Calcified Tissue International

, Volume 84, Issue 4, pp 249–256

Osteoarthritis and Risk of Fractures

Authors

    • Department of Endocrinology and Metabolism CAarhus Amtssygehus, Aarhus University Hospital
    • The Osteoporosis ClinicAarhus Amtssygehus, Tage Hansens Gade 2
  • Lars Rejnmark
    • Department of Endocrinology and Metabolism CAarhus Amtssygehus, Aarhus University Hospital
  • Leif Mosekilde
    • Department of Endocrinology and Metabolism CAarhus Amtssygehus, Aarhus University Hospital
Article

DOI: 10.1007/s00223-009-9224-z

Cite this article as:
Vestergaard, P., Rejnmark, L. & Mosekilde, L. Calcif Tissue Int (2009) 84: 249. doi:10.1007/s00223-009-9224-z

Abstract

We conducted a case–control study on the effects of osteoarthritis (OA) on the risk of fractures. There were 124,655 fracture cases and 373,962 age- and gender-matched controls. The main exposure was OA, and the main confounders were use of diuretics, antihypertensive drugs, and pain medication. OA was associated with a decreased risk of any fracture and of hip, forearm, and spine fractures. In general a decreasing trend in the risk of fractures was present with increasing time sine diagnosis of OA. The effect on fractures in areas rich in cortical bone such as the hip in general was larger than effects on skeletal sites rich in trabecular bone such as the spine. OA in the hip and knee, in general, was associated with a decreasing risk of fractures with time since diagnosis of OA, while this was not the case for OA in other locations. OA seems to be associated with a decreased risk of fractures at multiple skeletal sites as well as sites far from the location of OA. This may indicate systemic effects on bone strength, especially in areas rich in cortical bone.

Keywords

OsteoarthritisFractureHipKneeAnalgesic

Osteoarthritis (OA) has been associated with increased bone mineral density (BMD) even at skeletal locations far from the site of OA [1, 2], although this observation could not be reproduced in all studies [3]. Usually, increased BMD is associated with a decreased risk of fractures [4]. However, the mechanisms behind the increased BMD in OA are not completely understood at present [5], and artefacts in BMD from osteophytes may be present in some, but not all, skeletal sites [2]. These osteophytes may not contribute to bone strength, and it is not known in detail if the increased BMD in OA is associated with a true increase in bone biomechanical competence or not. Prior studies have indicated increased trabecular thickness and decreased risk of trabecular microfractures in the bone area just adjacent to the joint affected by OA, but it is not known if this process affects other skeletal sites [5]. A recent study indicated that OA of the spine was not associated with increased volumetric BMD but, rather, increased bone area in the tibia [6]. According to this study, increased volumetric BMD may not contribute to any change in the risk of fractures. However, the increased bone area may be linked to a decreased risk of fractures because the energy of a trauma is distributed over a larger area, leading to a lower force on each square centimeter.

Three prior studies have shown that patients with OA had a lower risk of fractures than control patients [79]. However, not all studies have shown a reduction in the risk of fractures [10, 11], some even showing an increased risk of fractures perhaps related to an increased severity of falls [10]. The studies showing a decreased risk of fractures in patients with OA indicate that OA may indeed change bone biomechanical competence even at skeletal sites far from where OA is localized. However, OA may be associated with heavy lifts [12], hard work [12], and increased body weight [13]; thus, OA may more likely be seen in patients who are physically active, and physical activity may counter development of osteoporosis.

The prior studies have not adjusted for use of drugs that may affect the risk of fractures such as pain medication and other important confounders for fractures such as alcoholism and have not in detail analyzed the effect on risk of fractures at multiple skeletal sites and the effect of duration of OA on risk of fractures. The current study thus analyzes the effect of OA per se on the risk of fractures.

The aims of the present study were to assess the risk of fractures in multiple skeletal sites usually associated with osteoporotic fractures in patients with OA and to determine if there was a time relationship between a diagnosis of OA and risk of fractures.

Materials and Methods

Study Design

The study was designed as a case–control study. All subjects sustaining a fracture during the year 2000 in Denmark were included as cases (n = 124,655), and for each case three subjects of the same age (same birth year) and gender were randomly selected from the background population as controls (n = 373,962). Controls were selected from computerized population lists based on intensity density sampling. Prior fractures were not excluded, and no racial matching was performed as the population is very homogeneous (>95% Caucasians).

