Osteoporosis International

, Volume 17, Issue 3, pp 348–354

Risk of new clinical fractures within 2 years following a fracture

Authors

    • Department of Trauma SurgeryAcademic Hospital Maastricht
  • Jochen Cals
    • Department of Trauma SurgeryAcademic Hospital Maastricht
  • Fons Kessels
    • KEMTA (Klinische Epidemiologie en Medical Technology Assessment)
  • Peter Brink
    • Department of Trauma SurgeryAcademic Hospital Maastricht
  • Geert Jan Dinant
    • Department of General PracticeUniversity Maastricht
  • Piet Geusens
    • Department of Internal MedicineAcademic Hospital Maastricht
    • Limburg University Center
Original Article

DOI: 10.1007/s00198-005-2026-x

Cite this article as:
van Helden, S., Cals, J., Kessels, F. et al. Osteoporos Int (2006) 17: 348. doi:10.1007/s00198-005-2026-x

Abstract

Introduction:Clinical fractures are associated with an increased relative risk of future fractures, but the absolute risk and timing of new clinical fractures immediately after a clinical fracture have not been reported extensively. The study objective was to determine the absolute risk of subsequent clinical fractures within 2 years after a clinical fracture. Methods:We analyzed clinical fracture data from a university hospital recruiting all fractures in the area between January 1999 and December 2001. Subjects were 2,419 male and female patients aged 50 years and older, with a total of 2,575 fractures. There were 139 patients with more than one simultaneous fracture. Mean age was 66 years for males and 72 for females. Results:The cumulative incidence of patients with new clinical fractures over 2 years was 10.8% (262/2,419). In the 262 patients with subsequent fractures, we observed a higher mean age, more females and more often multiple baseline fractures compared with the 2,157 patients without subsequent fractures. Kaplan-Meier analysis indicated that age, gender and having multiple baseline fractures contributed significantly to cumulative new fracture incidence. Cox regression showed that these variables independently contributed to a higher subsequent fracture incidence. New fracture incidence was higher with increasing age ( p <0.001; hazard ratio [HR] 1.2 per decade; confidence interval [CI] 1.1–1.3). Females had a new fracture incidence of 12.2% compared with 7.4% in males ( p =0.015; HR 1.5; CI 1.1–2.0). Patients with multiple baseline fractures had a new fracture incidence of 17.3% compared with 10.4% for subjects with one baseline fracture ( p =0.006; HR 1.8; CI 1.2–2.7). Of all clinical fractures occurring within 2 years after a clinical fracture, 60% occurred during the first year and 40% during the second year ( p =0.005). The absolute risk to develop an incident clinical fracture within 2 years after any clinical fracture was 10.8%. Increased age, female gender and the presence of multiple simultaneous fractures at baseline each independently increased the risk of incident fracture. Significantly more fractures occurred in the first year following the index fracture than in the second year. Conclusion: Altogether, these data support the need for early prevention of future fracture among individuals with a fracture after age 50, using interventions which have been shown to have a rapid anti-fracture benefit.

Keywords

FractureIncidenceTiming

Introduction

Several studies indicate that having a fracture during adulthood is associated with an increased relative risk of developing a subsequent fracture, both in men and in women [14]. Klotzbuecher showed in an extensive meta-analysis a significant increased relative risk of 1.7–2.6 for any subsequent fractures among people with a history of any fracture. No absolute fracture risk calculation was given, nor information about the timing of the next fracture. Johnell et al. found the relative risk in their population to be the highest during the first years after the event [5].

Studies on the short-term absolute fracture risk and timing of the subsequent fractures during the first years after a fracture are scarce. Prior morphometric vertebral fractures have a striking effect on future risk of new morphometric vertebral fractures within 1 year, with an absolute risk of 20% for new vertebral fractures (25% in the presence of >2 vertebral fractures at baseline and 25% for any fracture within 1 year) [6]. In Johnell’s study, the absolute risk estimate was given for patients who developed a fracture of the hip, forearm, spine or shoulder as first and second fracture, and the incidence was 12.0% ( n =230/1,918) during 2-year follow-up.

There is a tendency to progress from relative risk calculation to the 5-year or 10-year absolute risk evaluation in the individual patient, in order to recognize patients at highest risk for fractures within a short time [7]. We therefore investigated the incidence and timing of subsequent clinical fractures within 2 years after a clinical fracture in men and women aged over 50 years.

Materials and methods

All patients aged over 50 years with a clinical fracture, who were treated in a European level-one trauma center between January 1999 and December 2001, were included. Fractures were confirmed on X-ray. Patients involved in a car traffic accident and those with pathological fractures were excluded, as well as patients not currently living either in the Netherlands or in the Belgian border area in close proximity to Maastricht.

A fracture questionnaire was sent to all survivors. Self-reported fractures were confirmed by review of the radiography reports in the hospital database. Self-reported fractures treated in other hospitals could not be verified with radiography reports.

