- Email: [email protected]
Identification, risk assessment, and management of patients with atrial fibrillation in a large primary care cohort
International Journal of Cardiology 254 (2018) 119–124
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Identification, risk assessment, and management of patients with atrial fibrillation in a large primary care cohort Katrina K. Poppe a,b,⁎,1, Robert N. Doughty b,1, Matire Harwood c,1, P. Alan Barber b,1, Jeff Harrison d,1, Rod Jackson a,1, Sue Wells a,1 a
School of Population Health, University of Auckland, New Zealand Department of Medicine, University of Auckland, New Zealand c Te Kupenga Hauora Māori, University of Auckland, New Zealand d School of Pharmacy, University of Auckland, New Zealand b
a r t i c l e
i n f o
Article history: Received 22 August 2017 Received in revised form 2 November 2017 Accepted 13 November 2017 Keywords: Atrial fibrillation Stroke risk Cardiovascular risk Primary care Electronic health record
a b s t r a c t Background: Atrial fibrillation (AF) is associated with increased risk of cardiovascular disease (CVD) complications including stroke. We investigated the assessment and management of cardiovascular risk among patients with AF aged 35–74 years, by ethnic group, in a large cohort of people receiving a CVD risk assessment in primary care (PREDICT). Methods: PREDICT was linked to national dispensing, hospitalisation and mortality records. AF was present if recorded in PREDICT or during a prior hospitalisation; medications were those dispensed ≤6 months before or after a PREDICT assessment; the CHA2DS2-VASc score and a New Zealand (NZ) adjusted Framingham CVD risk were calculated. Data were linked to outcomes of stroke or major adverse cardiovascular event (MACE). Results: 12,739 (2.8%) of 447,020 people aged 35–74 years had AF. Māori, the indigenous population of NZ, had the highest proportion of AF, which by age group, was similar to that among Europeans 10 years older. 77% were at high stroke risk, of whom 42% received anticoagulation; 54% were at high CVD risk, of whom 67% received both lipid- and blood pressure-lowering medication. Per category of predicted risk, stroke risk was overestimated and risk of MACE was underestimated. Conclusions: The burden of AF and risk factors differed by ethnic group thus recommendations to screen for AF above a universal age threshold may introduce inequity in the detection and management of associated risk. The high burden of comorbidities at younger ages among many ethnic groups contributes to the poor performance of available risk assessment tools, further compounding potential inequity. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Atrial fibrillation (AF) is the most common abnormal heart rhythm encountered in clinical practice. An estimated 33 million people worldwide have diagnosed AF although many more will have undiagnosed or subclinical AF [1,2] and the prevalence increases with age. Compared with the general population, AF is associated with at least a 5-fold increased risk of stroke [3,4], but also with a significantly greater risk of heart failure, admission to hospital, or death [4]. A focus of management for patients with AF is the assessment of thromboembolic risk, in particular that of stroke; however the assessment and management of all cardiovascular disease (CVD) risk is at least as important [5]. Atrial fibrillation and vascular disease share ⁎ Corresponding author at: Epidemiology and Biostatistics, University of Auckland, Building 730 Tamaki Campus, Morrin Road, Private Bag 92019, Auckland 1142, New Zealand. E-mail address: [email protected] (K.K. Poppe). 1 This author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.
https://doi.org/10.1016/j.ijcard.2017.11.045 0167-5273/© 2017 Elsevier B.V. All rights reserved.