End Points

The study end points were occurrence of any fracture (ICD10 codes S02.0–S02.9, S07.0–S07.9, S12.0–S12.9, S22.0–S22.9, S32.0–S32.8, S42.0–S42.9, S52.0–S52.9, S62.0–S62.9, S72.0–S72.9, S82.0–S82.9, S92.0–S92.9) between January 1, 2000, and December 31, 2000. In Denmark almost all patients with fractures are managed in the hospital system (also including emergency rooms) [14] and even fractures sustained abroad are registered upon return for insurance reasons. The capture of fractures is thus high [15, 16].

Exposure Variables

The primary exposure was a diagnosis of OA, especially with reference to OA localization in the skeleton. The diagnosis of OA was based on the ICD8 and from 1994 the ICD10 system and was made by the physician treating the patient based on symptoms, objective findings, X-rays, and need for supplemental examinations such as CT or MRI scans. The validity of diagnoses of diseases of the musculoskeletal system, including OA, was high, with a precision of diagnoses of at least 81% [16].

The other exposure variables were (1) use of drugs known to be associated with fracture risk (corticosteroids, antiepileptic drugs, diuretics, and painkillers [strong and weak analgesics]), (2) number of contacts to the health service (hospitals, general practitioners, or specialists) as a proxy variable for disease severity [17], and (3) social variables [18]. The strong analgesics included morphine, fentanyl, methadone, oxycodone, nicomorphine, ketobemidone, buprenorphine, pethidine, tramadol, codeine, and dextropropoxyphene. The weak analgesics included acetaminophen, acetylsalicylic acid, and any nonsteroidal anti-inflammatory drugs (NSAIDs; aceclofenac, celecoxib, diclofenac, diflunisal, etodolac, flurbiprofen [Flurofen], ibuprofen, indomethacin, ketoprofen, lornoxicam, meloxicam, nabumetone, naproxen, phenylbutazone, piroxicam, rofecoxib, sulindac, tenoxicam, tiaprofen, and tolfenam).

These factors were chosen as they were known to potentially affect fracture risk and were regarded as important potential confounders in a setting where many variables besides the main factor may influence the risk of fractures (confounding by indication). The variables were entered into the statistical analysis, and analyses for interaction were performed. Other important disease confounders included alcoholism [19] and occurrence of a prior fracture or not [20].

The social variables were working or not, income in the year of the fracture (dichotomized by average income), and living alone or together with another person. These factors were included as prior studies have indicated that living in a relationship rather than living alone may be associated with a decreased risk of fractures, that having a job may be associated with fewer fractures than being retired or out of a job, and that income in some settings may be a predictor of fracture risk [18].

Registers

The information on fracture occurrence and occurrence of other diseases, prior fractures, and alcoholism came from two registers: the National Hospital Discharge Register [15] and the Psychiatric Central Register (data on alcohol abuse as many patients with this disorder are registered via this system) [21].

The National Hospital Discharge Register was founded in 1977 [15]. It covers all inpatient contacts from 1977 to 1994 and from 1995 also all outpatient visits to hospitals, outpatient clinics, and emergency rooms [15]. Upon discharge, the physician codes the reason for the contact using the ICD system. The code used is at the discretion of the individual physician. The register has a nationwide coverage and an almost 100% capture of contacts [15]. In general, the validity of registrations is high [16], especially for fractures, where a precision of 97% has been reported for fractures treated both on an inpatient basis and on an outpatient basis via emergency rooms (e.g., a forearm fracture) [22]. The cases occurred only once in the analyses with the first occurrence of an incident fracture during the year 2000.

The National Health Service keeps a register of all contacts to general practitioners for reimbursement purposes. The register does not contain ICD codes for the contacts but codes for the nature of the contact (regular checkup visit, routine vaccination in children).

The number of bed days in the year 1999 was counted as the number of days the patient spent on an inpatient basis in any hospital in 1999. The number of contacts to a general practitioner or specialist was counted as the total number of reimbursement codes issued by the general practitioner or specialist in the year 1999 for each patient.