Additionally, the nonresponders and deceased patients were individually entered into the hospital database, and subsequent fractures were investigated by studying the radiography reports in the 2 years after the baseline fracture until April 2004.

Clinical fractures were classified into 15 categories, according to the International Classification of Disease (ICD-9): skull (800–804), vertebra (805–806), clavicle (810), thorax (807, 811), pelvis (808), humerus (812), forearm (813), wrist (813), hand (814–817), hip (820), femur (821), patella (822), tibia/fibula (823), ankle (824), foot (825–826). The 15 categories were subsequently divided over five large groups for further analysis: humerus, radius/ulna, hip, other fractures and multiple fractures. Fractures at the classic osteoporotic sites (humerus, wrist, hip and vertebra) were also grouped for subsequent fracture analysis. All study subjects were followed forward in time until death or the most recent clinical contact, with a maximum follow-up of 2 years.

Demographic and baseline characteristics were summarized using descriptive statistics. The cumulative number of patients with new fractures, based on time between the first and second fracture, was analyzed using Kaplan-Meier methods with stratification for age, gender, number of baseline fractures and location of fractures. Cox regression was used to analyze independent contribution of those risk factors significant in Kaplan-Meier analysis to new fracture incidence and calculation of hazard ratios (HR) and their confidence intervals. Interactions were also investigated in the Cox regression.

Results

The study included 2,419 patients with 2,575 fractures. There were 139 patients with more than one simultaneous fracture. The group consisted of 1,745 females and 674 males, with a mean age and range 72 years (50–103) and 66 years (50–94), respectively. Fracture location and distribution of these subjects are shown in Table 1. Fifteen patients from abroad and ten patients with pathologic fractures were excluded. Single humerus, radius/ulna and hip fractures formed the majority of fractures, with a total of 1,392 fractures.
Table 1

Mean age and percentage in female and male group by fracture site

Fracture site

Patient number

Female group ( n =1,745)

Male group ( n =674)

Mean age (SD)

Humerus

256

11.8

7.4

73 (11)

Radius/ulna

611

29.5

14.2

69 (11)

Hip

525

22.2

20.3

80 (9)

Multiple

139

5.3

7

71 (12)

Other fractures

888

-

-

66 (12)

Skull ( n =9)

 

-

0.2

0.7

-

Spine ( n =54)

 

1.7

3.7

Clavicle ( n =42)

 

1.3

3

Thorax ( n =81)

 

2.1

6.7

Hand ( n =246)

 

7.7

16.5

Femur ( n =42)

 

1.9

1.3

Patella ( n =19)

 

0.7

0.9

Tibia/fibula ( n =68)

 

2.5

3.6

Ankle ( n =192)

 

8.1

7.4

Foot ( n =135)

 

4.9

7.3

Total

2,419

100%

100%

At the time of the study, 585 patients had died. The remaining 1,834 subjects received the questionnaire. The response rate was 70% (1,276 patients). X-ray reports of all 2,419 subjects were studied for subsequent fractures in our own hospital database (Fig. 1). In the responder group, a total of 139 patients reported new fractures in the questionnaire. In 104 patients, this could be confirmed in the database of our hospital. The other 35 patients were treated in other hospitals. All the patients in the responder group for which the hospital database reported a fracture reported this on the questionnaire.
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-005-2026-x/MediaObjects/198_2005_2026_Fig1_HTML.gif
Fig. 1

Flowchart of patients included in the study

Questionnaire and database research together showed during the 2-year follow-up 262 patients with 286 subsequent fractures. The absolute risk for new fractures within 2 years was 10.8% (262/2,419). Mean overall interval between first fracture and subsequent fracture was 10.8 months. Of all clinical fractures occurring within 2 years after a clinical fracture, 60% occurred during the first year and 40% during the second year ( p =0.005). Men tend to develop a new clinical fracture earlier in the 2-year follow-up (mean time to fracture: 9.5 months) than women do (11.1 months). Seventy percent of all men who developed a new clinical fracture did so in the first year after their first fracture. In women this was 58%. In the 262 patients with subsequent fractures, we observed a higher mean age, more females and more often multiple baseline fractures compared with the 2,157 patients without subsequent fractures. Kaplan-Meier analysis indicated that age, gender and having multiple baseline fractures significantly contribute to new fracture incidence. Absolute fracture risk according to prior fracture type showed the highest risk in subjects with multiple baseline fractures (Table 2). The most frequent new fractures were hip fractures (22.1%) followed by spine (16%), wrist (15.3%), humerus (8%), foot (6.5%). and hand fractures (6.1%). Of the patients with new fractures, 7.6% had multiple fractures.
Table 2

New fracture sites according to site of prior fracture by time and absolute fracture risk

 

Year of study

Absolute new fracture risk (2 year)

