- Email: [email protected]
Discrepancies in risk age and relative risk estimations of cardiovascular disease in patients with inflammatory joint diseases
International Journal of Cardiology 252 (2018) 201–206
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Discrepancies in risk age and relative risk estimations of cardiovascular disease in patients with inflammatory joint diseases Grunde Wibetoe a,⁎, Eirik Ikdahl a, Silvia Rollefstad a, Inge C. Olsen b, Kjetil Bergsmark b, Tore K. Kvien b, Anne Salberg c, Dag M. Soldal d, Gunnstein Bakland e, Åse Lexberg f, Bjørg-Tilde Fevang g, Hans Christian Gulseth h, Glenn Haugeberg i, Anne Grete Semb a a
Preventive Cardio-Rheuma Clinic, Department of Rheumatology, Diakonhjemmet Hospital, Oslo, Norway Department of Rheumatology, Diakonhjemmet Hospital, Oslo, Norway c Lillehammer Hospital for Rheumatic Diseases, Lillehammer, Norway d Department of Rheumatology, Hospital of Southern Norway, Kristiansand, Norway e Department of Rheumatology, University Hospital of Northern Norway, Tromsø, Norway f Department of Rheumatology, Vestre Viken Hospital, Drammen, Norway g Department of Rheumatology, Haukeland University Hospital, Bergen, Norway h Department of Rheumatology, Betanien Hospital, Skien, Norway i Department of Rheumatology, Martina Hansens Hospital, Bærum, Norway b
a r t i c l e
i n f o
Article history: Received 31 July 2017 Accepted 5 October 2017 Keywords: Cardiovascular diseases Risk factors Risk Arthritis
a b s t r a c t Objective: The European guidelines on cardiovascular disease (CVD) prevention advise use of relative risk and risk age algorithms for estimating CVD risk in patients with low estimated absolute risk. Patients with inflammatory joint diseases (IJD) are associated with increased risk of CVD. We aimed to estimate relative risk and risk age across IJD entities and evaluate the agreement between ‘cardiovascular risk age’ and ‘vascular age models’. Methods: Using cross-sectional data from a nationwide project on CVD risk assessment in IJD, risk age estimations were performed in patients with low/moderate absolute risk of fatal CVD. Risk age was calculated according to the cardiovascular risk age and vascular age model, and risk age estimations were compared using regression analysis and calculating percentage of risk age estimations differing ≥5 years. Results: Relative risk was increased in 53% and 20% had three times or higher risk compared to individuals with optimal CVD risk factor levels. Furthermore, 20–42% had a risk age ≥5 years higher than their actual age, according to the specific risk age model. There were only minor differences between IJD entities regarding relative risk and risk age. Discrepancies ≥5 years in estimated risk age were observed in 14–43% of patients. The largest observed difference in calculated risk age was 24 years. Conclusion: In patients with low estimated absolute risk, estimation of relative CVD risk and risk age may identify additional patients at need of intensive CVD preventive efforts. However, there is a substantial discrepancy between the risk age models. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Patients with inflammatory joint diseases (IJD), including rheumatoid arthritis (RA), axial spondyloarthritis (axSpA) and psoriatic arthritis (PsA) have increased risk of cardiovascular disease (CVD) compared to the general population [1–3]. Conventional cardiovascular disease (CVD) risk factors (CVD-RFs) have been shown to be prevalent in IJD populations [4–8], thus efficient and accurate CVD risk assessment may be particularly important in IJD [9]. Several CVD risk algorithms have been developed [10] and the Systematic Coronary Risk Evaluation ⁎ Corresponding author at: Preventive Cardio-Rheuma Clinic, Department of Rheumatology, Diakonhjemmet Hospital, PO Box 23, Vinderen, N-0319, Oslo, Norway. E-mail address: [email protected] (G. Wibetoe).
https://doi.org/10.1016/j.ijcard.2017.10.038 0167-5273/© 2017 Elsevier B.V. All rights reserved.
(SCORE) algorithm has been validated for estimation of absolute 10-year risk of fatal CVD in the general, European population [11,12]. Unfortunately, SCORE and other CVD risk algorithms developed for the general population have been proven to inaccurately predict the risk of CVD in patients with RA [13–17] and validated CVD risk models specifically targeted at IJD patients are currently missing. Awaiting the development of more precise CVD risk algorithms for RA patients, the European League against Rheumatism (EULAR) advocate modifying the SCORE (mSCORE) algorithm. This was based on reported standardised mortality ratios and consensus agreement, and the EULAR recommendations advocate applying a 1.5 multiplication factor to the estimated risk of future CVD in patients with RA [18,19]. The latest European guidelines on CVD prevention recommends estimation of 10-year absolute risk of a fatal atherosclerotic event using
202
G. Wibetoe et al. / International Journal of Cardiology 252 (2018) 201–206
the SCORE algorithm to guide treatment decisions regarding initiation of CVD preventive medication [20]. Furthermore, it is stated in the guidelines that in younger individuals “use of the relative risk chart or calculation of their ‘risk age’ may help in advising them of the need for intensive preventive efforts” [20], and there are indications that the concept of risk age improves risk communication [21]. The relative risk chart is presented in the ESC 2016 guidelines which also refers to two different risk age algorithms; the ‘vascular age’ and the ‘cardiovascular risk age’ model [22,23]. While relative risk is a ratio of the absolute risk of CVD in an individual to the CVD risk given optimal CVD-RF levels with same age and sex, the risk age concept denotes a specific age associated with equal absolute risk given ideal CVD-RFs with the same sex. We questioned whether relative risk and risk age estimation could identify individuals at increased CVD risk among IJD patients who represents a high-CVD-risk population in which validated risk calculators are still missing. Thus, the aims of this study were to estimate relative risk of CVD as well as cardiovascular risk age and vascular age compared to chronologic age in IJD patients. Furthermore, we aimed to evaluate the level of agreement and/or discrepancies between CVD risk estimations according to the different risk age algorithms. In addition, we evaluated if rheumatic disease related variables were associated with estimated relative risk and the difference between risk age and chronologic age. 2. Patients and methods Patients were recruited from the NOrwegian Collaboration on Atherosclerosis in patients with Rheumatic joint diseases (NOCAR) project [24]. Being a quality assurance project, informed consent was not collected and NOCAR was not submitted for approval by ethics boards since this was not required neither by Norwegian law nor the institution policy. However, the project received the appropriate approvals by Data Protection Officers (ref 2014/11741). RA/axSpA/PsA patient were included in the current analyses if they had a lowmoderate absolute risk, corresponding to a 10-year risk of a fatal CVD events b5% as estimated by applying the SCORE algorithm that includes high-density lipoproteincholesterol (HDL-c). For RA patients, the 1.5 multiplication factor was employed. Inclusion was further restricted to patients who were eligible for cardiovascular risk age estimations by being aged 37.5 to 67.5 years. Established atherosclerotic CVD and/or current use of antihypertensive (AntiHT) and/or lipid-lowering therapy (LLT) were exclusion criteria. In NOCAR, systematic CVD risk assessments are implemented in the follow-up of IJD patients who are attending rheumatology outpatient clinics across Norway. So far, data have been recorded in seven clinics (Oslo [Diakonhjemmet Hospital], Lillehammer [Hospital for Rheumatic Diseases], Kristiansand [Hospital of Southern Norway], Skien [Betanien Hospital], Bergen [Haukeland University Hospital], Drammen (Vestre Viken Hospital) and Tromsø [University Hospital of Northern Norway]). Data on self-reported CVD comorbidities, history of diabetes mellitus, and use of AntiHT and/or LLT were recorded, lipids (total cholesterol [TC] and HDL-c) were added to routine laboratory tests, and blood pressure (BP) was measured as part of the clinical examination. In case of initial elevated systolic (sBP) or diastolic BP (N140/90 mm Hg), three BP measurements were performed and the average of the last two were recorded. In addition to CVD related variables, data also included demographic (sex and age), socioeconomic (work status and number of years of education) and rheumatic disease related variables. The latter included specific IJD diagnosis, onset of rheumatic symptoms, serologic markers (rheumatoid factor [RF], anti-citrullinated peptide antibodies [ACPA] and human leukocyte antigen B27 [HLAB27]), markers of inflammation (C-reactive protein [CRP] and erythrocyte sedimentation rate [ESR]), and composite disease activity scores (Disease Activity Score in 28 joints [25] using ESR [DAS28] and Ankylosing Spondylitis Disease Activity Score [26] using CRP [ASDAS]). Lastly, status of anti-rheumatic treatment (glucocorticoids, synthetic and biologic disease-modifying anti-rheumatic drugs [sDMARDs and bDMARDs]) was also recorded. Relative risk was calculated according to the relative risk chart published in the ESC guidelines [20]. In detail, relative risk is estimated separately for daily smokers and nonsmokers, by finding the nearest corresponding pre-defined sBP levels (120/140/160/ 180 mm Hg) and TC levels (4/5/6/7/8 mmol/L) in which specific combinations of these risk factors, yield 40 unique risk cells corresponding to particular relative risks ranging from 1 to 12 [20]. Consequently, patients can have one to twelve times the estimated risk compared to an individual of the same age and sex but with optimal CVD-RF levels (non-smoking, sBP of 120 mm Hg and TC of 4 mmol/L). No classification of relative risk levels have previously been defined, thus we defined patients as having no (relative risk = 1), moderately (relative risk = 2) or highly increased relative risk (≥3), respectively. Similarly, cardiovascular risk age was calculated for males and females by finding the combination of nearest pre-defined age (40/45/50/55/60/65 years), smoking habits (daily smoker/non-smoker), sBP (120/140/160/180 mm Hg) and TC (4/5/6/7/8 mmol/L) levels [23]. For instance, a non-smoking individual with a sBP of 120 mm Hg and TC of
4 mmol/L will have a risk age equal to his/her chronologic age truncated to the nearest 5 year increment. In the development of the vascular age table Cuende et al. imputed TC at 5 mmol/L, sBP of 120 mm Hg and non-smoking in the SCORE algorithm to derive a reference table of absolute risk in individuals classified as having non-elevated CVD risk factors, for each age from 40 and up to 65 years [22]. Consequently, by calculating the absolute risk, a patient's vascular age may be estimated. In the following analyses, 10-year risk of fatal CVD events was calculated according to four different methods: 1) the former SCORE algorithm without HDL-c (SCORE), 2) the latest SCORE algorithm with HDL-c (SCORE-HDL-c), 3) the mSCORE without HDL-c (mSCORE) and 4) mSCORE with HDL-c (mSCORE-HDL-c). For each of these risk calculations, the estimated absolute risk was compared to the vascular age table [22], to find the estimated vascular age. For non-RA individuals, risk age as calculated using the SCORE algorithm without HDL-c would equal chronologic age if they were non-smokers, had sBP of 120 mm Hg and 5 mmol/L of TC. 2.1. Statistics Nominal data are presented as numbers and percentages. Continuous variables are presented as mean with standard deviation (SD) for normally distributed data, and as median with inter-quartile range (IQR) for non-normally distributed data. Group differences were evaluated using Chi-square test for dichotomous endpoints. In cases of low cell counts, Fisher's exact test was applied. For continuous dependent variables, one-way analysis of variance (ANOVA) was conducted, whereas Welch ANOVA was used if homogeneity of variance was violated. Furthermore, Kruskal-Wallis tests were used for continuous variables with non-normal distributions. The difference in years between estimated risk age and chronologic age was calculated for each individual according to the cardiovascular risk age model and the four different vascular age models. For cardiovascular risk age estimations, gap years was calculated by subtracting the nearest corresponding pre-defined age level (40/45/50/55/60/65 years) from the estimated cardiovascular risk age. Since no limits have been previously defined, risk ages ≥5 and ≥10 years above the patient's chronologic age was arbitrarily predefined as moderately and highly elevated risk age, respectively. In a similar fashion, a discrepancy of ≥5 years in risk age estimations between the risk age models was chosen as a substantial level of difference. Level of agreement between risk age models was investigated using linear regression calculating R square (R2). Percentage of observations in which the risk age models displayed minor (b5 years) and major (≥5 years) discrepancies was calculated. Lastly, median difference between estimated risk age and chronologic age was calculated for different levels of estimated relative risk. Association between rheumatic disease and antirheumatic treatment related variables to estimated relative risk and difference between risk age was investigated using linear regression models and Kruskal Wallis H test. Statistical significance was set at p b 0.05, and all statistical analyses were performed using STATA version 14.