The Danish Medicines Agency keeps a nationwide register of all drugs sold at pharmacies throughout the country from 1996 onward (National Pharmacological Database run by the Danish Medicines Agency, http://www.dkma.dk). Any drugs bought are registered with ATC code, dosage sold, and date of sale for the period January 1, 1996, to December 31, 2000. As all sales are registered to the individual who redeemed the prescription, the capture and validity are high.

Information on income was obtained from the tax authorities and information on working status and marital status from the National Bureau of Statistics (Statistics Denmark).

It is possible to link these sources of information through the Central Person Register number, which is a unique registration code given to every inhabitant—to some degree similar to the American social security number—that allows registration on an individual basis.

The project was approved and controlled by the National Board of Health, the Danish Data Protection Agency, and the directory board of the Psychiatric Central Register.

Statistical Analyses

Mean and standard deviation were used as descriptive statistics. Crude and adjusted odds ratios (ORs), and 95% confidence intervals (CIs) were calculated. A conditional logistic regression analysis was used to assess the association between any fracture and the exposure variable. Crude and multiply adjusted ORs were calculated. Analyses were performed using STATA 8.2 (StataCorp, College Station, TX) and SPSS 14.0 (SPSS Inc., Chicago IL), both in the UNIX version.

Results

Table 1 shows baseline characteristics of the fracture cases and controls. Patients and controls were well-matched concerning age and gender. Fracture cases more often than controls had comorbid conditions, had a diagnosis of alcoholism, had had prior fractures, and more often had used drugs.
Table 1

Baseline characteristics of fracture cases and controls

Variable

Cases (n = 124,655)

Controls (n = 373,962)

P

Age (years)

43.44 ± 27.39

43.44 ± 27.39

Gender

  

    Men

60,107 (48.2%)

180,321 (48,2%)

 

    Women

64,548 (51.8%)

193,641 (51.8%)

 

Annual income (DKR)

161,036 ± 138,789

172,322 ± 193,704

<0.01

Previous fracture

41,315 (33.1%)

56,200 (15.0%)

<0.01

Number of bed days in hospital in 1999

9.7 ± 39.7

4.2 ± 20.3

<0.01

Charlson indexa

  

<0.01

    0

97,256 (78.0%)

314,099 (84.0%)

 

    1–2

19,634 (16.8%)

47,745 (12.8%)

 

    3–4

5,450 (4.4%)

9,132 (2.4%)

 

    ≥5

2,315 (1.9%)

2,986 (0.8%)

 

Number of contacts to GP or specialist in 1999

23.9 ± 43.3

18.1 ± 31.4

<0.01

Alcoholism

8,863 (7.1%)

9,473 (2.5%)

<0.01

Ever use of antiepileptic drugs

7,091 (5.7%)

10,974 (2.9%)

<0.01

Ever use of any glucocorticoid

67,695 (54.3%)

189,636 (50.7%)

<0.01

Ever use of weak analgesics

65,787 (52.8%)

155,668 (41.6%)

<0.01

Ever use of strong analgesics

38,896 (31.2%)

64,522 (17.3%)

<0.01

Ever use of diuretics

26,466 (21.2%)

68,871 (18.4%)

<0.01

OA

7,091 (5.7%)

16,888 (4.2%)

<0.01

OA of the hip

2,749 (2.2%)

7,291 (1.9%)

<0.01

OA of the knee

2,986 (2.4%)

7,259 (1.9%)

<0.01

OA in other locations

2,384 (1.9%)

4,776 (1.3%)

<0.01

aA composite index of 19 comorbid conditions (see text)

GP general practitioner; DKR Danish crowns (5.5 DKR equals approximately US$1)

In an analysis including ever diagnoses with OA, weak analgesics, strong analgesics, and number of bed days, the ORs associated with risk of any fracture were as follows: OA 0.77 (95% CI 0.75–0.80), strong analgesics 1.80 (95% CI 1.77–1.83), weak analgesics 1.15 (95% CI 1.13–1.16), and number of bed days in hospital in 1999 1.68 (95% CI 1.65–1.72) for 1–3 bed days, 1.75 (95% CI 1.72–1.79) for 3–9 bed days, and 2.03 (95% CI 1.99–2.08) for ≥10 bed days. For hip fractures, the corresponding figures were as follows: 0.66 (95% CI 0.61–0.70) for OA, 2.98 (95% CI 2.83–3.13) for strong analgesics, 1.81 (95% CI 1.69–1.94) for weak analgesics, 1.39 (95% CI 1.28–1.51) for 1–3 bed days, 1.18 (95% CI 1.10–1.27) for 3–9 bed days, and 1.93 (95% CI 1.82–2.04) for ≥10 bed days.

