Fragility Fractures Are Associated with an Increased Risk for Cardiovascular Events in Women and Men with Rheumatoid Arthritis: A Population-Based Study

Abstract

Objective

Women and men with rheumatoid arthritis (RA) have an increased risk for fragility fractures and cardiovascular disease (CVD), each of which has been reported to contribute to excess morbidity and mortality in these patients. Fragility fractures share similar risk factors for CVD but may occur at relatively younger ages in RA patients. We aimed to determine if a fragility fracture predicts the development of CVD in women and men with RA.

Methods

We studied a population-based cohort with incident RA in 1955–2007 and compared them with age- and sex-matched non-RA subjects. We identified fragility fractures and CVD events following the RA incidence/index date, along with relevant risk factors. We used Cox models to examine the association between fractures and the development of CVD, where fractures and CVD risk factors were modeled as time-dependent covariates.

Results

There were 1171 subjects (822 women; 349 men) in each of the RA and non-RA cohorts. Over follow-up, there were 406 and 346 fragility fractures and 286 and 225 CVD events, respectively. The overall CVD risk was increased significantly for RA subjects following a fragility fracture (Hazard Ratio [HR]: 1.81; 95% confidence interval [CI]: 1.38–2.37) but not for non-RA subjects (HR [95% CI]: 1.18 [0.85, 1.63]). Results were similar for women and men with RA.

Conclusion

Fragility fractures in both women and men with RA are associated with an increased risk for CVD events and should raise an alert to clinicians to target these individuals for further screening and preventive strategies for CVD.

INTRODUCTION

Cardiovascular disease (CVD) is the leading cause of morbidity and mortality in women and men with rheumatoid arthritis (RA).(1) Patients with RA have an increased risk of CVD, including both ischemic heart disease (IHD) and heart failure (HF), independent of traditional cardiovascular risk factors.(2) Improved identification of RA patients at greatest cardiovascular risk would help better target these individuals for additional screening and preventive strategies.

It is well-recognized that both women and men with RA are also at increased risk for fragility fractures (3–6), including those diagnosed with RA in young adulthood.(5, 6) Indeed, we have reported that women with RA are at increased risk for fragility fractures even before they reach the age of 50 years. (6) Interestingly, there is emerging evidence suggesting that fragility fractures may be a sentinel event that could prove valuable in recognizing RA patients at increased risk for CVD. Bone loss and CVD share common risk factors, including cigarette smoking, sedentary lifestyle and glucocorticoid use.(7, 8) In RA, ongoing chronic inflammation may be another prominent factor contributing to both outcomes. In the general population, elevated levels of pro-inflammatory molecules such as C-reactive protein (CRP), interleukin (IL)-6 and tumor necrosis factor alpha (TNF-α) are increasingly recognized as playing a key role in the pathogenesis of both bone loss (9, 10) and CVD.(11–13) Collectively, these findings suggest that common pathological mechanisms may link fragility fractures and CVD among those with RA.(14–16)

Given the relatively younger age at which fragility fractures can occur in RA, we hypothesized that the presence of a fragility fracture following their diagnosis of RA could be a predictor of future cardiovascular events. We therefore studied a population-based cohort of women and men with an incident diagnosis of RA, in whom all fractures and cardiovascular outcomes have been identified over follow-up. We sought to determine if a fragility fracture in women and men with RA predicts the subsequent risk of CVD, specifically IHD and HF, independent of traditional cardiovascular risk factors.

METHODS

Study Subjects

We studied a well-characterized population-based cohort of Olmsted County, Minnesota, women and men who had an incident diagnosis of RA made in 1955–2007. (17) The study population was assembled using the Rochester Epidemiology Project (REP), a unique medical records linkage system which makes population-based epidemiologic research possible in Olmsted County.(18) Through the resources of the REP, the comprehensive (inpatient and outpatient) medical records for all Olmsted County residents, at any local provider, are available for review.