3. Results In total, 1826 IJD patients (RA: 899; axSpA: 506; PsA: 421) without established CVD and/or current use of AntiHT/LLT had a low/moderate 10-year risk of CVD (mSCORE-HDL-c b 5%). Patient characteristics are presented in Table 1. Overall, 59% were female (RA: 75%; axSpA: 37%; PsA: 52%) and median (inter-quartile range) age and disease duration was 51 (45, 58) and 8 (4, 16) years, respectively. Fifty-one percent of all IJD patients were current users of bDMARDs. In patients with RA, disease remission (DAS28 b 2.6) was present in 55%, while 39% of axSpA patients had inactive disease (ASDAS b 1.3). While 46% of the total IJD population had an estimated relative risk of 1 (no increased risk of CVD), 33% had twice that risk and 20% had a CVD risk that was three times or higher than the risk given optimal CVD-RFs (Table 2). The highest relative risk calculated was 8. Distribution of relative risk levels (1–8) was similar across IJD entities. Difference between risk age and chronologic age, according to 1) cardiovascular risk age estimations and vascular age estimations derived by using the 2) SCORE, 3) SCORE-HDL-c, 4) mSCORE and 5) mSCOREHDL-c algorithms is presented in the supplementary material (Fig. A.1). Depending on the specific risk age model, 19–33% had a risk age ≥ 5 to b10 years above their chronologic age, and 4–18% had a risk age 10 years or higher than their actual age. Among our patients, the largest difference between estimated risk age and chronologic age was 26 years. Using the vascular age estimations, 7–35% of the individuals had an estimated risk age below their chronologic age, depending on which specific model was applied. The most extreme observation in which estimated risk age was less than chronologic age was 13 years.
G. Wibetoe et al. / International Journal of Cardiology 252 (2018) 201–206
203
Table 1 Patient characteristics. IJD (n = 1826)
RA (n = 899)
axSpA (n = 506)
PsA (n = 421)
p-Values
Age, median (IQR) Female, n (%) Working/students, n (%) Education, mean ± SD (years)
51.3 (44.9, 57.7) 1076 (58.9) 1146 (65.2) 13.5 ± 3.1
54.4 (48.2, 59.8) 672 (74.7) 535 (61.6) 13.3 ± 3.1
46.9 (42.8, 53.3) 185 (36.6) 348 (71.2) 13.9 ± 3.1
49.5 (44.0, 56.5) 219 (52.0) 263 (65.4) 13.3 ± 3.0
b0.001 b0.001 0.002 0.001
Rheumatic disease-related variables Rheumatoid factor+, n (%) ACPA+, n (%) HLA-B27+, n (%) Disease duration, median (IQR) (years) ESR, median (IQR) CRP, median (IQR) DAS28 (ESR), mean ± SD Remission, n (%) Low activity, n (%) Moderate activity, n (%) High activity, n (%) ASDAS (CRP), median (IQR) Inactive disease, n (%) Moderate activity, n (%) High activity, n (%) Very high activity, n (%)
− − − 8.4 (3.7, 16.0) 8 (4, 16) 3 (1, 5) − − − − − − − − − −
364 (67.4) 498 (77.3) − 8.0 (3.8, 14.2) 9 (5, 17) 3 (1, 5) 2.6 ± 1.2 423 (55.3) 124 (16.2) 191 (25.0) 27 (3.5) − − − − −
− − 250 (86.7) 10.6 (3.8, 20.8) 7 (4, 14) 3 (1, 5) − − − − − 1.6 (1.0, 2.5) 147 (38.9) 98 (25.9) 107 (28.3) 26 (6.9)
− − − 7.6 (3.1, 14.7) 7 (4, 15) 3 (1, 5) 2.5 ± 1.3 189 (57.4) 48 (14.6) 82 (24.9) 10 (3.0) 1.5 (1.0, 2.4) 111 (42.7) 67 (25.8) 67 (25.8) 15 (5.8)
− − − b0.001 b0.001 0.064 − − − − − − − − − −
Antirheumatic medication, current use Glucocorticoids, n (%) Methotrexate, n (%) Other sDMARDs, n (%) bDMARDs, n (%)
274 (15.0) 698 (38.2) 917 (50.2) 936 (51.3)
241 (26.8) 497 (55.3) 658 (73.2) 416 (46.3)
6 (1.2) 40 (7.9) 51 (10.1) 300 (59.3)
27 (6.4) 161 (38.2) 208 (49.4) 220 (52.3)
b0.001 b0.001 b0.001 b0.001
Cardiovascular disease related variables Body mass index (BMI), mean ± SD kg/m) BMI ≥ 25 kg/m2, n (%) BMI ≥ 30 kg/m2, n (%) Daily smokers, n (%) Total cholesterol, mean ± SD (mmol/L) High-density lipoprotein-cholesterol, mean ± SD (mmol/L) Systolic blood pressure, mean ± SD (mm hg)
26.2 + 4.5 899 (54.8) 283 (17.3) 349 (19.1) 5.45 ± 1.03 1.57 ± 0.49 128.3 ± 15.4
25.6 + 4.3 410 (49.3) 124 (14.9) 174 (19.4) 5.48 ± 1.06 1.66 ± 0.51 127.4 ± 15.6
26.0 + 4.2 242 (53.8) 70 (15.6) 98 (19.4) 5.36 ± 1.00 1.49 ± 0.46 127.2 ± 15.4
27.8 + 5.1 247 (69.0) 89 (24.9) 77 (18.3) 5.50 ± 1.01 1.47 ± 0.47 131.4 ± 14.8
b0.001 b0.001 b0.001 0.887 0.072 b0.001 b0.001
Significance level between inflammatory joint disease (IJD) patients with rheumatoid arthritis (RA), axial spondyloarthritis (axSpA) and psoriatic arthritis (PsA) was set at p b 0.005. ACPA, anti-citrullinated peptide antibodies; ASDAS, ASAS-endorsed disease activity score; bDMARDs, biologic disease modifying antirheumatic drugs; HLA-B27, human leukocyte antigen B27; CRP, C-reactive protein; DAS28, disease activity score using 28 joint count; ESR, erythrocyte sedimentation rate; IQR, inter-quartile range; SD, standard deviation; sDMARDs, synthetic DMARDs.
The difference between risk age and chronologic age was also investigated for the specific IJD entities. Except for the application of the EULAR 1.5 multiplication factor in RA which resulted in an expected right-skewed distribution using the vascular age algorithms, the difference between risk age and chronologic age was comparable across RA, axSpA and PsA patients (Fig. A.2 in the supplementary material). Furthermore, the difference between risk age and chronologic age increased with higher estimated relative risk for all risk age models (Fig. 1). However, according to vascular age estimations using the SCORE-HDL-c algorithms, 29% of individuals with a relative risk of 2 to 5 had a risk age equal to, or less than, their chronologic age (data not shown). Agreement between the risk age estimations by the cardiovascular risk age and vascular age models are presented in Fig. 2, while agreement between the various vascular age algorithms are presented in the supplementary material (Fig. A.3). The largest observed difference between the cardiovascular risk age and vascular age models was 24 years, in which a female non-smoker aged 63 years old with HDL-c level at 3.4 mmol/L, TC level at 7.1 mmol/L and a sBP of 150 mm Hg had an estimated vascular age (using the SCORE-HDL-c algorithm) of 54 years and a cardiovascular risk age of 78 years (data not shown). Linear regression analyses on the associations between the cardiovascular risk age and the four vascular age models yielded a R2 ranging from 0.81 to 0.96. Discrepancies in risk age estimations by the cardiovascular risk age and vascular age models of ≥5 years were found in 10%–43% of observations (Table A.2 in the supplementary material). Furthermore, within the different vascular age models, R2 ranged 0.81–0.97, while
discrepancies of ≥5 years were found in 5.4–39.2% (Table A.3 in the supplementary material) and the largest discrepancy observed was 23 years (data not shown). Since an overall 7–35% of the total IJD population had a lower risk age than their chronological age, we evaluated if these patients had lower disease activity or were more frequent users of b-DMARDs. Except for current use of glucocorticoids and bDMARDs in RA, these additional analyses revealed no associations among rheumatic disease related variables having a significant and substantial impact on estimated relative risk or calculated risk age – chronologic age difference (Tables A.4–7 in the supplementary material). 4. Discussion To our knowledge, this is the first report evaluating estimated relative risk and risk age across IJD entities, and the first investigation of agreement between these risk age models [27]. Firstly, we found that more than half of all IJD patients with low/ moderate absolute risk of CVD who attended rheumatology outpatient clinics in the NOCAR project had estimated CVD risks that were two times or higher (i.e. relative risk ≥2) than individuals of the same age and sex with ideal CVD-RFs. Moreover, one in every five IJD patients had an estimated CVD risk that was three times or higher than counterparts with optimal CVD-RF levels. These findings are in line with previous notions that while absolute risk is driven by advancing age, a low estimated absolute risk of CVD may conceal a high relative risk [20]. Thus, despite having low absolute risk, numerous young IJD patients
204
G. Wibetoe et al. / International Journal of Cardiology 252 (2018) 201–206
Table 2 Relative risk of cardiovascular disease in patients with inflammatory joint diseases.