Table 2 shows the relationship between time since diagnosis of OA and risk of fractures. In the crude analysis, the OR for fractures was higher in OA patients than in subjects without OA but with a declining trend with time for overall risk of any fracture (2< 0.01) and risk of spine fractures (2< 0.01). For hip fractures, an increased risk was seen with less than 2 years since OA diagnosis, but a decline below that of subjects without OA was seen more than 5 years after diagnosis in the crude analysis. For forearm fractures, no particular relationship was present in the crude analysis. For overall risk of fractures, there was a significantly decreasing trend with time in the multiply adjusted analysis, even for the time 2–5 vs. >10 years (< 0.05). For hip fractures, an increased risk was present within the first 2 years following the diagnosis of OA in the multiply adjusted analysis, with a significant decrease with more than 2 years since first diagnosis of OA. There was a borderline significant trend toward a decline in relative risk of fractures from 2–5 to >10 years after first diagnosis of OA for hip fractures (P = 0.08). For forearm fractures, the risk was significantly decreased in the multiply adjusted analysis even with ≤2 years of duration since first OA diagnosis. No trend with time was present. For spine fractures, a significant decrease was present in the multiply adjusted analysis with >5 years since first OA diagnosis. There was no particular trend with time, although a significant decrease with time was present for 2–5 years to >10 years after first OA diagnosis (< 0.05).
Table 2

Risk of fractures stratified by duration of OA

Fracture site

Time since diagnosis of OA (years)

Crude OR (95% CI)

Multiply adjusted OR (95% CI)a

Any

≤2

1.74 (1.63–1.86)*

1.05 (0.98–1.13)

2.1–5

1.26 (1.19–1.34)*

0.79 (0.75–0.85)*

5.1–10

1.19 (1.12–1.26)*

0.76 (0.72–0.81)*

>10

1.14 (1.09–1.20)*

0.71 (0.68–0.75)*

Hip

≤2

2.29 (2.00–2.63)*

1.42 (1.22–1.64)*

2.1–5

0.96 (0.83–1.10)

0.60 (0.51–0.70)*

5.1–10

0.86 (0.76–0.98)*

0.55 (0.48–0.64)*

>10

0.84 (0.75–0.93)*

0.52 (0.46–0.58)*

Forearm

≤2

1.06 (0.88–1.28)

0.79 (0.65–0.96)*

2.1–5

1.09 (0.94–1.27)

0.83 (0.71–0.97)*

5.1–10

1.09 (0.94–1.26)

0.85 (0.73–0.98)*

>10

1.01 (0.88–1.15)

0.75 (0.65–0.86)*

Spine

≤2

1.84 (1.35–2.51)*

0.86 (0.61–1.22)

2.1–5

1.78 (1.36–2.33)*

0.95 (0.71–1.28)

5.1–10

1.25 (0.96–1.61)

0.57 (0.43–0.75)*

>10

1.10 (0.88–1.39)

0.62 (0.49–0.80)*

aAdjusted for prior fracture, alcoholism, Charlson index (a composite index of 19 comorbid conditions, see text), use of corticosteroids, use of drugs against epilepsy, use of diuretics, income, living alone or not, working status, use of strong analgesics, use of weak analgesics, number of bed days in 1999, and number of contacts to general practitioner or specialist in 1999

P < 0.01

Tables 36 show the risk of any fracture, hip fracture, forearm fracture, and spine fracture associated with OA of the hip, knee, and other locations. Upon adjustment for confounders, the risk of fractures declined significantly below that of patients without OA with increasing time since diagnosis. The risk factors associated with the largest change in OR were use of strong and weak analgesics and number of bed days in hospital in the preceding year. All other confounders only changed the ORs little (Table 3).
Table 3

Risk of any fracture associated with OA in the hip, knee, and other places

OA site

Time since diagnosis of OA (years)