Following approval by the Institutional Review Boards (IRB) of Mayo Clinic (09-001066) and the Olmsted Medical Center (013-OMC-09), REP resources were used to identify all Rochester residents (the central city of Olmsted County) who were ≥ 18 years of age when they fulfilled American College of Rheumatology (ACR; formerly, the American Rheumatism Association) 1987 criteria for RA between January 1, 1955, and December 31, 1994.(19) The cohort was subsequently expanded, using the same methodology, to include all Olmsted County residents fulfilling ACR criteria for RA from January 1, 1980 to December 31, 2007.(17) Potential RA subjects were identified by searching the computerized diagnostic index for any diagnosis of arthritis (excluding degenerative arthritis or osteoarthritis) made for residents during these time frames. The complete medical record for each potential RA subject was then reviewed by trained nurse abstractors using a pretested data collection form to confirm or reject the diagnosis, with RA incidence defined as the date of first fulfillment of 4 of the 7 ACR classification criteria. For each subject identified with incident RA, an individual without RA from the same population was randomly selected, matched for sex and birth year (± 3 years). Subjects in the non-RA cohort were assigned an index date corresponding to the RA incidence date of their matched pair.

Ascertainment of Fractures

After additional approval by the respective IRBs, these subjects were followed from the date of RA incidence (or corresponding index date for non-RA subjects) until death or last clinical contact through their linked medical records in the community (historical cohort study), and their records were searched by trained nurse abstractors for the occurrence of any fracture.(6) Ascertainment of all clinically evident fractures is believed to be complete.(6) Records at Mayo Clinic, for example, contain the details of every hospitalization and outpatient visit, all emergency room and nursing home care, as well as all radiographic and pathology reports, including autopsies, and all correspondence with each patient.(18) By convention, fractures occurring during daily activities and falls from standing height or less were considered to have resulted from no more than moderate trauma, whereas fractures resulting from motor vehicle accidents and falls from a greater height were deemed from severe trauma. In addition, we are able to distinguish fractures that were due to a specific bone lesion, such as metastatic disease (pathologic fractures), as well as fractures only discovered because of radiographic tests performed in the clinical setting for unrelated indications (incidental fractures). From all fractures identified, we defined a subset of fragility fractures (i.e., all non-pathologic fractures occurring as a result of no more than moderate trauma or identified incidentally), as well as a subset of traditional major osteoporotic fractures (i.e., fragility fractures of the proximal femur [hip], thoracic/lumbar vertebrae [spine], distal forearm [wrist] or proximal humerus [shoulder]).

Ascertainment of Cardiovascular Outcomes

Cardiovascular outcomes were defined as the earliest of the following cardiovascular events: IHD or HF. IHD included documentation of angina, myocardial infarction (MI; including silent events) and coronary revascularization procedures (i.e., coronary artery bypass graft, percutaneous angioplasty, insertion of stents, and atherectomy). MI was defined using standardized epidemiologic criteria.(20) Silent MI was considered as present as of the date of the first documentation of a characteristic electrocardiogram or a recorded physician’s diagnosis in a patient with no documented history of MI. HF was defined according to Framingham criteria and could be of any etiology.(21)

Risk Factors for CVD

Cardiovascular risk factors were defined according to standard epidemiological criteria. Smoking history was collected as never, current or former. Hypertension was defined as two or more ambulatory blood pressure readings ≥140 mm Hg systolic and/or ≥90 mm Hg diastolic obtained during a 1-year period, or a physician’s diagnosis or documented use of antihypertensive medications.(22) Body mass index (BMI) was documented at baseline; obesity was defined as BMI ≥30 kg/m². Diabetes mellitus was defined as fasting plasma glucose ≥126 mg/dl, a physician’s diagnosis or documented use of insulin and/or oral hypoglycemic agents.(23) Dyslipidemia was defined as a low-density lipoprotein (LDL) cholesterol ≥160mg/dl, a total cholesterol ≥240mg/dl, a high-density lipoprotein (HDL) cholesterol of <40mg/dl or triglycerides ≥150mg/dl, a physician’s documentation of dyslipidemia or treatment with a lipid-lowering medication.(24).

RA Disease Characteristics

For the RA cohort, information on RA disease severity had been collected previously through medical record review.(25) Disease severity measures included rheumatoid factor (RF) positivity, as well as the presence of joint erosions/destructive changes and rheumatoid nodules during the first year following RA diagnosis. Information on glucocorticoid use and hormone replacement therapy was also collected.