many of the same risk factors and often co-exist [6]. Among patients with atherosclerosis in the REACH Registry, CVD risk was higher among patients with AF than those without [7]. Conventional risk assessment scores may not reliably incorporate AF in risk prediction. Clinical guidelines for the CVD risk management of patients with AF include strategies that can substantially decrease morbidity and mortality, particularly reducing the risk of stroke. Anticoagulation is advised for all people with AF unless they're at clearly low thromboembolic risk [8–10], and BP- and lipid-lowering medication is recommended for patients at high CVD risk [11–13]. The prevalence of AF is highest in the elderly, increasing from 0.7% among Europeans aged 55–59 years, to over 18% for those aged N85 years [2], and recommendations are to screen for AF in people aged ≥65 years or older [8,13,14]. Yet there is a lack of evidence about AF among younger adults, whether management of stroke and CVD risk is optimised in these patients, and about the accuracy of risk prediction. Similarly, much of our understanding about the prevalence and impact of AF comes from white or Caucasian populations. Data are limited for non-Caucasians; however Indigenous Australians develop
120
K.K. Poppe et al. / International Journal of Cardiology 254 (2018) 119–124
AF approximately 20 years earlier than non-Indigenous Australians [15], and among Māori, the indigenous population of New Zealand (NZ), 30% of people aged 80–90 years had AF compared with 21% of non-Māori aged 85 years [16]. As AF is a significant risk factor for adverse vascular outcomes, and the prevalence of AF appears to differ by age among some indigenous populations, AF is likely to contribute to disparities in clinical outcomes in these groups. In common with many countries in the world, NZ has a multi-ethnic population, with differing age distributions, co-morbidity burdens, and disparities in clinical outcomes by ethnicity. Internationally, a “racial paradox” between risk factors and AF has been described, where the non-Caucasian groups that have been studied tend to have a higher burden of risk factors for AF than Caucasian or Western populations, yet have a lower prevalence of AF [17]. Such discordance between risk factors and AF has not been observed in Indigenous Australians [18], and is not suggested among NZ Māori, who are identified at a population level as having a high burden of risk factors for CVD [13]. Using a large primary care cohort of routine CVD risk assessments which had been linked to national hospitalisation and dispensing records, we investigated the identification, vascular risk assessment and management of patients with known AF aged 35–74 years, by ethnic group.
loop diuretic on at least 3 occasions in the 5 years prior to the risk assessment, and hypertension was defined as dispensing of at least one BP lowering medication in the 6 months prior to the index risk assessment or a mean systolic BP of ≥150 mm Hg at the time of risk assessment. Vascular disease was defined as a history of CHD, PVD, or revascularisation procedure. A NZ-adjusted Framingham risk score was used to assess 5-year CVD risk (of CV death, non-fatal myocardial infarction (MI), stroke, or other vascular event) [13,21]. The NZ adjustments are described in Appendix A. A 5-year CVD risk ≥15% was used to indicate high CVD risk, 5–15% is intermediate risk and b5% is low risk. Assessment of CVD risk is recommended for people aged 35–74 years depending on sex and ethnicity [13]. Ethnicity is self-reported and defined according to a national prioritisation protocol in the order: Māori, Pacific, Indian, Chinese/other non-Indian Asian, European (Caucasians of European descent). In addition to lifestyle modification, clinical guidelines advise offering pharmacotherapy with blood pressure (BP) lowering and lipid lowering medications for patients at N15% 5-year CVD risk and consideration of aspirin for those at N20% risk (roughly equivalent to 40% 10-year CVD risk) [13]. Outcomes assessed for this study were non-fatal ischaemic or haemorrhagic stroke, and a composite of major adverse CVD events (MACE) defined as non-fatal MI, stroke, HF, or all-cause death. Analyses of outcomes were limited to patients who experienced an event or who had at least 1 year of follow-up since CVD risk assessment. Analyses were performed using R statistical software v3.2.3. Proportions were compared using the chi-squared test of proportions. The cohort study and research process was approved by the Northern Region Ethics Committee Y in 2003 (AKY/03/12/314) with subsequent annual approval by the National Multi Region Ethics Committee since 2007 (MEC/07/19/EXP).