Inflammatory joint disease Rheumatoid arthritis Axial spondyloarthritis Psoriatic arthritis Female, n (%) Age (years), n (%) 40 45 50 55 60 65 Daily smokers, n (%) Systolic blood pressure (mm Hg), n (%) 120 140 160 180 Total cholesterol (mmol/L), n (%) 4 5 6 7 8
Relative risk = 1
Relative risk = 2
Relative risk 3 ≤ 12
847 (46.4)a 432 (48.1)b 248 (49.0)c 167 (39.7)d 498 (58.8)
610 (33.4)a 279 (31.0)b 169 (33.4)c 162 (38.5)d 355 (58.2)
369 (20.2)a 188 (20.9)b 89 (17.6)c 92 (21.9)d 223 (60.4)
164 (19.4) 194 (22.9) 162 (19.1) 145 (17.1) 114 (13.5) 68 (8.0) 0 (0)
67 (11.0) 120 (19.7) 117 (19.2) 128 (21.0) 103 (16.9) 75 (12.3) 110 (18.0)
31 (8.4) 66 (17.9) 76 (20.6) 88 (23.8) 75 (20.3) 33 (8.9) 239 (64.8)
686 (81.0) 159 (18.8) 2 (0.2) 0 (0)
206 (33.8) 368 (60.3) 36 (5.9) 0 (0)
88 (23.8) 147 (39.8) 107 (29.0) 27 (7.3)
250 (29.5) 335 (39.6) 262 (30.9) 0 (0) 0 (0)
44 (7.2) 232 (38.0) 135 (22.1) 179 (29.3) 20 (3.3)
17 (4.6) 82 (22.2) 151 (40.9) 76 (20.6) 43 (11.7)
Distribution of relative risk levels in patients with inflammatory joint diseases, including separate values for rheumatoid arthritis, axial spondyloarthritis and psoriatic arthritis. The table also displays the distribution of cardiovascular risk factors implemented in the relative risk chart by predefined age strata (40/45/50/55/60/65 years) and level of systolic blood pressure (sBP 120/140/160/180 mm Hg) and total cholesterol (TC 4/5/6/7/ 8 mmol/L) presented with n (numbers) and percentage (%). a n = 1826. b n = 899. c n = 506. d n = 421.
have high relative risk, indicating that they may benefit from optimizing CVD-RF levels. When calculating absolute 10-year risk of fatal CVD events, it is worth noting that risk of non-fatal and fatal CVD events may be three- or four-fold higher than the risk of fatal CVD in men and females, respectively [28].
Secondly, our analyses also revealed differences in estimated risk between the cardiovascular risk age model and the vascular age models. In general, there was a high rate of risk age estimations that differed ≥5 years between the two different risk age algorithms. We also found that several individuals had high relative risks despite having a risk age that was comparable to their chronologic age in the two vascular risk age models employing the SCORE algorithms including HDL-c. These findings demonstrate frequent and substantial diverging risk estimations between these different risk algorithms. In detail, due to the large increments between predefined sBP and TC levels in the risk age chart by Cooney et al., cardiovascular risk age estimations are vulnerable to substantial discrepancies between individuals with highly comparable CVD-RFs. For instance, two male smokers with only slightly differing sBP (129 and 131 mm Hg) and TC (7.4 and 7.6 mmol/L) levels have 5 year difference in estimated cardiovascular risk age [23]. Interestingly, since the vascular age model was developed using 5 mmol/L and 120 mm Hg as reference values, any patient with lower TC and sBP levels may have an estimated risk age below their chronologic age according to this risk age model. Moreover, while Cuende et al. [22] developed the vascular age chart based on calculations using the former SCORE calculator, the latest SCORE calculator includes HDL-c. Furthermore, since application of the EULAR 1.5 multiplication factor is advised in RA, vascular age was calculated in four different manners for RA patients. Lastly, whereas Norway is a low-risk country, SCORE algorithms with or without HDL-c have been developed for high risk countries as well [20]. A 10-year risk of a fatal CVD event of ≥ 5% denotes a threshold of when patients are considered high risk individuals who “may be candidates for drug treatment”, whereas those with low to moderate risk should be offered lifestyle advice [20]. We estimated relative risk and risk age in all IJD patients with low-to moderate absolute risk in which a high frequency of asymptomatic atherosclerosis has been found previously [29,30]. The observation that IJD patients with low/moderate CVD risk according to SCORE frequently have elevated risk ages and relative risks, indicates that there may be a substantial potential for optimization of CVD-RF levels in these individuals. Use of concepts such as relative risk and risk age may improve CVD risk communication and thus possibly patient adherence to CVD preventive measures. However, validation
Fig. 1. Association of relative risk and difference between risk age and chronologic age. Boxplots of median difference in risk age and chronologic age according to cardiovascular age and vascular age estimations for patients with different levels of estimated relative risk. The vascular age estimations were calculated by using the SCORE (Systematic Coronary Risk Evaluation), SCORE-HDL-c (SCORE with high-density lipoprotein-cholesterol), mSCORE (modified SCORE by using the EULAR 1.5 multiplication factor in RA) and mSCORE-HDL-c (modified SCORE-HDL-c) algorithms.
G. Wibetoe et al. / International Journal of Cardiology 252 (2018) 201–206
205
Fig. 2. Agreement between cardiovascular risk age and vascular age estimations. Scatter plot of cardiovascular age and vascular age estimations. The latter were calculated using the SCORE (Systematic Coronary Risk Evaluation), SCORE-HDL-c (SCORE with high-density lipoprotein-cholesterol), mSCORE (modified SCORE by using the EULAR 1.5 multiplication factor in RA) and mSCORE-HDL-c (modified SCORE-HDL-c) algorithms.
of the discriminatory abilities of such models in terms of prediction of observed events is needed prior to full-scale implementation of these risk algorithms in CVD preventive programs. Until then, clinicians should be aware of the discrepancies between the cardiovascular risk age and vascular age models. In this paper, we conducted our analysis in IJD patients. However, our findings of a high grade of discrepancy between different risk age algorithms would similarly be demonstrated with data from non-IJD patients since risk age is calculated based solely on age, sex, smoking status and levels of systolic blood pressure and cholesterol levels. In additional analyses, with the exception of current use of glucocorticoids and bDMARDs in RA, we did not find a strong association between rheumatic disease activity and/or treatment and estimated CVD risk according to the relative risk and risk age algorithms. However, several studies have found that CVD risk in IJD is also altered due to rheumatic disease related factors, such as the duration and severity of systemic inflammation [31,32] as well as use of anti-rheumatic medications, including NSAIDs [33,34], glucocorticoids [33,35], methotrexate [33,36,37] and bDMARDs [33,38,39]. It is worth noting that the vascular age models using the mSCORE/mSCORE-HDL-c algorithms applies a 1.5 multiplication factor in RA, whereas the cardiovascular risk age and relative risk algorithms only incorporates conventional CVD-RFs. Careful interpretations of our findings are required due to some methodological limitations. The main purpose of this paper was 1) to estimate CVD risk according to relative risk and risk age algorithms in accordance to conventional CVD-RFs and 2) evaluating the agreement between different risk age models. We report on predicted CVD risk according to different risk algorithms using cross-sectional data and not on actual observed CVD events. Moreover, we do not compare estimated risk between IJD and non-IJD patients which is beyond the scope of this paper. Any data are prone to error of misclassification and while self-reported smoking habits have been shown to be quite accurate [40], sBP measurements only conducted at a single visit may not reflect
overall BP. The external validity of estimated risk in IJD patients outside Northern Europe also needs consideration since Norway represents a low-CVD-risk country [20] with a predominant Caucasian population. In conclusion, numerous IJD patients with low estimated absolute risk have increased relative risk and/or risk age due to high levels of CVD-RFs. There are only minor differences in estimated relative risk and risk age between IJD entities. Poor comparability of risk age models further underlines the importance of validation of these risk age algorithms before their full-scale implementation in CVD preventive programs.
Acknowledgments We are thankful to all study nurses, medical doctors and health personnel for participating and facilitating the NOCAR project. Funding The South Eastern Norwegian Regional Health Authority (2015055 and 2014111) provided funding for 2 PhD students. In addition funding was provided by Grethe Harbitz legacy and Olav Raagholt and Gerd Meidel Raagholts Foundation. Conflicts of interest The authors report no relationships that could be construed as a conflict of interest. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.ijcard.2017.10.038.