Site of new fracture

0–1

1–2

 

   Prior humerus fracture ( n =256)

-

12.1%

      Humerus

0

3

      Ulna/radius

1

3

      Hip

3

2

      Other

12

5

      Multiple

2

0

   Prior ulna/radius fracture ( n =611)

-

9.2%

      Humerus

4

0

      Ulna/radius

8

6

      Hip

7

7

      Other

12

8

      Multiple

1

3

   Prior hip fracture ( n =525)

-

11.4%

      Humerus

2

5

      Ulna/radius

2

3

      Hip

10

7

      Other

19

9

      Multiple

2

1

   Prior other fracture ( n =888)

-

10.2%

      Humerus

3

1

      Ulna/radius

11

2

      Hip

11

6

      Other

26

25

      Multiple

3

3

   Prior multiple fractures ( n =139)

-

17.3%

      Humerus

3

0

      Ulna/radius

2

2

      Hip

4

1

      Other

7

0

      Multiple

3

2

Total

158

104

10.80%

Females had a new fracture incidence of 12.2% compared with males with 7.4% (Fig. 2; p =0.001). New fracture incidence was higher with increasing age (Fig. 3; p <0.001). Patients with multiple baseline fractures showed a new fracture incidence of 17.3% compared with 10.4% for subjects with one baseline fracture (Fig. 4; p =0.007). Cox regression showed age, gender and having multiple baseline fractures as variables independently contributing to a higher subsequent fracture incidence (Table 3). Interactions were considered in the Cox regression but were not significant.
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-005-2026-x/MediaObjects/198_2005_2026_Fig2_HTML.gif
Fig. 2

Subsequent fracture incidence by gender ( n =2,419)

https://static-content.springer.com/image/art%3A10.1007%2Fs00198-005-2026-x/MediaObjects/198_2005_2026_Fig3_HTML.gif
Fig. 3

Subsequent fracture incidence by age ( n =2,419)

https://static-content.springer.com/image/art%3A10.1007%2Fs00198-005-2026-x/MediaObjects/198_2005_2026_Fig4_HTML.gif
Fig. 4

Subsequent fracture incidence by number of baseline fractures ( n =2,419)

Table 3

Cox regression with age, gender and number of baseline fractures

 

Hazard ratio

Confidence interval

p value

Age (per decade)

1.2

1.1–1.3

<0.001

Gender

1.5

1.1–2.0

0.015

Number of baseline fractures

1.8

1.2–2.7

0.006

Discussion

To our knowledge, this is the first study to identify the number of concomitant baseline fractures as an independent risk factor for new fractures. In the meta-analysis of Klotzbuecher, it was suggested that the number of fractures in patient history could be more significant then the location of the fracture [8,9]. This differs from our result in the fact that we did not look at the number of fractures in patient history but at the number of concomitant fractures present at the same time.

The absolute new fracture risk of 10.8% for any clinical fracture and the observation that 60% of these new fractures occurred within 1 year were quite similar to the study of Johnell et al., who reported on fractures of hip, forearm, shoulder and vertebrae [10]. They found an incidence of 12.0% for the same subsequent fractures over 2 years and the highest incidence immediately after the first fracture. Our study indicates that the same incidence is found for all clinical fractures. This underscores the importance of a prevention strategy starting soon after the first fracture for all fracture types.

One may wonder why patients with a fracture are at an increased risk for developing a new fracture immediately following a prior fracture. Silman et al. proposed several possible mechanisms in this prospect [2]. Firstly, bone loss occurs after a fracture, independent of bone mineral density at the time of the fracture. Secondly, risk factors for the development of the first fracture most often remain untreated and increase the susceptibility to subsequent fractures [3,9]. Thirdly, any mechanically induced deformity alters weight distribution in that particular bone, such as in vertebrae, and therefore is associated with an increased risk of fracture at another site [2,6]. Furthermore, as most fractures occurred after a trauma such as a fall, all such patients are also at increased risk for further falls.

This study has several limitations. Firstly, some subsequent fractures may have been missed in the patient group of nonresponders to the questionnaire and the patients that had already died at the time of the study. Systematic research in our own hospital database showed subsequent fractures in these two groups, but certainly some patients were treated in other hospitals, as was the case in the responder group. This results in an underestimation of the new fracture incidence. Secondly, it is possible that some people did not correctly answer the questionnaire. However, a study performed by Ismail assessed the validity of self-reported incident non-spine fractures using a postal questionnaire and concluded that this method was accurate for obtaining information about the occurrence of most fractures, especially hip and distal forearm fractures, including their timing [10]. It was also stated that, if possible, self-reported fractures ideally should be verified from other sources, which we did in our own database research in the responder group.

We conclude that increasing age, female gender and having multiple concomitant fractures independently influence the short-term risk to develop a new clinical fracture after any clinical fracture. This risk is highest in the first year following the fracture.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2005