Crude OR (95% CI)

Adjusted for weak analgesics

Adjusted for weak and strong analgesics

Adjusted for weak and strong analgesics and number of bed days in hospital

Multiply adjusted OR (95% CI)a

Hip

≤2

1.71 (1.55–1.88)*

1.42 (1.29–1.57)*

1.18 (1.06–1.33)*

1.01 (0.91–1.12)

1.03 (0.93–1.15)

2.1–5

1.05 (0.96–1.16)

0.88 (0.80–0.96)*

0.72 (0.66–0.79)*

0.65 (0.59–0.71)*

0.67 (0.61–0.74)*

5.1–10

0.99 (0.91–1.08)

0.85 (0.78–0.92)*

0.72 (0.66–0.79)*

0.69 (0.63–0.75)*

0.70 (0.64–0.77)*

>10

0.88 (0.81–0.95)*

0.77 (0.71–0.83)*

0.67 (0.62–0.73)*

0.64 (0.59–0.69)*

0.66 (0.61–0.72)*

Knee

≤2

1.42 (1.29–1.56)*

1.20 (1.09–1.31)*

1.05 (0.96–1.16)

0.90 (0.82–0.99)*

0.96 (0.87–1.06)

2.1–5

1.16 (1.07–1.26)*

0.98 (0.90–1.06)

0.84 (0.77–0.92)*

0.76 (0.70–0.83)*

0.82 (0.75–0.89)*

5.1–10

1.08 (1.00–1.18)*

0.92 (0.85–1.00)

0.79 (0.72–0.86)*

0.75 (0.69–0.82)*

0.77 (0.71–0.84)*

>10

1.15 (1.06–1.26)*

1.00 (0.91–1.09)

0.86 (0.79–0.94)*

0.81 (0.74–0.88)*

0.79 (0.72–0.86)*

Other

≤2

1.92 (1.70–2.16)*

1.63 (1.45–1.84)*

1.41 (1.25–1.59)*

1.19 (1.05–1.34)*

1.25 (1.10–1.42)*

2.1–5

1.55 (1.39–1.73)*

1.31 (1.18–1.47)*

1.13 (1.01–1.26)*

1.03 (0.92–1.15)

1.02 (0.91–1.14)

5.1–10

1.51 (1.36–1.67)

1.28 (1.15–1.42)*

1.05 (0.95–1.17)

0.99 (0.89–1.11)

0.97 (0.87–1.09)

>10

1.29 (1.19–1.39)*

1.11 (1.03–1.20)*

0.93 (0.86–1.00)

0.89 (0.82–0.96)*

0.87 (0.80–0.94)*

aAdjusted for prior fracture, alcoholism, Charlson index (a composite index of 19 comorbid conditions, see text), use of corticosteroids, use of drugs against epilepsy, use of diuretics, income, living alone or not, working status, use of strong analgesics, use of weak analgesics, number of bed days in 1999, and number of contacts to general practitioner or specialist in 1999

P < 0.01

Table 4

Risk of hip fractures associated with OA in the hip, knee, and other places

OA site

Time since diagnosis of OA (years)

Multiply adjusted OR (95% CI)a

Hip

≤2

1.97 (1.62–2.40)*

2.1–5

0.51 (0.40–0.64)*

5.1–10

0.39 (0.31–0.48)*

>10

0.34 (0.28–0.41)*

Knee

≤2

0.98 (0.78–1.22)

2.1–5

0.70 (0.57–0.86)*

5.1–10

0.71 (0.58–0.86)*

>10

0.86 (0.71–1.03)

Other

≤2

0.82 (0.59–1.15)

2.1–5

0.70 (0.52–0.94)*

5.1–10

0.85 (0.66–1.10)

>10

0.78 (0.65–0.93)*

aAdjusted for prior fracture, alcoholism, Charlson index (a composite index of 19 comorbid conditions, see text), use of corticosteroids, use of drugs against epilepsy, use of diuretics, income, living alone or not, working status, use of strong analgesics, use of weak analgesics, number of bed days in 1999, and number of contacts to general practitioner or specialist in 1999

P < 0.01

Table 5

Risk of forearm fractures associated with OA in the hip, knee, and other places

OA site

Time since diagnosis of OA (years)