2. Method
3. Results
The PREDICT cohort study contains information obtained during routine risk assessment for CVD and has been described previously [19]. Briefly, PREDICT is a web-based decision support programme that has been integrated with the most commonly used electronic practice management systems in NZ primary care. When PREDICT is used by a practitioner to estimate CVD risk for a patient, a risk profile is electronically stored both in the patient record and anonymously on a central database. With the permission of health providers, this profile is linked to an encrypted National Health Index number (eNHI) and made available to researchers at the University of Auckland. For the current study, PREDICT records were limited to patients with AF aged 35–74 years, between August 2002–October 2015, and linked to national pharmaceutical dispensing and ICD-10-AM coded national public hospital discharge records via the individual's eNHI. Only the first risk assessment for each patient was included. Atrial fibrillation was present if “ECG confirmed AF” had been recorded in PREDICT, or if AF or atrial flutter had been coded (ICD-10 I48) during a prior hospitalisation. Anticoagulation was defined as dispensing of warfarin or dabigatran, as other non-vitamin K antagonist oral anticoagulants were not widely available in NZ during the time of data collection. Similarly, the majority of anticoagulation was with warfarin (86%) as dabigatran became available for clinical use in July 2011. Medications were included if they were dispensed within 6 months before or after the PREDICT risk assessment. The CHA2DS2-VASc score [20] was calculated using information available in the database and used to assess thromboembolic risk. The score assigns points on the basis of age, sex, and medical history, specifically heart failure (HF), hypertension, diabetes, previous stroke, transient ischaemic attack (TIA) or systemic embolism, and vascular disease. A CHA2DS2-VASc score ≥ 2 was used to indicate a high thromboembolic risk. Cliniciandefined diagnoses of HF or hypertension were not available in the PREDICT database. Thus the presence of HF was defined as a prior hospitalisation for HF or dispensing of a
12,739 of 447,020 (2.8%) people aged 35–74 years who had received a routine CVD risk assessment in primary care had a recorded diagnosis of AF. Half (54%) of those with AF were aged b 65 years and one third were female, 78% had hypertension (as defined in the methods), 40% a history of vascular disease, 31% a history of HF, and 30% had diabetes (Table 1). 3.1. Proportion of AF and co-morbidities AF occurred more frequently among Māori compared with nonMāori at all ages, with the proportion of Māori with AF similar to that among Europeans 10 years older (Fig. 1, Supplementary Table 1). Māori, Pacific and Indian patients were generally younger than European and Chinese/Other Asian patients (consistent with national CVD risk screening recommendations [13]), with two thirds of these groups aged b65 years compared with half of Europeans (Table 1). They had a significantly greater burden of comorbidities: 48%, 42%, and 59% of Māori, Pacific and Indian patients had known vascular disease compared with 36% of Europeans and 33% of Chinese/Other Asians (all p-values b 0.001), and 44%, 48% and 56% had diabetes compared with 20–29% of Europeans and Chinese/Other Asians
Table 1 Characteristics of the cohort with atrial fibrillation.
n Age, years, mean (SD) b65 years Female Current smoker Hypertension Prior vascular disease Prior stroke or TIA Heart failure Diabetes HbA1c ≥ 64 mmol/mol Duration ≥ 10 years With nephropathy SBP, mm Hg, mean (SD) Extreme BP* TC:HDL, mean (SD) Extreme cholesterol**
Total
Māori
Pacific
Indian
Chinese/other Asian
European
12,739 62 (8.9) 6833 (54) 4361 (34) 1446 (11) 9986 (78) 5073 (40) 1670 (13) 3943 (31) 3759 (30) 990 (26) 850 (23) 338 (9) 129 (15.9) 407 (3) 3.9 (1.20) 149 (1)
2669 60 (9.3) 1721 (64) 1110 (42) 602 (23) 2227 (84) 1276 (48) 504 (19) 1309 (49) 1169 (44) 359 (31) 270 (23) 162 (14) 129 (17.1) 138 (5) 4.0 (1.26) 46 (2)
1436 58 (10.0) 986 (69) 631 (44) 181 (13) 1169 (81) 608 (42) 256 (18) 678 (47) 682 (48) 229 (34) 149 (22) 93 (14) 128 (16.8) 68 (5) 3.9 (1.17) 11 (1)
302 60 (9.4) 186 (62) 112 (37) 16 (5) 265 (88) 179 (59) 41 (14) 117 (39) 169 (56) 57 (34) 52 (31) 16 (10) 130 (18.0) 92 (3) 3.8 (1.14) 3 (1)
474 63 (8.8) 249 (52) 185 (39) 28 (6) 365 (77) 154 (33) 48 (10) 74 (16) 136 (29) 19 (14) 20 (15) 2 (2) 125 (15.9) 5 (1) 3.7 (1.04) 2 (0.4)
7858 64 (8.1) 3691 (47) 2323 (30) 619 (8) 5960 (76) 2856 (36) 821 (10) 1765 (23) 1603 (20) 326 (20) 359 (22) 65 (4) 130 (15.2) 187 (2) 3.9 (1.20) 87 (1)
Values are n (%) unless otherwise stated. BP = blood pressure, SBP = systolic BP, TC:HDL = ratio of total cholesterol to high density lipoprotein, TIA = transient ischaemic attack. *BP consistently ≥ 170/100; **TC ≥ 8 mmol/L or TC:HDL ≥ 8.