Statistical Analysis

Descriptive statistics were used to summarize subject characteristics. Subjects with CVD events before the index date were excluded from analyses as they were not at risk of a first CVD event. Cox models, adjusted for traditional cardiovascular risk factors (age, sex, calendar year of RA incidence/index date, current smoking, hypertension, obesity, diabetes mellitus and dyslipidemia) were used to examine the association between either a fragility fracture or major osteoporotic fracture and the development of CVD in each cohort. In addition to the combined CVD outcome, the outcomes of IHD and HF were also examined separately. Time-dependent covariates were used to represent fractures and CVD risk factors in these analyses, which allowed subjects to be modeled as unexposed over follow-up then change to exposed following development of a risk factor. Interactions between sex and fractures were used to determine whether the association between fractures and CVD differed in women compared to men. Interactions between calendar year of RA incidence/index date and fracture were used to examine potential time trends in the association between fractures and CVD.

RESULTS

The RA and non-RA cohorts each contained 1171 subjects (822 women; 349 men) with a mean (± SD) age that was identical for each cohort as of the RA incidence/index date at 57 ± 16 years. Both cohorts were predominately white (93% in RA vs. 94% in non-RA). The median duration (range) of follow-up was 10.0 (0.02–45.7) years for RA subjects and 11.8 (0.01–47.3) years in the non-RA subjects. RA subjects were more likely to be current smokers at index date than non-RA subjects. At the index date, RA subjects were more likely than non-RA subjects to have hypertension, obesity, and dyslipidemia but not diabetes mellitus. Among RA subjects, 67% were RF seropositive, while 26% had erosions and 15% had nodules within the first year after RA diagnosis. The majority of RA subjects (68%) were exposed to glucocorticoids over the course of their follow-up, while 31% of the women had been administered hormone replacement therapy at some point.

Risk for Any CVD Events Following Fracture

There were 137 RA subjects with CVD prior to their diagnosis compared to 139 non-RA subjects who had CVD prior to their index date. These subjects were excluded from analyses examining the risk of CVD following a fracture. Characteristics of the remaining 1034 RA and 1032 non-RA subjects are reported in Table 1. The excluded subjects in both groups were older, more likely to be male and had higher frequencies of CVD risk factors than subjects who were not excluded. Among the included subjects, the differences in characteristics between the RA and non-RA subjects remained similar to the original cohort. Loss to follow-up was similar in both groups (20% in RA vs. 22% in non-RA).

Table 1.

Descriptive characteristics of 1034 Olmsted County women and men age ≥18 years with incident rheumatoid arthritis (RA) in 1955–2007 and 1032 age- and sex-matched non-RA subjects

Characteristics	RA (N=1034)	Non-RA (N=1032)	P Value
Age at index date, years (mean±SD)	54.6 ± 15.0	54.5 ±14.9	0.84
Length of follow-up, years (median [range])	10.5 (0.02–45.7)	12.5 (0.01–47.3)	–
Female sex	739 (71%)	744 (72%)	0.75
White race	959 (93%)	967 (94%)	0.39
Current smoker at incidence/index date	261 (25%)	221 (21%)	0.04
Diabetes mellitus
At incidence/index date	81 (8%)	60 (6%)	0.07
Evera	193 (25%)	198 (28%)	0.99
Hypertension
At incidence/index date	586 (57%)	476 (46%)	<0.001
Evera	877 (92%)	777 (89%)	<0.001
Dyslipidemia
At incidence/index date	410 (40%)	354 (34%)	0.012
Evera	654 (69%)	687 (78%)	0.054
Obesity
At incidence/index date	319 (31%)	265 (26%)	0.009
Evera	398 (41%)	389 (42%)	0.70
Prior major osteoporotic fracture	89 (9%)	99 (10%)	0.44
Rheumatoid factor positivity	675 (67%)	–
Erosionsb	269 (26%)	–
Nodulesb	163 (16%)	–
Evera glucocorticoids	706 (68%)	–
Evera hormone replacement therapy	284 (27%)	–

^aEver percentages represent the cumulative incidence at 30 years after incidence/index date, adjusted for the competing risk of death

^bIn the 1^st year after incidence date

Over 14,125 person-years (p-y) of follow-up following the RA incidence date, there were 406 RA subjects (301 women, 105 men) who had a fragility fracture, and 318 RA subjects (234 women, 84 men) who had a major osteoporotic fracture. There were 286 RA subjects (183 women, 103 men) who developed CVD over follow-up. The rates of CVD among RA subjects after and before/without fractures are presented in Table 2. The median (interquartile range) time to CVD among RA subjects who fractured was 4.7 (1.8, 9.3) years after a fragility fracture and 4.4 (1.6, 7.7) years after a major osteoporotic fracture.