Multiply adjusted OR (95% CI)a

Hip

≤2

0.69 (0.50–0.96)*

2.1–5

0.74 (0.58–0.94)*

5.1–10

0.79 (0.63–0.99)*

>10

0.80 (0.66–0.99)*

Knee

≤2

0.77 (0.58–1.02)

2.1–5

0.86 (0.69–1.07)

5.1–10

0.82 (0.67–1.02)

>10

0.65 (0.51–0.84)*

Other

≤2

1.21 (0.88–1.65)

2.1–5

1.07 (0.80–1.42)

5.1–10

0.99 (0.72–1.36)

>10

0.88 (0.71–1.10)

aAdjusted for prior fracture, alcoholism, Charlson index (a composite index of 19 comorbid conditions, see text), use of corticosteroids, use of drugs against epilepsy, use of diuretics, income, living alone or not, working status, use of strong analgesics, use of weak analgesics, number of bed days in 1999, and number of contacts to general practitioner or specialist in 1999

P < 0.01

Table 6

Risk of spine fractures associated with OA in the hip, knee, and other places

OA site

Time since diagnosis of OA (years)

Multiply adjusted OR (95% CI)a

Hip

≤2

1.07 (0.67–1.71)

2.1–5

0.96 (0.63–1.45)

5.1–10

0.48 (0.30–0.76)*

>10

0.67 (0.39–0.83)*

Knee

≤2

0.78 (0.46–1.33)

2.1–5

0.79 (0.53–1.19)

5.1–10

0.59 (0.38–0.91)*

>10

0.61 (0.39–0.97)*

Other

≤2

0.85 (0.45–1.61)

2.1–5

1.05 (0.56–1.94)

5.1–10

1.02 (0.64–1.61)

>10

0.88 (0.59–1.29)

aAdjusted for prior fracture, alcoholism, Charlson index (a composite index of 19 comorbid conditions, see text), use of corticosteroids, use of drugs against epilepsy, use of diuretics, income, living alone or not, working status, use of strong analgesics, use of weak analgesics, number of bed days in 1999, and number of contacts to general practitioner or specialist in 1999

P < 0.01

OA of the hip was associated with decreased risk of fractures with time for overall risk of fractures, hip fractures, forearm fractures, and spine fractures. For hip fractures, an increased risk was present for less than 2 years since first OA diagnosis. For forearm fractures, the decrease in risk was present with all durations.

OA of the knee was associated with a decreased risk of fractures with time for overall risk of fractures (>2 years), hip (2–10, but not >10, years), forearm (>10 years), and spine fractures (>5 years).

OA of other locations was associated with a decreased risk of any fracture with >10 years since first OA diagnosis in the multiply adjusted analysis, while forearm and spine fractures were not associated with OA at other locations. For hip fractures, a trend toward a decrease was present but achieved statistical significance only for 2–5 and >10 years.

Discussion

In this large-scale population-based study we have shown a decreased risk of fractures in patients with OA, especially patients with OA of the knee and hip. This decrease in risk of fractures is in accordance with the observations of some prior studies [79] but in contrast with others that have demonstrated an increased risk of fractures [10, 11].