K.K. Poppe et al. / International Journal of Cardiology 254 (2018) 119–124
20
Men
20
Women
121
Maori
15
European Indian
35-44
10 0
0
5
10
Asian
5
Atrial fibrillation, %
15
Pacific
45-54
55-64
65-74
35-44
45-54
Age, years
55-64
65-74
Age, years
Fig. 1. Proportion of people in the PREDICT cohort with atrial fibrillation, by sex, ethnicity and age group.
respectively (all p-values b0.001). Diabetes was less well controlled (HbA1c ≥64 mmol/mol among 31–34% of Māori, Pacific and Indian vs 14–20% of Europeans and Chinese/Other Asians, all p-values b 0.001) and renal complications of diabetes more prevalent (nephropathy in 10–14% vs 2–4%, all p-values b0.003). These differences were present in all age groups (Supplementary Table 2). 3.2. Thromboembolic risk High thromboembolic risk (CHA2DS2-VASc score ≥ 2) was seen in 77% of the cohort. A higher proportion of Māori, Pacific, and Indian people were at high risk compared with Chinese/Other Asian and
European patients, consistent with the higher burden of comorbidities seen in these groups (Table 2, Supplementary Fig. 1). This difference was demonstrated at all ages, although Chinese/Other Asian had a higher proportion of each age group at high risk than Europeans. Anticoagulation therapy had been dispensed to less than half of those with AF, with only 49% of those at high thromboembolic risk and 19% of the 2997 patients who were at low risk (CHA2DS2-VASc b2) dispensed warfarin or dabigatran. Ethnic differences in those treated with anticoagulation therapy were also seen, with 34% of Indian people and 50% of Māori treated (Table 2). 3508 (70%) of the high stroke risk patients not treated with anticoagulation were prescribed aspirin.
Table 2 Vascular risk and management in patients with atrial fibrillation.