Table 2.

Incidence of cardiovascular disease (CVD) outcomes among subjects with and without rheumatoid arthritis (RA) during follow-up time before and after fractures (any fragility or major osteoporotic [OP]^a fracture).

		CVD rate per 100 person-years
		RA Cohort		Non-RA Cohort
Outcome	Fracture Definition	Nob Fracture	After Fracture	Nob Fracture	After Fracture
Any CVD	Fragility	1.99	4.84	1.42	3.21
	Major OP	2.04	5.79	1.50	3.36
Ischemic heart disease	Fragility	1.36	3.22	1.20	2.16
	Major OP	1.44	3.50	1.24	2.25
Heart failure	Fragility	1.28	3.37	0.82	2.06
	Major OP	1.32	4.03	0.87	2.24

^aMajor OP fracture defined as a fragility fracture at the hip, thoracic or lumbar spine, wrist or shoulder.

^bPerson-years of follow-up and CVD events occurring before fracture are included in the “no fracture” computations.

Compared to RA subjects who did not experience a fracture, RA subjects who sustained a fragility fracture had a significantly increased risk for any CVD (Hazard Ratio [HR]: 1.81; 95% confidence interval [CI]: 1.38–2.37). Similar findings were noted following major osteoporotic fractures (Table 3). Furthermore, when adjusted for glucocorticoid use and hormone replacement therapy, the association between fractures and CVD in RA subjects showed little difference [data not shown].

Table 3.

Risk (hazard ratio) for cardiovascular disease (CVD) following either any fragility fracture or major osteoporotic (OP)^a fracture among subjects with and without rheumatoid arthritis (RA). In each cohort, subjects without the defined fracture served as the referent group.

Outcome	Fracture Definition	Hazard Ratio (95 % CI)b
		RA Cohort	Non-RA Cohort
Any CVD (N=286 RA; N=225 non-RA)	Fragility	1.81 (1.38, 2.37)	1.18 (0.85, 1.63)
	Major OP	1.80 (1.35, 2.40)	1.12 (0.77, 1.62)
Ischemic heart disease (N=202 RA; N=182 non-RA)	Fragility	1.72 (1.24, 2.36)	1.10 (0.75, 1.60)
	Major OP	1.56 (1.10, 2.20)	1.09 (0.71, 1.68)
Heart failure (N=220 RA; N=156 non-RA)	Fragility	1.83 (1.35, 2.49)	1.12 (0.76, 1.63)
	Major OP	1.81 (1.32, 2.47)	1.07 (0.70, 1.63)

^aMajor OP fracture defined as a fragility fracture at the hip, thoracic or lumbar spine, wrist or shoulder.

^bAdjusted for age, sex, calendar year of RA incidence/index date, current smoking at incidence/index date, diabetes mellitus, hypertension, dyslipidemia and obesity.

No significant differences were observed between women and men with RA for the development of CVD following either a fragility fracture (HR: 1.68, 95%CI: 1.12–2.34 for women; HR: 2.28, 95% CI: 1.43–3.65 for men; interaction p-value=0.26) or for major osteoporotic fractures (HR: 1.69, 95%CI: 1.20–2.38 for women; HR: 2.38, 95%CI: 1.43–3.97 for men; interaction p-value=0.35). Similarly, the age at RA diagnosis being ≥ or < 50 years did not appear to have any significant influence on the association between fractures and CVD (interaction p-value>0.4 for both fracture definitions). There were no apparent time trends in the association between fractures and CVD (interaction p-value>0.8 for both fracture definitions).

Markers for RA severity including RF positivity, presence of erosions in the first year and presence of nodules in the first year also did not appear to have any significant impact on the associations between fractures and CVD (interaction p-values ranged from 0.18 to 0.96).