206
G. Wibetoe et al. / International Journal of Cardiology 252 (2018) 201–206
References [1] J.A. Avina-Zubieta, J. Thomas, M. Sadatsafavi, A.J. Lehman, D. Lacaille, Risk of incident cardiovascular events in patients with rheumatoid arthritis: a meta-analysis of observational studies, Ann. Rheum. Dis. 71 (2012) 1524–1529. [2] S. Mathieu, B. Pereira, M. Soubrier, Cardiovascular events in ankylosing spondylitis: an updated meta-analysis, Semin. Arthritis Rheum. 44 (2015) 551–555. [3] C. Horreau, C. Pouplard, E. Brenaut, T. Barnetche, L. Misery, B. Cribier, et al., Cardiovascular morbidity and mortality in psoriasis and psoriatic arthritis: a systematic literature review, J. Eur. Acad. Dermatol. Venereol. 27 (Suppl. 3) (2013) 12–29. [4] S. Castaneda, M.A. Martin-Martinez, C. Gonzalez-Juanatey, J. Llorca, M.J. GarciaYebenes, S. Perez-Vicente, et al., Cardiovascular morbidity and associated risk factors in Spanish patients with chronic inflammatory rheumatic diseases attending rheumatology clinics: baseline data of the CARMA project, Semin. Arthritis Rheum. 44 (2015) 618–626. [5] C. Han, D.W. Robinson Jr., M.V. Hackett, L.C. Paramore, K.H. Fraeman, M.V. Bala, Cardiovascular disease and risk factors in patients with rheumatoid arthritis, psoriatic arthritis, and ankylosing spondylitis, J. Rheumatol. 33 (2006) 2167–2172. [6] M. Labitigan, A. Bahce-Altuntas, J.M. Kremer, G. Reed, J.D. Greenberg, N. Jordan, et al., Higher rates and clustering of abnormal lipids, obesity, and diabetes mellitus in psoriatic arthritis compared with rheumatoid arthritis, Arthritis Care Res. 66 (2014) 600–607. [7] C.C. Mok, G.T.C. Ko, L.Y. Ho, Yu KL, P.T. Chan, C.H. To, Prevalence of atherosclerotic risk factors and the metabolic syndrome in patients with chronic inflammatory arthritis, Arthritis Care Res. 63 (2011) 195–202. [8] G. Wibetoe, E. Ikdahl, S. Rollefstad, I.C. Olsen, K. Bergsmark, T.K. Kvien, et al., Cardiovascular disease risk profiles in inflammatory joint disease entities, Arthritis Res. Ther. 19 (2017) 153. [9] A.G. Semb, S. Rollefstad, P. van Riel, G.D. Kitas, E.L. Matteson, S.E. Gabriel, Cardiovascular disease assessment in rheumatoid arthritis: a guide to translating knowledge of cardiovascular risk into clinical practice, Ann. Rheum. Dis. 73 (2014) 1284–1288. [10] J.A. Damen, L. Hooft, E. Schuit, T.P. Debray, G.S. Collins, I. Tzoulaki, et al., Prediction models for cardiovascular disease risk in the general population: systematic review, BMJ i2416 (2016) 353. [11] R.M. Conroy, K. Pyorala, A.P. Fitzgerald, S. Sans, A. Menotti, G. De Backer, et al., Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project, Eur. Heart J. 24 (2003) 987–1003. [12] M.K. Aktas, V. Ozduran, C.E. Pothier, R. Lang, M.S. Lauer, Global risk scores and exercise testing for predicting all-cause mortality in a preventive medicine program, JAMA 292 (2004) 1462–1468. [13] E.E. Arts, C. Popa, A.A. Den Broeder, A.G. Semb, T. Toms, G.D. Kitas, et al., Performance of four current risk algorithms in predicting cardiovascular events in patients with early rheumatoid arthritis, Ann. Rheum. Dis. 74 (2015) 668–674. [14] E.E. Arts, C.D. Popa, A.A. Den Broeder, R. Donders, A. Sandoo, T. Toms, et al., Prediction of cardiovascular risk in rheumatoid arthritis: performance of original and adapted SCORE algorithms, Ann. Rheum. Dis. 75 (2016) 674–680. [15] C. Gomez-Vaquero, A. Corrales, A. Zacarias, J. Rueda-Gotor, R. Blanco, C. GonzalezJuanatey, et al., SCORE and REGICOR function charts underestimate the cardiovascular risk in Spanish patients with rheumatoid arthritis, Arthritis Res. Ther. 15 (2013) R91. [16] C.S. Crowson, E.L. Matteson, V.L. Roger, T.M. Therneau, S.E. Gabriel, Usefulness of risk scores to estimate the risk of cardiovascular disease in patients with rheumatoid arthritis, Am. J. Cardiol. 110 (2012) 420–424. [17] V.K. Kawai, C.P. Chung, J.F. Solus, A. Oeser, P. Raggi, C.M. Stein, Brief report: the ability of the 2013 American College of Cardiology/American Heart Association cardiovascular risk score to identify rheumatoid arthritis patients with high coronary artery calcification scores, Arthritis Rheum. 67 (2015) 381–385. [18] R. Agca, S.C. Heslinga, S. Rollefstad, M. Heslinga, I.B. McInnes, M.J. Peters, et al., EULAR recommendations for cardiovascular disease risk management in patients with rheumatoid arthritis and other forms of inflammatory joint disorders: 2015/2016 update, Ann. Rheum. Dis. (2016). [19] S. Dadoun, N. Zeboulon-Ktorza, C. Combescure, M. Elhai, S. Rozenberg, L. Gossec, et al., Mortality in rheumatoid arthritis over the last fifty years: systematic review and meta-analysis, Joint Bone Spine 80 (2013) 29–33. [20] M.F. Piepoli, A.W. Hoes, S. Agewall, C. Albus, C. Brotons, A.L. Catapano, et al., European guidelines on cardiovascular disease prevention in clinical practice: the Sixth Joint Task Force of the European Society of Cardiology and other societies on cardiovascular disease prevention in clinical practice (constituted by representatives of 10 societies and by invited experts): developed with the special contribution of the European Association for Cardiovascular Prevention & Rehabilitation (EACPR), Eur. Heart J. (2016) 2016.
[21] A. Soureti, R. Hurling, P. Murray, W. van Mechelen, M. Cobain, Evaluation of a cardiovascular disease risk assessment tool for the promotion of healthier lifestyles, Eur. J. Cardiovasc. Prev. Rehabil. 17 (2010) 519–523. [22] J.I. Cuende, N. Cuende, J. Calaveras-Lagartos, How to calculate vascular age with the SCORE project scales: a new method of cardiovascular risk evaluation, Eur. Heart J. 31 (2010) 2351–2358. [23] M.T. Cooney, E. Vartiainen, T. Laatikainen, D. De Bacquer, C. McGorrian, A. Dudina, et al., Cardiovascular risk age: concepts and practicalities, Heart 98 (2012) 941–946. [24] E. Ikdahl, S. Rollefstad, G. Wibetoe, A. Salberg, D.M. Soldal, I.C. Olsen, T.K. Kvien, G. Haugeberg, A.G. Semb, How to implement cardiovascular disease risk assessment for patients with inflammatory joint diseases in daily rheumatology practice: an overview of a nationwide Norwegian project [abstract], Arthritis Rheum. 68 (suppl 10) (2016) http://acrabstracts.org/abstract/how-to-implement-cardiovascular-disease-risk-assessment-for-patients-with-inflammatory-joint-diseases-indaily-rheumatology-practice-an-overview-of-a-nationwide-norwegian-project/ (Accessed October 24, 2017). [25] M.L. Prevoo, M.A. van't Hof, H.H. Kuper, M.A. van Leeuwen, L.B. van de Putte, P.L. van Riel, Modified disease activity scores that include twenty-eight-joint counts. Development and validation in a prospective longitudinal study of patients with rheumatoid arthritis, Arthritis Rheum. 38 (1995) 44–48. [26] C. Lukas, R. Landewe, J. Sieper, M. Dougados, J. Davis, J. Braun, et al., Development of an ASAS-endorsed disease activity score (ASDAS) in patients with ankylosing spondylitis, Ann. Rheum. Dis. 68 (2009) 18–24. [27] K. Groenewegen, H. Den Ruijter, G. Pasterkamp, J. Polak, M. Bots, S.A. Peters, Vascular age to determine cardiovascular disease risk: a systematic review of its concepts, definitions, and clinical applications, Eur. J. Prev. Cardiol. 23 (2016) 264–274. [28] I. van Dis, J.M. Geleijnse, J.M. Boer, D. Kromhout, H. Boshuizen, D.E. Grobbee, et al., Effect of including nonfatal events in cardiovascular risk estimation, illustrated with data from The Netherlands, Eur. J. Prev. Cardiol. 21 (2014) 377–383. [29] A.G. Semb, E. Ikdahl, J. Hisdal, I.C. Olsen, S. Rollefstad, Exploring cardiovascular disease risk evaluation in patients with inflammatory joint diseases, Int. J. Cardiol. 223 (2016) 331–336. [30] A. Corrales, C. Gonzalez-Juanatey, M.E. Peiro, R. Blanco, J. Llorca, M.A. Gonzalez-Gay, Carotid ultrasound is useful for the cardiovascular risk stratification of patients with rheumatoid arthritis: results of a population-based study, Ann. Rheum. Dis. 73 (2014) 722–727. [31] D.H. Solomon, J. Kremer, J.R. Curtis, M.C. Hochberg, G. Reed, P. Tsao, et al., Explaining the cardiovascular risk associated with rheumatoid arthritis: traditional risk factors versus markers of rheumatoid arthritis severity, Ann. Rheum. Dis. 69 (2010) 1920–1925. [32] P.M. Ridker, M. Cushman, M.J. Stampfer, R.P. Tracy, C.H. Hennekens, Inflammation, aspirin, and the risk of cardiovascular disease in apparently healthy men, N. Engl. J. Med. 336 (1997) 973–979. [33] C. Roubille, V. Richer, T. Starnino, C. McCourt, A. McFarlane, P. Fleming, et al., The effects of tumour necrosis factor inhibitors, methotrexate, non-steroidal antiinflammatory drugs and corticosteroids on cardiovascular events in rheumatoid arthritis, psoriasis and psoriatic arthritis: a systematic review and meta-analysis, Ann. Rheum. Dis. 74 (2015) 480–489. [34] S. Trelle, S. Reichenbach, S. Wandel, P. Hildebrand, B. Tschannen, P.M. Villiger, et al., Cardiovascular safety of non-steroidal anti-inflammatory drugs: network metaanalysis, BMJ c7086 (2011) 342. [35] L. Wei, T.M. MacDonald, B.R. Walker, Taking glucocorticoids by prescription is associated with subsequent cardiovascular disease, Ann. Intern. Med. 141 (2004) 764–770. [36] S.L. Westlake, A.N. Colebatch, J. Baird, P. Kiely, M. Quinn, E. Choy, et al., The effect of methotrexate on cardiovascular disease in patients with rheumatoid arthritis: a systematic literature review, Rheumatology (Oxford) 49 (2010) 295–307. [37] H.K. Choi, M.A. Hernán, J.D. Seeger, J.M. Robins, F. Wolfe, Methotrexate and mortality in patients with rheumatoid arthritis: a prospective study, Lancet 359 (2002) 1173–1177. [38] C. Barnabe, B.J. Martin, W.A. Ghali, Systematic review and meta-analysis: anti-tumor necrosis factor α therapy and cardiovascular events in rheumatoid arthritis, Arthritis Care Res. 63 (2011) 522–529. [39] S.L. Westlake, A.N. Colebatch, J. Baird, N. Curzen, P. Kiely, M. Quinn, et al., Tumour necrosis factor antagonists and the risk of cardiovascular disease in patients with rheumatoid arthritis: a systematic literature review, Rheumatology (Oxford) 50 (2011) 518–531. [40] L.G. Kvalvik, R.M. Nilsen, R. Skjaerven, S.E. Vollset, O. Midttun, P.M. Ueland, et al., Self-reported smoking status and plasma cotinine concentrations among pregnant women in the Norwegian Mother and Child Cohort Study, Pediatr. Res. 72 (2012) 101–107.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Discrepancies in risk age and relative risk estimations of cardiovascular disease in patients with inflammatory joint diseases Grunde Wibetoe a,⁎, Eirik Ikdahl a, Silvia Rollefstad a, Inge C. Olsen b, Kjetil Bergsmark b, Tore K. Kvien b, Anne Salberg c, Dag M. Soldal d, Gunnstein Bakland e, Åse Lexberg f, Bjørg-Tilde Fevang g, Hans Christian Gulseth h, Glenn Haugeberg i, Anne Grete Semb a a
Preventive Cardio-Rheuma Clinic, Department of Rheumatology, Diakonhjemmet Hospital, Oslo, Norway Department of Rheumatology, Diakonhjemmet Hospital, Oslo, Norway c Lillehammer Hospital for Rheumatic Diseases, Lillehammer, Norway d Department of Rheumatology, Hospital of Southern Norway, Kristiansand, Norway e Department of Rheumatology, University Hospital of Northern Norway, Tromsø, Norway f Department of Rheumatology, Vestre Viken Hospital, Drammen, Norway g Department of Rheumatology, Haukeland University Hospital, Bergen, Norway h Department of Rheumatology, Betanien Hospital, Skien, Norway i Department of Rheumatology, Martina Hansens Hospital, Bærum, Norway b
a r t i c l e
i n f o
Article history: Received 31 July 2017 Accepted 5 October 2017 Keywords: Cardiovascular diseases Risk factors Risk Arthritis