n Thromboembolic risk At high risk Women at high risk At high risk by age group 35–44 years 45–54 years 55–64 years 65–74 years Anticoagulation In those at high risk Cardiovascular risk At high risk Women at high risk At high risk by age group 35–44 years 45–54 years 55–64 years 65–74 years Blood pressure lowering In those at high risk Lipid lowering In those at high risk Aspirin In those at high risk
Total
Māori
Pacific
Indian
Chinese/other Asian
European
12,739
2669
1436
302
474
7858
9742 (77) 4010 (41)
2279 (85) 1050 (46)
1203 (84) 599 (50)
264 (87) 105 (40)
357 (75) 162 (45)
5639 (72) 2094 (37)
240 (52) 1100 (52) 2811 (66) 5591 (95) 5350 (42) 4766 (49)
99 (58) 429 (72) 822 (86) 929 (98) 1325 (50) 1225 (54)
92 (64) 282 (77) 386 (81) 443 (98) 673 (47) 620 (52)
8 (44) 52 (84) 89 (84) 115 (99) 104 (34) 98 (37)
9 (45) 27 (43) 105 (63) 216 (96) 180 (38) 159 (45)
32 (30) 310 (30) 1409 (55) 3888 (93) 3068 (39) 2664 (47)
6838 (54) 2113 (31)
1807 (68) 691 (38)
854 (60) 337 (40)
208 (69) 66 (32)
206 (44) 69 (34)
3763 (48) 950 (25)
152 (33) 788 (37) 2108 (50) 3790 (64) 10,252 (81) 6117 (90) 6922 (54) 4945 (72) 6033 (47) 3933 (58)
74 (43) 307 (52) 676 (71) 750 (79) 2295 (86) 1644 (91) 1563 (59) 1240 (69) 1294 (49) 1006 (56)
47 (33) 162 (44) 297 (63) 348 (77) 1199 (84) 771 (90) 821 (57) 599 (70) 673 (47) 466 (55)
2 (11) 34 (55) 76 (72) 96 (83) 274 (91) 201 (97) 229 (76) 182 (88) 184 (61) 145 (70)
6 (30) 15 (24) 58 (35) 127 (56) 381 (80) 187 (91) 264 (56) 161 (78) 220 (46) 123 (60)
23 (21) 270 (26) 1001 (39) 2469 (59) 6103 (78) 3314 (88) 4045 (52) 2763 (73) 3662 (47) 2193 (58)
Values are n (%) unless otherwise specified.
122
K.K. Poppe et al. / International Journal of Cardiology 254 (2018) 119–124 Table 3 Event rates among patients with AF and at least one year of follow-up (n = 12,326).
Stroke rates within 1 year Overall event rate, % By risk groupa, % 0.8 1.8 2.7 ≥3.2 MACE rates within 5 years Overall event rate, % By risk group, % b5 5–10 10–15 ≥15 Known N20% risk
Whole group n = 12,326
Māori n = 2573
Pacific n = 1384
Indian n = 291
Chinese/other Asian n = 458
European n = 7620
1.2
1.9
1.4
1.0
2.0
0.9
0.2 0.3 0.8 1.7
1.8 0.8 1.3 2.3
0 0 0 2.1
0 0 0 1.5
0 1.3 2.5 2.2
0 0.2 0.7 1.4
25
38
36
24
12
19
8 14 20 26 36
0 20 31 35 48
0 27 26 34 45
0 10 5 17 33
8 11 3 22 16
8 10 15 20 30
a Stroke risk groups are represented by the mean risk where 0.8% risk correlates to a CHADS-VASc score of 0, 1.8% risk to a score of 1, 2.7% risk to a score of 2, and ≥3.2 to a score of ≥3 [8].
Stroke occurred in 709 of 12,326 (6%) patients with AF who either experienced an event or had at least one year of follow-up within 1 to 4122 days of baseline risk assessment. For the entire cohort, 150 (1.2%) had a stroke in the first year of follow-up, with Māori and Chinese/Other Asian having the highest proportion of stroke events (1.9% and 2.0% respectively; Table 3). Results differed slightly for the sub-group of patients who were not receiving anticoagulant therapy (n = 7160): 63 (0.9%) had a stroke however the highest rates occurred among Māori and Pacific (1.4% and 1.6% respectively; Supplementary Table 3a). The CHA2DS2-VASc score overestimated stroke risk in all ethnic groups, with the exception of Māori with a CHA2DS2-VASc score of 0 (approximately 0.8% risk), where the actual event rate was 1.8% (Table 3). The degree of overestimation was least for Chinese/ Other Asian patients.