In comparison, over 16,151 p-y of follow-up following the index date, there were 346 non-RA subjects (260 women, 86 men) who had a fragility fracture and 245 non-RA subjects (186 women, 59 men) with a major osteoporotic fracture, while 225 non-RA subjects (151 women, 74 men) had CVD events over follow-up. CVD event rates in patients with RA were significantly higher than in non-RA subjects, both before/without and after fracture (Table 2). When compared with non-RA subjects who did not have a fracture, those with a fragility fracture did not have any significant increase in subsequent risk for CVD (HR: 1.18; 95%CI: 0.85–1.63). Again, similar findings were noted when considering major osteoporotic fractures (Table 3).

Risk of Ischemic Heart Disease and Heart Failure Following Fracture

In addition to analyzing the first CVD event of any type, we also analyzed IHD events and HF events separately. There were 123 RA subjects who had IHD prior to their diagnosis, while 131 non-RA subjects had IHD prior to their matched index date. When these subjects were excluded from analyses, there were 202 RA subjects (125 women, 77 men) and 182 non-RA subjects (118 women, 64 men) who developed IHD over follow-up. Similarly, there were 38 RA and 34 non-RA subjects who had HF prior to RA incidence/index date; these subjects were excluded from analyses of HF. There were 220 RA subjects (146 women, 74 men) and 156 non-RA subjects (100 women, 56 men) who developed HF over follow-up. Rates of IHD and HF outcomes in each of these groups are presented in Table 2.

RA subjects who sustained a fragility fracture had a significantly increased risk for subsequent IHD (HR: 1.72; 95%CI: 1.24–2.36) as well as HF (HR: 1.83; 95%CI: 1.35–2.49) when compared with RA subjects who did not fracture (Table 3). In contrast, non-RA subjects with a fragility fracture were not at any significant increase in subsequent risk for IHD (HR: 1.10; 95%CI: 0.75–1.60) or HF (HR: 1.12; 95%CI: 0.76–1.63) when compared with those who did not fracture. Similar findings were noted following major osteoporotic fractures in each of the RA and non-RA cohorts (Table 3).

DISCUSSION

While RA is an established risk factor for both fragility fractures and CVD, the association between the development of CVD in patients with RA following any fragility fracture has not previously been established. Our findings support our hypothesis that women and men with RA who develop a fragility fracture have a higher risk of subsequently developing cardiovascular events. This was observed for both IHD and HF outcomes. This increased risk of CVD following fragility fractures also appeared to be independent of many of the established cardiovascular risk factors, as well as glucocorticoid use and hormone replacement therapy. In contrast, in non-RA subjects, we observed no association between fractures and CVD following adjustment for CVD risk factors. As we have previously reported, the Framingham risk score underestimates CVD risk among patients with RA, suggesting that other mechanisms, such as RA disease activity, may contribute to the excess risk of CVD in RA. (26) Fractures may be a surrogate marker for these other mechanisms.

There is growing recognition on the importance of establishing recommendations for cardiovascular risk assessment and prevention in patients with RA. However, fragility fractures are not considered in current risk assessment guidelines for CVD in RA. (27, 28) Our findings suggest that a fragility fracture in a RA patient may signal an individual at heightened risk for CVD events, who should be targeted for more aggressive management of any modifiable risk factors and for primary preventative strategies to help lower their CVD risk.

The risk of new CVD events, both IHD and HF, following a fragility fracture in RA is a novel finding. While shared risk factors likely account for this observed association, additional work is necessary on understanding the potential pathogenic link between fragility fractures in RA and subsequent risk for CVD, both for IHD and HF, as that may help identify novel methods for decreasing CVD events in RA patients. Such work may also have implications for CVD management generally. While a number of studies have demonstrated that CVD is associated with future fractures in the general population (16, 29–31), one study did report that HF patients were also more likely to have a history of fractures prior to their HF diagnosis (31), again suggesting the role of shared chronic risk factors. That said, we did not observe an increased risk for CVD following a fragility fracture in our non-RA cohort, which would indicate unique shared risk factors for CVD in RA.

