11

35

  75–84

71

45

63

35

49

Education

  Primary or some high school

33

19

58

13

39

  High school graduate

79

52

66

38

48

  College graduate

23

11

48

7

30

Ever smoked

71

46

65

30

42

Never smoked

67

38

57

29

43

Currently smokes

14

8

57

5

36

Prior fracture

39

28

72

21

54

Fracture after index fracture

15

12

80

9

60

Maternal fracture

17

11

65

11

65

Maternal osteoporosis

10

6

67

6

67

Diagnosed with osteoporosis before fracture

23

21

91*

20

87*

*p<0.05

Those reporting osteoporosis diagnosis before the fracture had the highest prevalence of treatment according to either treatment definition (91% for any treatment, 87% for prescription treatment), and this difference was significant at p<0.05.

Discussion

We observed that only 42% of eligible patients were prescribed medication to treat osteoporosis after a hip or wrist fracture. The prevalence of treatment after fracture was even lower among women who reported no treatment prior to their fracture (33%). We also found little difference in treatment prevalence based on risk factors for osteoporotic fractures. In addition, the number of patients reporting any type of treatment after fracture compared with before increased by only 8%.

These mail survey results are consistent with the results of previous studies that used questionnaires, telephone interview, medical record review, or insurance claims data to determine the prevalence of treatment for osteoporosis after fracture. A pharmacy database analysis by Castel et al. showed 62% of inpatient and 70% of outpatients (n=183) at an Israeli hospital received no osteoporosis medication 6 months after fracture [27]. Kamel et al. found documentation in records indicating possible osteoporosis for less than 10% of hip fracture patients (n=170), at a Nassau county, New York hospital [28]. In a retrospective cohort study of 1,162 female distal radius fracture patients 55 years of age and older, only 23% of subjects received treatment with a prescription osteoporosis medication within 6 months after the fracture [31].

Cross-sectional studies using questionnaires or telephone interviews found similarly low rates of treatment. Only 38% of 112 women 40 years and older surveyed by Khan et al. in Alberta, Canada, were being treated with either bisphosphonates or hormone replacement therapy 6 months to 3 years after low-trauma hip fracture [22]. In a cross-sectional telephone survey of post–fragility fracture patients ages 18 years and older from three Toronto area hospitals (n=108), Hajcsar et al. found under 20% of patients had been treated for osteoporosis within 1 year of the fracture [29]. Pal surveyed 82 postmenopausal female hip fracture patients who had been hospitalized at a single hospital in Manchester, England, in 1996, 6 months after fracture [25]. Only 13 reported a prescription for HRT, and 12 for bisphosphonates.

The prevalence of treatment after fracture in this study (41% for prescription medication, 59% for any medication) was somewhat higher than in these other studies. This could be partially explained by the later time period of our survey, which took place 3–5 years after the approval of oral bisphosphonates for the treatment of osteoporosis in the United States and Canada (1995), whereas for previous studies, these drugs had been available for only 1–3 years. Differences in prevalence may also be accounted for by inclusion of younger fracture patients [29] and males [27] in previous studies, whereas our study included only older women. Studies using medical record and/or claims review of hospital or clinic patients also found lower prevalence of treatment (<10%–30%) [27, 28, 29, 31], a possible effect of the underreporting of osteoporosis treatment in hospital and physician medical records compared with self-report.

Differences in the respondent and nonrespondent population could also have affected our estimates. We could have overestimated treatment prevalence if the respondent population was overly representative of women who were more likely to attend to health care needs than the nonrespondent population. The use of self-report of diagnosis and treatment may also be considered a potential weakness, as self-report may be more prone to reporting bias than data derived from medical records or claims. It is also possible that a follow-up survey more accurately ascertains diagnosis and treatment prevalence than do claims or medical record review. Documentation in medical records can be inconsistent, and claims data does not take into account nonprescription treatment [32, 33, 34]. The correspondence between physician records and self-report of prescription medication has been found to be reasonably strong for some medications, and varies based on the type of medication, and has not been assessed explicitly for all of the treatments described in this study [34, 35]. Still, it is important to note that if the treatment prevalence is overestimated in this study, it does not alter our main finding, that women at high risk of future fractures are undertreated for osteoporosis.

Treatment could have been underestimated if those who did not respond to the survey were more likely to have been treated than nonrespondents. If hip fracture patients, or older subjects, whose response rates were the lowest, were more likely than other subjects to have been treated after their fracture, then our estimates of prevalence may be low. Still, even if prescription treatment prevalence for those women was as high as 80% after the fracture, the weighted average prevalence for the entire sample would have been only 65%.

One clinically based explanation for a low prevalence of treatment is that physicians tend to restrict treatment to those at highest risk of a future fracture. In such a case we would expect to see a higher prevalence of treatment among those with risk factors such as maternal history of fracture, maternal history of osteoporosis, prior fracture, or smoking. While our study sample may have been too small to detect a significant difference in treatment according to these risk factors, it is notable that no trend was observed toward higher prevalence of treatment for women with any of these risk factors.

By comparing treatment after fracture to treatment before the fracture we attempted to assess the degree to which physicians respond to a sentinel event (fracture) by changing treatment recommendation to the patient. Some physicians may have decided for clinically sound reasons not to alter treatment after osteoporotic fracture, but given that the most common treatment recommendations before fracture were HRT (21%), and calcium/vitamin D (19%), we might have expected to see more patients with a change in treatment (28% for prescriptions, 34% for all medications). HRT is prescribed to treat conditions other than osteoporosis, and calcium/vitamin D is not considered as effective as prescription medications in reducing fracture risk or increasing bone mass [4]. The low proportion of patients for whom a change in treatment was made, therefore, suggests a sizeable opportunity for intervention to reduce risk of subsequent fractures. One recent study has demonstrated that a significant increase in treatment is possible after a community education initiative, though overall treatment prevalence at discharge was still lower than that found in our survey [36].

While scientific advances are made through both basic and clinical research, the quick and consistent application of this knowledge to all patients is problematic. For osteoporosis treatment, the issue of slow translation of research into practice seems to impair the most up-to-date treatments from reaching patients.

Acknowledgements

The authors would like to acknowledge the assistance of the following persons in completing this study: Janet Kellogg, Ann DeRoode, Eleanor Vanetzian, and the Department of Healthcare Quality, Baystate Health Systems. Disclosure: Partial funding for this study was provided under a grant from Procter & Gamble Pharmaceuticals.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2004