3.3. Cardiovascular risk Estimated 5-year CVD risk was 15% or more, indicating high CVD risk, in 54% of the cohort (Table 2). As seen for thromboembolic risk, and consistent with their higher burden of comorbidities, a higher proportion of Māori, Pacific, and Indian patients were at high risk compared with Chinese/Other Asian and European patients (Table 2). This difference was demonstrated at all ages, particularly among Māori and Indian patients aged ≥45 years. Over half of the patients with AF were dispensed at least one BP lowering or lipid lowering medication, and 47% received aspirin. Of those at high CVD risk, 90% received BP lowering and 72% received lipid lowering medication (Table 2); 4561 (67%) received both lipidand BP-lowering however 451 (7%) received neither. The highest rates of medication use, including for those at high CVD risk, occurred among Indian patients. A major adverse cardiovascular event (MACE) occurred in 3575 (29%) patients, within 0–4167 days of their baseline risk assessment. The 5-year rate of a MACE was 25%, with the highest rates occurring among Māori and Pacific (38 and 36% respectively) (Table 3). In contrast to stroke risk, 5-year CVD risk was underestimated: 125 (8%) of patients at low risk (b5%) experienced a MACE within 5 years, as did 672 (16%) of patients at intermediate risk (5–15%). Patients at high risk, whether estimated by the CVD risk equation or categorised as high risk due to existing CVD or other high risk conditions, had MACE rates of 26% and 36% respectively. Risk was similarly underestimated among patients who did not receive BP- and lipidlowering medication, particularly among Māori and Pacific patients (Supplementary Table 3b).
4. Discussion This study has shown that the burden of AF in a primary care cohort differs not only by age, but by age within ethnic group. Contrary to the racial paradox that has been observed in some non-European populations [17], Māori, the indigenous population of NZ, and Pacific peoples, had both a higher burden of risk factors for AF and a higher burden of AF than Europeans/Caucasians at all ages. In contrast, Indians had a higher burden of risk factors and a lower burden of AF than Europeans. Our cohort had a generally higher prevalence of risk factors than was described in a study of AF in NZ practices not included in our cohort [22]; however that work used history as recorded in the primary care record only (which are not connected to hospitalisation records) and had significantly lower representation of non-European ethnic groups. For all patients with AF, there are clear disparities between estimated vascular risk and medication use. Over three quarters of the patients in this study were estimated to be at high risk of stroke within one year, and half were at high risk of a CVD event in the next five years. The majority of these high risk patients received a BP lowering medication, three quarters received a statin but only half of those at estimated high stroke risk received anticoagulation therapy (only one third of Indian patients at high risk). The associations between AF and vascular disease, including but not limited to stroke, reinforce the importance of aggressive management of common risk factors [4]. These findings show that there is a clear opportunity to improve evidence-based practice. Current risk assessment tools were inaccurate in this population: three quarters were predicted to be at high risk of stroke in the next year yet only 1.2% experienced a stroke; conversely, a MACE occurred in 34% of those estimated at high 5-year CVD risk but also in 16% at intermediate (5–15%) risk and 8% at low (b5%) CVD risk. Similar risk factors are important in predicting each outcome and are included in each risk score, yet one score overestimates risk (of stroke) and the other underestimates risk (of MACE). In a recent meta-analysis of over 580,000 subjects with AF, Odutayo and colleagues [4] noted that evidencebased strategies to reduce the burden of events in people with AF should include discussion of predicted CVD risk. We agree, however our data demonstrate that the current risk assessment tools need to be optimised for patients with AF in order to more accurately identify those at higher risk. Thromboembolic risk for people with AF is typically assessed using the CHA2DS2-VASc score [20]. While this scoring system is used in many countries, it is important to note that it is based on data from European populations. Yet as we have shown, age, risk, and risk profiles
K.K. Poppe et al. / International Journal of Cardiology 254 (2018) 119–124
differ for different ethnic groups. Compared with Caucasians, AfricanAmericans are less likely to develop AF [23], yet once they have it, have a higher risk of stroke [24] and adding African-American ethnicity to the CHA2DS2-VASc score more accurately predicted their higher risk in a multi-ethnic cohort of people aged N 65 [24]. Risk scores can be modified in a number of ways to suit a population that is different from that used to derive it. Recalibration applies a global factor to the score to increase or decrease results in everyone, and adding new risk factors, such as ethnicity, improves discrimination of the group affected by that new variable. Although useful, these are crude ways to improve risk prediction at the level of the individual. Among Europeans, the proportion with AF tends to increase above the age of 65 years; among Asians, an annual stroke risk of approximately 1.8% for younger AF patients led to the development of a modified CHA2DS2-VASc score that assigned 1 point for age between 50 and 74 years rather than 65–74 years [25]. Older people are more likely to have had a longer exposure to comorbidities than younger people, thus the risk associated with each comorbidity is greater in a score developed for people presenting with AF later in life. The high stroke risk seen in our non-European patients aged b 65, where no points are assigned for age in the standard