The prior studies that have shown a decrease in risk of fractures associated with OA have not adjusted for use of drugs, comorbid conditions, or duration of disease. Our study thus adds significant insight into the effect of OA per se without confounding from comorbidity and drug use. One of the studies showing an increased risk of fractures in patients with OA reported no increase in the risk of falls but, rather, that the severity of falls was increased and that knee pain was an independent risk factor for falls and, thus, fractures [10]. This may explain the link with pain medication found in our study as analgesics have been associated with an increased risk of fractures [9, 23]. One study not reporting any overall association between OA and risk of fractures reported an increased risk of spine fractures with increasing disk narrowing [11]. This could indicate that loading of the individual vertebral bodies changed with spine OA and that Schmorl nodes may indent the vertebral end plates and thus weaken them. However, in our study we did show a trend toward a decrease in the risk of spine fractures with increasing duration of OA. The difference between our study and the study by Sornay-Rendu et al. [11] may be that the latter study was based not on clinically diagnosed OA but on a grading system. It is well-known that differences exist between the degree of OA and disk degeneration on an X-ray and the clinical sensation of pain in patients. Furthermore, the duration of OA was not clear in the study by Sornay-Rendu et al. [11] and may have been shorter than in our study. Some studies have shown an increased bone loss rate over time in patients with OA [24], which should suggest an increased risk of fractures with time, whereas the opposite was observed in our study. However, discrepancies exist between studies on BMD, some reporting an increased BMD [25], and thus a larger reservoir to lose from [24], while some have suggested an increased loss rate [24]. More research is thus needed to solve this discrepancy. Although correlations between BMD and degree of OA exist [1], few studies have reported the actual BMD status of patients with and without OA. One study reported a difference in spine BMD of around 0.5 Z scores in men and 0.4 Z scores in women with knee OA [24]. This would translate into a decrease in overall risk of fractures of 1.5−0.5 = 0.82 using the estimates of Marshall et al. [4], which is in the range observed in this study (Tables 2 and 3). It thus appears that the change in BMD may account for some of the alterations in risk of fractures. The mechanisms behind the alterations in BMD are not completely understood [5], but prior research has shown increased formation of subchondral sclerotic bone and increased trabecular thickness with a decreased tendency to microfractures due to increased osteoblastic activity in the area affected by OA [5]. The mechanisms behind the generalized increase in BMD seen in some studies are not understood in detail, but the differences in time until risk of fractures changed may be linked to differences in the response of cortical and trabecular bone to OA. However, in contrast to the studies mentioned above, one study has indicated that volumetric BMD, and thus potentially bone biomechanical competence, was not changed in patients with OA; but an increased bone area was observed [6], and the increased bone area may decrease the risk of fractures. The study on association between OA and volumetric BMD was cross-sectional [6] and may thus have been subject to bias from lack of information on data from other sites than the tibia and lack of information on a number of important confounders such as drug use.

In the crude analysis an increased fracture risk was present, which disappeared upon adjustment for confounders. The increase in the crude analysis may be linked to use of pain medication, which per se may be associated with an increased risk of fractures [9, 23]. In our study, fractures in the spine, which consists mainly of trabecular bone, responded much slower to OA than did fractures of the hip, which consists mainly of cortical bone. Perhaps the localized increase in formation of bone close to the joint surface affected by OA may lead to a more generalized increase in cortical thickness. The early decline in the risk of forearm fractures may be linked both to changes in BMD and to changes in the risk of falls. Patients with OA especially of the lower extremities may move less than other patients, which may decrease the risk of falls and thus fractures.

Overweight may be linked to OA, and higher body mass index (BMI) is associated with increased BMD and a decreased risk of fractures [26]. However, overweight precedes OA; and if high BMI was the sole cause for the increased BMD and decreased risk of fractures, the decrease in risk of fractures should be present already at the time of diagnosis. This was not the case except for forearm fractures.

The confounders that changed OR the most were use of strong analgesics and number of bed days in the year prior to the fracture. This indicates that strong analgesics, although effective in alleviating pain, may have a serious side effect in causing more fractures, probably from falls due to dizziness. Use of strong analgesics should thus be carefully considered in patients with benign conditions such as OA. The association with number of bed days in the year prior to the fracture could indicate an effect of the severity of the comorbid conditions. The fact that these two confounders had such a large effect on the risk of fractures shows the importance of adjustments as OA might otherwise erroneously be associated with an increased risk of fractures.

The major strengths of our study are the large sample size and the near completeness of the register with nationwide coverage. Furthermore, we adjusted for use of drugs which may counter some of the decrease in the risk of fractures [9, 23, 27] and have thus tried to estimate the “true” risk of fractures associated with OA.

The major weaknesses relate to lack of individual information on BMI, physical activity, vitamin D status, dietary intakes, and smoking. Further studies including these factors are thus necessary.

In conclusion, OA seems to be associated with a decreased risk of fractures at multiple skeletal sites and at sites far from the location of OA. This may indicate systemic effects on bone strength, especially in areas rich in cortical bone.

Acknowledgements

Danmarks Statistik (Statistics Denmark) is acknowledged for the help without which this project would not have been possible. Research librarian Ms. Edith Clausen is acknowledged for invaluable help with the references. The Danish Medical Research Council granted financial support (grant 22-04-0495).

Copyright information

© Springer Science+Business Media, LLC 2009