a b s t r a c t Objective: The European guidelines on cardiovascular disease (CVD) prevention advise use of relative risk and risk age algorithms for estimating CVD risk in patients with low estimated absolute risk. Patients with inflammatory joint diseases (IJD) are associated with increased risk of CVD. We aimed to estimate relative risk and risk age across IJD entities and evaluate the agreement between ‘cardiovascular risk age’ and ‘vascular age models’. Methods: Using cross-sectional data from a nationwide project on CVD risk assessment in IJD, risk age estimations were performed in patients with low/moderate absolute risk of fatal CVD. Risk age was calculated according to the cardiovascular risk age and vascular age model, and risk age estimations were compared using regression analysis and calculating percentage of risk age estimations differing ≥5 years. Results: Relative risk was increased in 53% and 20% had three times or higher risk compared to individuals with optimal CVD risk factor levels. Furthermore, 20–42% had a risk age ≥5 years higher than their actual age, according to the specific risk age model. There were only minor differences between IJD entities regarding relative risk and risk age. Discrepancies ≥5 years in estimated risk age were observed in 14–43% of patients. The largest observed difference in calculated risk age was 24 years. Conclusion: In patients with low estimated absolute risk, estimation of relative CVD risk and risk age may identify additional patients at need of intensive CVD preventive efforts. However, there is a substantial discrepancy between the risk age models. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Patients with inflammatory joint diseases (IJD), including rheumatoid arthritis (RA), axial spondyloarthritis (axSpA) and psoriatic arthritis (PsA) have increased risk of cardiovascular disease (CVD) compared to the general population [1–3]. Conventional cardiovascular disease (CVD) risk factors (CVD-RFs) have been shown to be prevalent in IJD populations [4–8], thus efficient and accurate CVD risk assessment may be particularly important in IJD [9]. Several CVD risk algorithms have been developed [10] and the Systematic Coronary Risk Evaluation ⁎ Corresponding author at: Preventive Cardio-Rheuma Clinic, Department of Rheumatology, Diakonhjemmet Hospital, PO Box 23, Vinderen, N-0319, Oslo, Norway. E-mail address: [email protected] (G. Wibetoe).
https://doi.org/10.1016/j.ijcard.2017.10.038 0167-5273/© 2017 Elsevier B.V. All rights reserved.
(SCORE) algorithm has been validated for estimation of absolute 10-year risk of fatal CVD in the general, European population [11,12]. Unfortunately, SCORE and other CVD risk algorithms developed for the general population have been proven to inaccurately predict the risk of CVD in patients with RA [13–17] and validated CVD risk models specifically targeted at IJD patients are currently missing. Awaiting the development of more precise CVD risk algorithms for RA patients, the European League against Rheumatism (EULAR) advocate modifying the SCORE (mSCORE) algorithm. This was based on reported standardised mortality ratios and consensus agreement, and the EULAR recommendations advocate applying a 1.5 multiplication factor to the estimated risk of future CVD in patients with RA [18,19]. The latest European guidelines on CVD prevention recommends estimation of 10-year absolute risk of a fatal atherosclerotic event using
202
G. Wibetoe et al. / International Journal of Cardiology 252 (2018) 201–206
the SCORE algorithm to guide treatment decisions regarding initiation of CVD preventive medication [20]. Furthermore, it is stated in the guidelines that in younger individuals “use of the relative risk chart or calculation of their ‘risk age’ may help in advising them of the need for intensive preventive efforts” [20], and there are indications that the concept of risk age improves risk communication [21]. The relative risk chart is presented in the ESC 2016 guidelines which also refers to two different risk age algorithms; the ‘vascular age’ and the ‘cardiovascular risk age’ model [22,23]. While relative risk is a ratio of the absolute risk of CVD in an individual to the CVD risk given optimal CVD-RF levels with same age and sex, the risk age concept denotes a specific age associated with equal absolute risk given ideal CVD-RFs with the same sex. We questioned whether relative risk and risk age estimation could identify individuals at increased CVD risk among IJD patients who represents a high-CVD-risk population in which validated risk calculators are still missing. Thus, the aims of this study were to estimate relative risk of CVD as well as cardiovascular risk age and vascular age compared to chronologic age in IJD patients. Furthermore, we aimed to evaluate the level of agreement and/or discrepancies between CVD risk estimations according to the different risk age algorithms. In addition, we evaluated if rheumatic disease related variables were associated with estimated relative risk and the difference between risk age and chronologic age. 2. Patients and methods Patients were recruited from the NOrwegian Collaboration on Atherosclerosis in patients with Rheumatic joint diseases (NOCAR) project [24]. Being a quality assurance project, informed consent was not collected and NOCAR was not submitted for approval by ethics boards since this was not required neither by Norwegian law nor the institution policy. However, the project received the appropriate approvals by Data Protection Officers (ref 2014/11741). RA/axSpA/PsA patient were included in the current analyses if they had a lowmoderate absolute risk, corresponding to a 10-year risk of a fatal CVD events b5% as estimated by applying the SCORE algorithm that includes high-density lipoproteincholesterol (HDL-c). For RA patients, the 1.5 multiplication factor was employed. Inclusion was further restricted to patients who were eligible for cardiovascular risk age estimations by being aged 37.5 to 67.5 years. Established atherosclerotic CVD and/or current use of antihypertensive (AntiHT) and/or lipid-lowering therapy (LLT) were exclusion criteria. In NOCAR, systematic CVD risk assessments are implemented in the follow-up of IJD patients who are attending rheumatology outpatient clinics across Norway. So far, data have been recorded in seven clinics (Oslo [Diakonhjemmet Hospital], Lillehammer [Hospital for Rheumatic Diseases], Kristiansand [Hospital of Southern Norway], Skien [Betanien Hospital], Bergen [Haukeland University Hospital], Drammen (Vestre Viken Hospital) and Tromsø [University Hospital of Northern Norway]). Data on self-reported CVD comorbidities, history of diabetes mellitus, and use of AntiHT and/or LLT were recorded, lipids (total cholesterol [TC] and HDL-c) were added to routine laboratory tests, and blood pressure (BP) was measured as part of the clinical examination. In case of initial elevated systolic (sBP) or diastolic BP (N140/90 mm Hg), three BP measurements were performed and the average of the last two were recorded. In addition to CVD related variables, data also included demographic (sex and age), socioeconomic (work status and number of years of education) and rheumatic disease related variables. The latter included specific IJD diagnosis, onset of rheumatic symptoms, serologic markers (rheumatoid factor [RF], anti-citrullinated peptide antibodies [ACPA] and human leukocyte antigen B27 [HLAB27]), markers of inflammation (C-reactive protein [CRP] and erythrocyte sedimentation rate [ESR]), and composite disease activity scores (Disease Activity Score in 28 joints [25] using ESR [DAS28] and Ankylosing Spondylitis Disease Activity Score [26] using CRP [ASDAS]). Lastly, status of anti-rheumatic treatment (glucocorticoids, synthetic and biologic disease-modifying anti-rheumatic drugs [sDMARDs and bDMARDs]) was also recorded. Relative risk was calculated according to the relative risk chart published in the ESC guidelines [20]. In detail, relative risk is estimated separately for daily smokers and nonsmokers, by finding the nearest corresponding pre-defined sBP levels (120/140/160/ 180 mm Hg) and TC levels (4/5/6/7/8 mmol/L) in which specific combinations of these risk factors, yield 40 unique risk cells corresponding to particular relative risks ranging from 1 to 12 [20]. Consequently, patients can have one to twelve times the estimated risk compared to an individual of the same age and sex but with optimal CVD-RF levels (non-smoking, sBP of 120 mm Hg and TC of 4 mmol/L). No classification of relative risk levels have previously been defined, thus we defined patients as having no (relative risk = 1), moderately (relative risk = 2) or highly increased relative risk (≥3), respectively. Similarly, cardiovascular risk age was calculated for males and females by finding the combination of nearest pre-defined age (40/45/50/55/60/65 years), smoking habits (daily smoker/non-smoker), sBP (120/140/160/180 mm Hg) and TC (4/5/6/7/8 mmol/L) levels [23]. For instance, a non-smoking individual with a sBP of 120 mm Hg and TC of
4 mmol/L will have a risk age equal to his/her chronologic age truncated to the nearest 5 year increment. In the development of the vascular age table Cuende et al. imputed TC at 5 mmol/L, sBP of 120 mm Hg and non-smoking in the SCORE algorithm to derive a reference table of absolute risk in individuals classified as having non-elevated CVD risk factors, for each age from 40 and up to 65 years [22]. Consequently, by calculating the absolute risk, a patient's vascular age may be estimated. In the following analyses, 10-year risk of fatal CVD events was calculated according to four different methods: 1) the former SCORE algorithm without HDL-c (SCORE), 2) the latest SCORE algorithm with HDL-c (SCORE-HDL-c), 3) the mSCORE without HDL-c (mSCORE) and 4) mSCORE with HDL-c (mSCORE-HDL-c). For each of these risk calculations, the estimated absolute risk was compared to the vascular age table [22], to find the estimated vascular age. For non-RA individuals, risk age as calculated using the SCORE algorithm without HDL-c would equal chronologic age if they were non-smokers, had sBP of 120 mm Hg and 5 mmol/L of TC. 2.1. Statistics Nominal data are presented as numbers and percentages. Continuous variables are presented as mean with standard deviation (SD) for normally distributed data, and as median with inter-quartile range (IQR) for non-normally distributed data. Group differences were evaluated using Chi-square test for dichotomous endpoints. In cases of low cell counts, Fisher's exact test was applied. For continuous dependent variables, one-way analysis of variance (ANOVA) was conducted, whereas Welch ANOVA was used if homogeneity of variance was violated. Furthermore, Kruskal-Wallis tests were used for continuous variables with non-normal distributions. The difference in years between estimated risk age and chronologic age was calculated for each individual according to the cardiovascular risk age model and the four different vascular age models. For cardiovascular risk age estimations, gap years was calculated by subtracting the nearest corresponding pre-defined age level (40/45/50/55/60/65 years) from the estimated cardiovascular risk age. Since no limits have been previously defined, risk ages ≥5 and ≥10 years above the patient's chronologic age was arbitrarily predefined as moderately and highly elevated risk age, respectively. In a similar fashion, a discrepancy of ≥5 years in risk age estimations between the risk age models was chosen as a substantial level of difference. Level of agreement between risk age models was investigated using linear regression calculating R square (R2). Percentage of observations in which the risk age models displayed minor (b5 years) and major (≥5 years) discrepancies was calculated. Lastly, median difference between estimated risk age and chronologic age was calculated for different levels of estimated relative risk. Association between rheumatic disease and antirheumatic treatment related variables to estimated relative risk and difference between risk age was investigated using linear regression models and Kruskal Wallis H test. Statistical significance was set at p b 0.05, and all statistical analyses were performed using STATA version 14.