CHA2DS2-VASc score, reflect the high burden of comorbidities in this group. A high comorbidity burden may indeed increase risk of an adverse event, but the overestimation of 1-year stroke risk seen in our cohort suggests that the amount of risk assigned to each comorbidity by the CHA2DS2-VASc score was greater than needed. If the score was developed in a cohort that spanned the age range of adults with AF, the differing contributions of comorbidities by age may be better represented. Almost half of the cohort with AF had existing vascular disease, placing them at high risk of a future event. For those without CVD, the NZ adjusted Framingham score was used to estimate the 5-year risk of a CVD event; however this score does not include an adjustment for AF. Examining the rates of MACE by ethnicity showed that CVD risk among Indian people with AF was only marginally underestimated. The NZ adjustments to the Framingham score add 5% to the absolute risk for people of Indian descent in an effort to capture the known higher CVD risk among this group [13]. This 5% adjustment is also made for Māori and Pacific peoples, yet CVD risk was still underestimated for these patients in our cohort. The burden of comorbidities was similarly high across all three ethnic groups but a key difference is that a smaller proportion of Indian people had AF than Māori or Pacific (or Europeans) at each age group. As age is a strong driver of CVD risk, the relatively older age of Indian patients with AF in this cohort is accommodated by the current risk score. New risk equations for the primary prevention of CVD in NZ are being developed that include AF as a predictor, thus CVD risk assessment may improve in this patient group. Health policy and practice needs to shift away from its focus on silos of disease-specific risk and instead develop a more holistic, individualised approach to risk assessment and management that is relevant to the life-course of risk and disease. AF occurs for individuals in the setting of a number of CV risk factors and associated clinical manifestations of CV disease. Thus a process of systematic detection [14] and appropriate management in the context of overall CVD risk must be considered for people with AF. After analysing outcomes on almost 10,000,000 people, 6% of whom had AF, Odutayo and colleagues recommended that “the burden of non-stroke events in adults with AF would benefit from a focus on primary prevention and management of CV risk factors” [4]. Clinical practice guidelines need to reflect this life-course approach, and be augmented by continuous refinement of risk assessment in parallel to practice change.
123
eligible population to have had a CVD risk assessment in the past 5 years [26]. In addition, software integrated with the electronic health record has enabled the collection of relevant data on a large and representative sample of adults in contact with the health system, which can be accurately linked to national databases of dispensing and outcome using a unique patient identification number. As a result, these data represent routine clinical practice, include large numbers of non-European patients, and have complete data on many factors, including outcomes. It remains, though, that the cohort represents people who are eligible for a CVD risk assessment (based on age criteria) and in whom AF has been diagnosed. Although representative of the ethnic mix undergoing CVD risk assessment, Indian and Chinese/Other Asian groups are represented by a relatively small number of patients in this cohort. These groups also had a low event rate, limiting the robustness of the comparisons between these and the other ethnic groups. All outcomes are defined from national administrative databases of mortality and public hospital admissions, so if a person experienced an event when out of the country, or if it resulted in admission only to a private hospital, it will not be recorded. Fewer than 10% of all women were defined as being at low thromboembolic risk. The 2016 ESC guidelines for the management of AF effectively remove the additional risk ascribed to women by the CHA2DS2-VASc score by defining high risk among women as a score of 3 or more (and among men as a score of 2 or more) [1]. The current study defined high stroke risk as a score of 2 or more, regardless of sex, as that was the recommendation during the period of data collection [8,10]. 5. Conclusions These data, from a contemporary multi-ethnic population, demonstrate substantial variability by ethnic group in the age at which AF occurs and in the associated risk factors for vascular events. Overestimation of stroke risk and underestimation of CVD risk, contributed to by the high burden of comorbidities at younger ages in many ethnic groups, signals a need to improve the accuracy of risk prediction in this patient group. Overall clinical event rates remain high, and while the majority of patients at high estimated vascular risk received BPlowering medication, use of anticoagulant and lipid-lowering medications for these patients was suboptimal. Systematic approaches to improving cardiovascular risk assessment and management for patients with AF in the community are urgently required. Author contributions KP and RD conceived and designed this work. SW, RJ, MH and JH were involved in the data collection process; KP designed and performed all analyses. KP and RD interpreted the results and drafted the manuscript. All authors critically revised the manuscript. All gave final approval and agree to be accountable for all aspects of work ensuring integrity and accuracy. Funding The PREDICT CVD study has been supported by funding from the Health Research Council of New Zealand via two project grants (HRC 03/183 and 08/121) and under a programme grant (HRC VIEW 11/800 programme). No funders had a role in the design of the study; the collection, analysis, or interpretation of the data; or the decision to approve publication of the finished manuscript.