Among the leading potential explanations for our study observations is ongoing chronic inflammation as a shared risk factor. Chronic inflammation appears to have a strong impact on the quality and quantity of bone. The induction of pro-inflammatory cytokines influences bone remodeling and structure, stimulating bone homeostasis in the direction of net bone loss and increasing the likelihood of fracture.(32, 33) It is primarily due to T-cell mediated stimulation of osteoclastogenesis by RANKL, although elevated TNF-α, IL-6, and IL-1, all key inflammatory markers in RA, also adversely affect bone homeostasis through stimulation of bone resorption and inhibition of bone formation.(32–35). Chronic inflammation is also a well-recognized cause of atherosclerosis, being responsible for the development and destabilization of arterial plaques, the cornerstone of IHD.(36) Pro-inflammatory cytokines that are attributed to the development of atherosclerosis are also those seen in RA-driven inflammation. Thus, TNF-α is responsible for a decrease in vascular adhesion molecules;(37) IL-1 is responsible for upregulation of endothelial adhesion molecules and activation of macrophages and vascular cells;(38, 39) and IL-6 has been shown to enhance fatty lesion development.(38, 39) Pro-inflammatory cytokines, including TNF-α and IL-6, are also associated with an increased risk for HF,(13) although the pathogenesis is not as well understood yet. Of note, the associations we identified between fragility fractures and subsequent CVD events were independent of potential markers of RA disease severity, but they were only considered at baseline. Cumulative RA disease activity over follow-up may therefore be more relevant.

There are other possible factors that might account for CVD following fragility fractures in RA patients. Frailty, which is a potentially modifiable risk factor,(40) is a predictor for both fractures and CVD.(41–43) Interestingly, chronic inflammation is also implicated as a risk factor for frailty.(44) Patients with RA are often administered calcium and vitamin D supplementation for osteoporosis management. Although somewhat controversial, recent reports have linked calcium supplementation, with or without vitamin D to the development of CVD.(45) Nonetheless, other studies have not observed an adverse effect of calcium on CVD, (46) while vitamin D supplementation may even be protective for HF.(47) In addition, NSAIDS may increase the risk for CVD (48), and, due to their effect on the release of prostaglandins, may have a negative effect on bone quality.(49, 50) Neither calcium and vitamin D supplementation nor NSAID use could be accounted for in our analyses, as these are often available over-the-counter and not well documented in the medical record.

Our study has a number of strengths. Firstly, the population-based design of the study with extensive follow-up and the use of complete (inpatient and outpatient) contemporary medical record documentation strengthens our work by providing complete ascertainment of study outcomes. Secondly, inclusion of the non-RA comparison cohort ascertained from the same population with identical data collection methods enables us to compare risks. As with all studies, our results also need to be interpreted in light of potential limitations. The Olmsted County population is predominately white, therefore our results may not be generalizable to other racial groups. Changes in management have occurred over the time period studied; however, we found no significant time trends in our results. Also, RA disease activity over follow-up was not consistently available for all subjects to be considered in analyses. Inflammatory cytokines were not available to specifically address whether they played a role in our study findings. If chronic inflammation is a key factor in the observations identified, newer biologic therapies for RA may better control the inflammatory state and thereby decrease the risk for both fractures and CVD. However, the majority of our RA subjects were diagnosed and treated in the pre-biologic era, so we are unable to address this possibility. Even with larger numbers of subjects in the post-biologic era, longer follow-up than is currently available would be needed to address this question, at least at this time. Nevertheless, it would be an important future question to address. Given that cumulative shared risk factors may be the key explanation for how fragility fractures predict future CVD, a fragility fracture in an RA patient, regardless of their current management or disease control, may still be a key signal for being at particularly high risk for CVD.

In conclusion, fragility fractures in patients with RA are associated with an increased risk for the development of future CVD events (including both IHD and HF), independent of traditional cardiovascular risk factors. While shared risk factors likely account for this association, inflammation is a key pathogenic mechanism that is associated with both fragility fractures and CVD and could be an especially important explanation for our findings. Further studies are required to better address this hypothesis. Based on our results, we suggest that patients with RA who have experienced a fragility fracture should be particularly screened for CVD and may warrant more aggressive preventative therapy.