3. Results In total, 1826 IJD patients (RA: 899; axSpA: 506; PsA: 421) without established CVD and/or current use of AntiHT/LLT had a low/moderate 10-year risk of CVD (mSCORE-HDL-c b 5%). Patient characteristics are presented in Table 1. Overall, 59% were female (RA: 75%; axSpA: 37%; PsA: 52%) and median (inter-quartile range) age and disease duration was 51 (45, 58) and 8 (4, 16) years, respectively. Fifty-one percent of all IJD patients were current users of bDMARDs. In patients with RA, disease remission (DAS28 b 2.6) was present in 55%, while 39% of axSpA patients had inactive disease (ASDAS b 1.3). While 46% of the total IJD population had an estimated relative risk of 1 (no increased risk of CVD), 33% had twice that risk and 20% had a CVD risk that was three times or higher than the risk given optimal CVD-RFs (Table 2). The highest relative risk calculated was 8. Distribution of relative risk levels (1–8) was similar across IJD entities. Difference between risk age and chronologic age, according to 1) cardiovascular risk age estimations and vascular age estimations derived by using the 2) SCORE, 3) SCORE-HDL-c, 4) mSCORE and 5) mSCOREHDL-c algorithms is presented in the supplementary material (Fig. A.1). Depending on the specific risk age model, 19–33% had a risk age ≥ 5 to b10 years above their chronologic age, and 4–18% had a risk age 10 years or higher than their actual age. Among our patients, the largest difference between estimated risk age and chronologic age was 26 years. Using the vascular age estimations, 7–35% of the individuals had an estimated risk age below their chronologic age, depending on which specific model was applied. The most extreme observation in which estimated risk age was less than chronologic age was 13 years.
G. Wibetoe et al. / International Journal of Cardiology 252 (2018) 201–206
203
Table 1 Patient characteristics. IJD (n = 1826)
RA (n = 899)
axSpA (n = 506)
PsA (n = 421)
p-Values
Age, median (IQR) Female, n (%) Working/students, n (%) Education, mean ± SD (years)
51.3 (44.9, 57.7) 1076 (58.9) 1146 (65.2) 13.5 ± 3.1
54.4 (48.2, 59.8) 672 (74.7) 535 (61.6) 13.3 ± 3.1
46.9 (42.8, 53.3) 185 (36.6) 348 (71.2) 13.9 ± 3.1
49.5 (44.0, 56.5) 219 (52.0) 263 (65.4) 13.3 ± 3.0
b0.001 b0.001 0.002 0.001
Rheumatic disease-related variables Rheumatoid factor+, n (%) ACPA+, n (%) HLA-B27+, n (%) Disease duration, median (IQR) (years) ESR, median (IQR) CRP, median (IQR) DAS28 (ESR), mean ± SD Remission, n (%) Low activity, n (%) Moderate activity, n (%) High activity, n (%) ASDAS (CRP), median (IQR) Inactive disease, n (%) Moderate activity, n (%) High activity, n (%) Very high activity, n (%)
− − − 8.4 (3.7, 16.0) 8 (4, 16) 3 (1, 5) − − − − − − − − − −
364 (67.4) 498 (77.3) − 8.0 (3.8, 14.2) 9 (5, 17) 3 (1, 5) 2.6 ± 1.2 423 (55.3) 124 (16.2) 191 (25.0) 27 (3.5) − − − − −
− − 250 (86.7) 10.6 (3.8, 20.8) 7 (4, 14) 3 (1, 5) − − − − − 1.6 (1.0, 2.5) 147 (38.9) 98 (25.9) 107 (28.3) 26 (6.9)
− − − 7.6 (3.1, 14.7) 7 (4, 15) 3 (1, 5) 2.5 ± 1.3 189 (57.4) 48 (14.6) 82 (24.9) 10 (3.0) 1.5 (1.0, 2.4) 111 (42.7) 67 (25.8) 67 (25.8) 15 (5.8)
− − − b0.001 b0.001 0.064 − − − − − − − − − −
Antirheumatic medication, current use Glucocorticoids, n (%) Methotrexate, n (%) Other sDMARDs, n (%) bDMARDs, n (%)
274 (15.0) 698 (38.2) 917 (50.2) 936 (51.3)
241 (26.8) 497 (55.3) 658 (73.2) 416 (46.3)
6 (1.2) 40 (7.9) 51 (10.1) 300 (59.3)
27 (6.4) 161 (38.2) 208 (49.4) 220 (52.3)
b0.001 b0.001 b0.001 b0.001
Cardiovascular disease related variables Body mass index (BMI), mean ± SD kg/m) BMI ≥ 25 kg/m2, n (%) BMI ≥ 30 kg/m2, n (%) Daily smokers, n (%) Total cholesterol, mean ± SD (mmol/L) High-density lipoprotein-cholesterol, mean ± SD (mmol/L) Systolic blood pressure, mean ± SD (mm hg)
26.2 + 4.5 899 (54.8) 283 (17.3) 349 (19.1) 5.45 ± 1.03 1.57 ± 0.49 128.3 ± 15.4
25.6 + 4.3 410 (49.3) 124 (14.9) 174 (19.4) 5.48 ± 1.06 1.66 ± 0.51 127.4 ± 15.6
26.0 + 4.2 242 (53.8) 70 (15.6) 98 (19.4) 5.36 ± 1.00 1.49 ± 0.46 127.2 ± 15.4
27.8 + 5.1 247 (69.0) 89 (24.9) 77 (18.3) 5.50 ± 1.01 1.47 ± 0.47 131.4 ± 14.8
b0.001 b0.001 b0.001 0.887 0.072 b0.001 b0.001
Significance level between inflammatory joint disease (IJD) patients with rheumatoid arthritis (RA), axial spondyloarthritis (axSpA) and psoriatic arthritis (PsA) was set at p b 0.005. ACPA, anti-citrullinated peptide antibodies; ASDAS, ASAS-endorsed disease activity score; bDMARDs, biologic disease modifying antirheumatic drugs; HLA-B27, human leukocyte antigen B27; CRP, C-reactive protein; DAS28, disease activity score using 28 joint count; ESR, erythrocyte sedimentation rate; IQR, inter-quartile range; SD, standard deviation; sDMARDs, synthetic DMARDs.
The difference between risk age and chronologic age was also investigated for the specific IJD entities. Except for the application of the EULAR 1.5 multiplication factor in RA which resulted in an expected right-skewed distribution using the vascular age algorithms, the difference between risk age and chronologic age was comparable across RA, axSpA and PsA patients (Fig. A.2 in the supplementary material). Furthermore, the difference between risk age and chronologic age increased with higher estimated relative risk for all risk age models (Fig. 1). However, according to vascular age estimations using the SCORE-HDL-c algorithms, 29% of individuals with a relative risk of 2 to 5 had a risk age equal to, or less than, their chronologic age (data not shown). Agreement between the risk age estimations by the cardiovascular risk age and vascular age models are presented in Fig. 2, while agreement between the various vascular age algorithms are presented in the supplementary material (Fig. A.3). The largest observed difference between the cardiovascular risk age and vascular age models was 24 years, in which a female non-smoker aged 63 years old with HDL-c level at 3.4 mmol/L, TC level at 7.1 mmol/L and a sBP of 150 mm Hg had an estimated vascular age (using the SCORE-HDL-c algorithm) of 54 years and a cardiovascular risk age of 78 years (data not shown). Linear regression analyses on the associations between the cardiovascular risk age and the four vascular age models yielded a R2 ranging from 0.81 to 0.96. Discrepancies in risk age estimations by the cardiovascular risk age and vascular age models of ≥5 years were found in 10%–43% of observations (Table A.2 in the supplementary material). Furthermore, within the different vascular age models, R2 ranged 0.81–0.97, while
discrepancies of ≥5 years were found in 5.4–39.2% (Table A.3 in the supplementary material) and the largest discrepancy observed was 23 years (data not shown). Since an overall 7–35% of the total IJD population had a lower risk age than their chronological age, we evaluated if these patients had lower disease activity or were more frequent users of b-DMARDs. Except for current use of glucocorticoids and bDMARDs in RA, these additional analyses revealed no associations among rheumatic disease related variables having a significant and substantial impact on estimated relative risk or calculated risk age – chronologic age difference (Tables A.4–7 in the supplementary material). 4. Discussion To our knowledge, this is the first report evaluating estimated relative risk and risk age across IJD entities, and the first investigation of agreement between these risk age models [27]. Firstly, we found that more than half of all IJD patients with low/ moderate absolute risk of CVD who attended rheumatology outpatient clinics in the NOCAR project had estimated CVD risks that were two times or higher (i.e. relative risk ≥2) than individuals of the same age and sex with ideal CVD-RFs. Moreover, one in every five IJD patients had an estimated CVD risk that was three times or higher than counterparts with optimal CVD-RF levels. These findings are in line with previous notions that while absolute risk is driven by advancing age, a low estimated absolute risk of CVD may conceal a high relative risk [20]. Thus, despite having low absolute risk, numerous young IJD patients
204
G. Wibetoe et al. / International Journal of Cardiology 252 (2018) 201–206
Table 2 Relative risk of cardiovascular disease in patients with inflammatory joint diseases.
Inflammatory joint disease Rheumatoid arthritis Axial spondyloarthritis Psoriatic arthritis Female, n (%) Age (years), n (%) 40 45 50 55 60 65 Daily smokers, n (%) Systolic blood pressure (mm Hg), n (%) 120 140 160 180 Total cholesterol (mmol/L), n (%) 4 5 6 7 8
Relative risk = 1
Relative risk = 2
Relative risk 3 ≤ 12
847 (46.4)a 432 (48.1)b 248 (49.0)c 167 (39.7)d 498 (58.8)
610 (33.4)a 279 (31.0)b 169 (33.4)c 162 (38.5)d 355 (58.2)
369 (20.2)a 188 (20.9)b 89 (17.6)c 92 (21.9)d 223 (60.4)
164 (19.4) 194 (22.9) 162 (19.1) 145 (17.1) 114 (13.5) 68 (8.0) 0 (0)
67 (11.0) 120 (19.7) 117 (19.2) 128 (21.0) 103 (16.9) 75 (12.3) 110 (18.0)
31 (8.4) 66 (17.9) 76 (20.6) 88 (23.8) 75 (20.3) 33 (8.9) 239 (64.8)
686 (81.0) 159 (18.8) 2 (0.2) 0 (0)
206 (33.8) 368 (60.3) 36 (5.9) 0 (0)
88 (23.8) 147 (39.8) 107 (29.0) 27 (7.3)
250 (29.5) 335 (39.6) 262 (30.9) 0 (0) 0 (0)
44 (7.2) 232 (38.0) 135 (22.1) 179 (29.3) 20 (3.3)
17 (4.6) 82 (22.2) 151 (40.9) 76 (20.6) 43 (11.7)
Distribution of relative risk levels in patients with inflammatory joint diseases, including separate values for rheumatoid arthritis, axial spondyloarthritis and psoriatic arthritis. The table also displays the distribution of cardiovascular risk factors implemented in the relative risk chart by predefined age strata (40/45/50/55/60/65 years) and level of systolic blood pressure (sBP 120/140/160/180 mm Hg) and total cholesterol (TC 4/5/6/7/ 8 mmol/L) presented with n (numbers) and percentage (%). a n = 1826. b n = 899. c n = 506. d n = 421.
have high relative risk, indicating that they may benefit from optimizing CVD-RF levels. When calculating absolute 10-year risk of fatal CVD events, it is worth noting that risk of non-fatal and fatal CVD events may be three- or four-fold higher than the risk of fatal CVD in men and females, respectively [28].
Secondly, our analyses also revealed differences in estimated risk between the cardiovascular risk age model and the vascular age models. In general, there was a high rate of risk age estimations that differed ≥5 years between the two different risk age algorithms. We also found that several individuals had high relative risks despite having a risk age that was comparable to their chronologic age in the two vascular risk age models employing the SCORE algorithms including HDL-c. These findings demonstrate frequent and substantial diverging risk estimations between these different risk algorithms. In detail, due to the large increments between predefined sBP and TC levels in the risk age chart by Cooney et al., cardiovascular risk age estimations are vulnerable to substantial discrepancies between individuals with highly comparable CVD-RFs. For instance, two male smokers with only slightly differing sBP (129 and 131 mm Hg) and TC (7.4 and 7.6 mmol/L) levels have 5 year difference in estimated cardiovascular risk age [23]. Interestingly, since the vascular age model was developed using 5 mmol/L and 120 mm Hg as reference values, any patient with lower TC and sBP levels may have an estimated risk age below their chronologic age according to this risk age model. Moreover, while Cuende et al. [22] developed the vascular age chart based on calculations using the former SCORE calculator, the latest SCORE calculator includes HDL-c. Furthermore, since application of the EULAR 1.5 multiplication factor is advised in RA, vascular age was calculated in four different manners for RA patients. Lastly, whereas Norway is a low-risk country, SCORE algorithms with or without HDL-c have been developed for high risk countries as well [20]. A 10-year risk of a fatal CVD event of ≥ 5% denotes a threshold of when patients are considered high risk individuals who “may be candidates for drug treatment”, whereas those with low to moderate risk should be offered lifestyle advice [20]. We estimated relative risk and risk age in all IJD patients with low-to moderate absolute risk in which a high frequency of asymptomatic atherosclerosis has been found previously [29,30]. The observation that IJD patients with low/moderate CVD risk according to SCORE frequently have elevated risk ages and relative risks, indicates that there may be a substantial potential for optimization of CVD-RF levels in these individuals. Use of concepts such as relative risk and risk age may improve CVD risk communication and thus possibly patient adherence to CVD preventive measures. However, validation
Fig. 1. Association of relative risk and difference between risk age and chronologic age. Boxplots of median difference in risk age and chronologic age according to cardiovascular age and vascular age estimations for patients with different levels of estimated relative risk. The vascular age estimations were calculated by using the SCORE (Systematic Coronary Risk Evaluation), SCORE-HDL-c (SCORE with high-density lipoprotein-cholesterol), mSCORE (modified SCORE by using the EULAR 1.5 multiplication factor in RA) and mSCORE-HDL-c (modified SCORE-HDL-c) algorithms.
G. Wibetoe et al. / International Journal of Cardiology 252 (2018) 201–206
205
Fig. 2. Agreement between cardiovascular risk age and vascular age estimations. Scatter plot of cardiovascular age and vascular age estimations. The latter were calculated using the SCORE (Systematic Coronary Risk Evaluation), SCORE-HDL-c (SCORE with high-density lipoprotein-cholesterol), mSCORE (modified SCORE by using the EULAR 1.5 multiplication factor in RA) and mSCORE-HDL-c (modified SCORE-HDL-c) algorithms.
of the discriminatory abilities of such models in terms of prediction of observed events is needed prior to full-scale implementation of these risk algorithms in CVD preventive programs. Until then, clinicians should be aware of the discrepancies between the cardiovascular risk age and vascular age models. In this paper, we conducted our analysis in IJD patients. However, our findings of a high grade of discrepancy between different risk age algorithms would similarly be demonstrated with data from non-IJD patients since risk age is calculated based solely on age, sex, smoking status and levels of systolic blood pressure and cholesterol levels. In additional analyses, with the exception of current use of glucocorticoids and bDMARDs in RA, we did not find a strong association between rheumatic disease activity and/or treatment and estimated CVD risk according to the relative risk and risk age algorithms. However, several studies have found that CVD risk in IJD is also altered due to rheumatic disease related factors, such as the duration and severity of systemic inflammation [31,32] as well as use of anti-rheumatic medications, including NSAIDs [33,34], glucocorticoids [33,35], methotrexate [33,36,37] and bDMARDs [33,38,39]. It is worth noting that the vascular age models using the mSCORE/mSCORE-HDL-c algorithms applies a 1.5 multiplication factor in RA, whereas the cardiovascular risk age and relative risk algorithms only incorporates conventional CVD-RFs. Careful interpretations of our findings are required due to some methodological limitations. The main purpose of this paper was 1) to estimate CVD risk according to relative risk and risk age algorithms in accordance to conventional CVD-RFs and 2) evaluating the agreement between different risk age models. We report on predicted CVD risk according to different risk algorithms using cross-sectional data and not on actual observed CVD events. Moreover, we do not compare estimated risk between IJD and non-IJD patients which is beyond the scope of this paper. Any data are prone to error of misclassification and while self-reported smoking habits have been shown to be quite accurate [40], sBP measurements only conducted at a single visit may not reflect
overall BP. The external validity of estimated risk in IJD patients outside Northern Europe also needs consideration since Norway represents a low-CVD-risk country [20] with a predominant Caucasian population. In conclusion, numerous IJD patients with low estimated absolute risk have increased relative risk and/or risk age due to high levels of CVD-RFs. There are only minor differences in estimated relative risk and risk age between IJD entities. Poor comparability of risk age models further underlines the importance of validation of these risk age algorithms before their full-scale implementation in CVD preventive programs.
Acknowledgments We are thankful to all study nurses, medical doctors and health personnel for participating and facilitating the NOCAR project. Funding The South Eastern Norwegian Regional Health Authority (2015055 and 2014111) provided funding for 2 PhD students. In addition funding was provided by Grethe Harbitz legacy and Olav Raagholt and Gerd Meidel Raagholts Foundation. Conflicts of interest The authors report no relationships that could be construed as a conflict of interest. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.ijcard.2017.10.038.
206
G. Wibetoe et al. / International Journal of Cardiology 252 (2018) 201–206
References [1] J.A. Avina-Zubieta, J. Thomas, M. Sadatsafavi, A.J. Lehman, D. Lacaille, Risk of incident cardiovascular events in patients with rheumatoid arthritis: a meta-analysis of observational studies, Ann. Rheum. Dis. 71 (2012) 1524–1529. [2] S. Mathieu, B. Pereira, M. Soubrier, Cardiovascular events in ankylosing spondylitis: an updated meta-analysis, Semin. Arthritis Rheum. 44 (2015) 551–555. [3] C. Horreau, C. Pouplard, E. Brenaut, T. Barnetche, L. Misery, B. Cribier, et al., Cardiovascular morbidity and mortality in psoriasis and psoriatic arthritis: a systematic literature review, J. Eur. Acad. Dermatol. Venereol. 27 (Suppl. 3) (2013) 12–29. [4] S. Castaneda, M.A. Martin-Martinez, C. Gonzalez-Juanatey, J. Llorca, M.J. GarciaYebenes, S. Perez-Vicente, et al., Cardiovascular morbidity and associated risk factors in Spanish patients with chronic inflammatory rheumatic diseases attending rheumatology clinics: baseline data of the CARMA project, Semin. Arthritis Rheum. 44 (2015) 618–626. [5] C. Han, D.W. Robinson Jr., M.V. Hackett, L.C. Paramore, K.H. Fraeman, M.V. Bala, Cardiovascular disease and risk factors in patients with rheumatoid arthritis, psoriatic arthritis, and ankylosing spondylitis, J. Rheumatol. 33 (2006) 2167–2172. [6] M. Labitigan, A. Bahce-Altuntas, J.M. Kremer, G. Reed, J.D. Greenberg, N. Jordan, et al., Higher rates and clustering of abnormal lipids, obesity, and diabetes mellitus in psoriatic arthritis compared with rheumatoid arthritis, Arthritis Care Res. 66 (2014) 600–607. [7] C.C. Mok, G.T.C. Ko, L.Y. Ho, Yu KL, P.T. Chan, C.H. To, Prevalence of atherosclerotic risk factors and the metabolic syndrome in patients with chronic inflammatory arthritis, Arthritis Care Res. 63 (2011) 195–202. [8] G. Wibetoe, E. Ikdahl, S. Rollefstad, I.C. Olsen, K. Bergsmark, T.K. Kvien, et al., Cardiovascular disease risk profiles in inflammatory joint disease entities, Arthritis Res. Ther. 19 (2017) 153. [9] A.G. Semb, S. Rollefstad, P. van Riel, G.D. Kitas, E.L. Matteson, S.E. Gabriel, Cardiovascular disease assessment in rheumatoid arthritis: a guide to translating knowledge of cardiovascular risk into clinical practice, Ann. Rheum. Dis. 73 (2014) 1284–1288. [10] J.A. Damen, L. Hooft, E. Schuit, T.P. Debray, G.S. Collins, I. Tzoulaki, et al., Prediction models for cardiovascular disease risk in the general population: systematic review, BMJ i2416 (2016) 353. [11] R.M. Conroy, K. Pyorala, A.P. Fitzgerald, S. Sans, A. Menotti, G. De Backer, et al., Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project, Eur. Heart J. 24 (2003) 987–1003. [12] M.K. Aktas, V. Ozduran, C.E. Pothier, R. Lang, M.S. Lauer, Global risk scores and exercise testing for predicting all-cause mortality in a preventive medicine program, JAMA 292 (2004) 1462–1468. [13] E.E. Arts, C. Popa, A.A. Den Broeder, A.G. Semb, T. Toms, G.D. Kitas, et al., Performance of four current risk algorithms in predicting cardiovascular events in patients with early rheumatoid arthritis, Ann. Rheum. Dis. 74 (2015) 668–674. [14] E.E. Arts, C.D. Popa, A.A. Den Broeder, R. Donders, A. Sandoo, T. Toms, et al., Prediction of cardiovascular risk in rheumatoid arthritis: performance of original and adapted SCORE algorithms, Ann. Rheum. Dis. 75 (2016) 674–680. [15] C. Gomez-Vaquero, A. Corrales, A. Zacarias, J. Rueda-Gotor, R. Blanco, C. GonzalezJuanatey, et al., SCORE and REGICOR function charts underestimate the cardiovascular risk in Spanish patients with rheumatoid arthritis, Arthritis Res. Ther. 15 (2013) R91. [16] C.S. Crowson, E.L. Matteson, V.L. Roger, T.M. Therneau, S.E. Gabriel, Usefulness of risk scores to estimate the risk of cardiovascular disease in patients with rheumatoid arthritis, Am. J. Cardiol. 110 (2012) 420–424. [17] V.K. Kawai, C.P. Chung, J.F. Solus, A. Oeser, P. Raggi, C.M. Stein, Brief report: the ability of the 2013 American College of Cardiology/American Heart Association cardiovascular risk score to identify rheumatoid arthritis patients with high coronary artery calcification scores, Arthritis Rheum. 67 (2015) 381–385. [18] R. Agca, S.C. Heslinga, S. Rollefstad, M. Heslinga, I.B. McInnes, M.J. Peters, et al., EULAR recommendations for cardiovascular disease risk management in patients with rheumatoid arthritis and other forms of inflammatory joint disorders: 2015/2016 update, Ann. Rheum. Dis. (2016). [19] S. Dadoun, N. Zeboulon-Ktorza, C. Combescure, M. Elhai, S. Rozenberg, L. Gossec, et al., Mortality in rheumatoid arthritis over the last fifty years: systematic review and meta-analysis, Joint Bone Spine 80 (2013) 29–33. [20] M.F. Piepoli, A.W. Hoes, S. Agewall, C. Albus, C. Brotons, A.L. Catapano, et al., European guidelines on cardiovascular disease prevention in clinical practice: the Sixth Joint Task Force of the European Society of Cardiology and other societies on cardiovascular disease prevention in clinical practice (constituted by representatives of 10 societies and by invited experts): developed with the special contribution of the European Association for Cardiovascular Prevention & Rehabilitation (EACPR), Eur. Heart J. (2016) 2016.
[21] A. Soureti, R. Hurling, P. Murray, W. van Mechelen, M. Cobain, Evaluation of a cardiovascular disease risk assessment tool for the promotion of healthier lifestyles, Eur. J. Cardiovasc. Prev. Rehabil. 17 (2010) 519–523. [22] J.I. Cuende, N. Cuende, J. Calaveras-Lagartos, How to calculate vascular age with the SCORE project scales: a new method of cardiovascular risk evaluation, Eur. Heart J. 31 (2010) 2351–2358. [23] M.T. Cooney, E. Vartiainen, T. Laatikainen, D. De Bacquer, C. McGorrian, A. Dudina, et al., Cardiovascular risk age: concepts and practicalities, Heart 98 (2012) 941–946. [24] E. Ikdahl, S. Rollefstad, G. Wibetoe, A. Salberg, D.M. Soldal, I.C. Olsen, T.K. Kvien, G. Haugeberg, A.G. Semb, How to implement cardiovascular disease risk assessment for patients with inflammatory joint diseases in daily rheumatology practice: an overview of a nationwide Norwegian project [abstract], Arthritis Rheum. 68 (suppl 10) (2016) http://acrabstracts.org/abstract/how-to-implement-cardiovascular-disease-risk-assessment-for-patients-with-inflammatory-joint-diseases-indaily-rheumatology-practice-an-overview-of-a-nationwide-norwegian-project/ (Accessed October 24, 2017). [25] M.L. Prevoo, M.A. van't Hof, H.H. Kuper, M.A. van Leeuwen, L.B. van de Putte, P.L. van Riel, Modified disease activity scores that include twenty-eight-joint counts. Development and validation in a prospective longitudinal study of patients with rheumatoid arthritis, Arthritis Rheum. 38 (1995) 44–48. [26] C. Lukas, R. Landewe, J. Sieper, M. Dougados, J. Davis, J. Braun, et al., Development of an ASAS-endorsed disease activity score (ASDAS) in patients with ankylosing spondylitis, Ann. Rheum. Dis. 68 (2009) 18–24. [27] K. Groenewegen, H. Den Ruijter, G. Pasterkamp, J. Polak, M. Bots, S.A. Peters, Vascular age to determine cardiovascular disease risk: a systematic review of its concepts, definitions, and clinical applications, Eur. J. Prev. Cardiol. 23 (2016) 264–274. [28] I. van Dis, J.M. Geleijnse, J.M. Boer, D. Kromhout, H. Boshuizen, D.E. Grobbee, et al., Effect of including nonfatal events in cardiovascular risk estimation, illustrated with data from The Netherlands, Eur. J. Prev. Cardiol. 21 (2014) 377–383. [29] A.G. Semb, E. Ikdahl, J. Hisdal, I.C. Olsen, S. Rollefstad, Exploring cardiovascular disease risk evaluation in patients with inflammatory joint diseases, Int. J. Cardiol. 223 (2016) 331–336. [30] A. Corrales, C. Gonzalez-Juanatey, M.E. Peiro, R. Blanco, J. Llorca, M.A. Gonzalez-Gay, Carotid ultrasound is useful for the cardiovascular risk stratification of patients with rheumatoid arthritis: results of a population-based study, Ann. Rheum. Dis. 73 (2014) 722–727. [31] D.H. Solomon, J. Kremer, J.R. Curtis, M.C. Hochberg, G. Reed, P. Tsao, et al., Explaining the cardiovascular risk associated with rheumatoid arthritis: traditional risk factors versus markers of rheumatoid arthritis severity, Ann. Rheum. Dis. 69 (2010) 1920–1925. [32] P.M. Ridker, M. Cushman, M.J. Stampfer, R.P. Tracy, C.H. Hennekens, Inflammation, aspirin, and the risk of cardiovascular disease in apparently healthy men, N. Engl. J. Med. 336 (1997) 973–979. [33] C. Roubille, V. Richer, T. Starnino, C. McCourt, A. McFarlane, P. Fleming, et al., The effects of tumour necrosis factor inhibitors, methotrexate, non-steroidal antiinflammatory drugs and corticosteroids on cardiovascular events in rheumatoid arthritis, psoriasis and psoriatic arthritis: a systematic review and meta-analysis, Ann. Rheum. Dis. 74 (2015) 480–489. [34] S. Trelle, S. Reichenbach, S. Wandel, P. Hildebrand, B. Tschannen, P.M. Villiger, et al., Cardiovascular safety of non-steroidal anti-inflammatory drugs: network metaanalysis, BMJ c7086 (2011) 342. [35] L. Wei, T.M. MacDonald, B.R. Walker, Taking glucocorticoids by prescription is associated with subsequent cardiovascular disease, Ann. Intern. Med. 141 (2004) 764–770. [36] S.L. Westlake, A.N. Colebatch, J. Baird, P. Kiely, M. Quinn, E. Choy, et al., The effect of methotrexate on cardiovascular disease in patients with rheumatoid arthritis: a systematic literature review, Rheumatology (Oxford) 49 (2010) 295–307. [37] H.K. Choi, M.A. Hernán, J.D. Seeger, J.M. Robins, F. Wolfe, Methotrexate and mortality in patients with rheumatoid arthritis: a prospective study, Lancet 359 (2002) 1173–1177. [38] C. Barnabe, B.J. Martin, W.A. Ghali, Systematic review and meta-analysis: anti-tumor necrosis factor α therapy and cardiovascular events in rheumatoid arthritis, Arthritis Care Res. 63 (2011) 522–529. [39] S.L. Westlake, A.N. Colebatch, J. Baird, N. Curzen, P. Kiely, M. Quinn, et al., Tumour necrosis factor antagonists and the risk of cardiovascular disease in patients with rheumatoid arthritis: a systematic literature review, Rheumatology (Oxford) 50 (2011) 518–531. [40] L.G. Kvalvik, R.M. Nilsen, R. Skjaerven, S.E. Vollset, O. Midttun, P.M. Ueland, et al., Self-reported smoking status and plasma cotinine concentrations among pregnant women in the Norwegian Mother and Child Cohort Study, Pediatr. Res. 72 (2012) 101–107.