4.1. Strengths and limitations Disclosures Over the last 5–10 years, NZ has developed a systematic approach to CVD risk assessment in primary care. This has been supported by a Ministry of Health-led national health target requiring 90% of the
Outside this study, SW has received a Fellowship from the Stevenson Foundation and a research grant from Roche Diagnostics Ltd. All other
124
K.K. Poppe et al. / International Journal of Cardiology 254 (2018) 119–124
authors declare no conflicts of interest. All authors have approved the final article. Appendix A. Cardiovascular risk assessment in New Zealand National clinical guidelines advise the following adjustments are made to the 1991 Framingham 5-year CVD risk estimate: A single 5% increase in absolute risk is added for one of: family history of premature CHD or ischaemic stroke; Māori, Pacific or Indo-Asian ethnicity; or diabetes with: microalbuminuria, duration N10 years, or HbA1c consistently N 64 mmol/mol. CVD risk is considered to be ≥ 15% if total or total: HDL cholesterol are ≥ 8 or if BP is consistently ≥ 170/100 mm Hg. The risk score is not applied to patients with: prior CVD, familial hypercholesterolaemia, defective ApoB or combined dyslipidaemia, or diabetes with overt nephropathy or severe renal impairment. These patients are considered as having a CVD-equivalent risk (i.e. at least N20% risk). The PREDICT cohort represents people who have undergone routine CVD risk assessment in accordance with national guidelines. As people of Māori, Pacific or Indian descent are known to be at high CVD risk (at a population level), guidelines advise that risk should start to be assessed in people identifying with these ethnicities 10 years earlier than for other ethnic groups, i.e. Maori, Pacific, Indian men from 35 years old; Maori, Pacific, Indian women from 45 years old; European, Asian men from 45 years old; European, Asian women from 55 years old. As a result, the PREDICT cohort over-represents high-risk ethnic groups compared with their distribution in the general population. Representation is also influenced by the geographic distribution of where people live, as seen in the relatively high proportions of Pacific, Indian and Asian people in PREDICT. The majority of people in these groups live in the upper half of the North Island of New Zealand, where PREDICT software is used. Aged 35–74 years 2013 NZ populationa PREDICT cohort PREDICT as % of NZ n (%) of total n (%) of total population All Māori Pacific Indian Asian European/Caucasian a
2,090,970 235,860 (11%) 99,645 (5%) 66,152 (3%) 143,303 (7%) 1,546,010 (74%)
446,961 59,185 (13%) 56,147 (13%) 37,170 (8%) 42,934 (10%) 251,525 (56%)
21 25 56 56 30 16
NZ Ministry of Health population using a single prioritised ethnicity per person.
Appendix B. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.ijcard.2017.11.045. References [1] P. Kirchhof, S. Benussi, D. Kotecha, et al., 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS, Eur. Heart J. 37 (2016) 2893–2962.
[2] J. Heeringa, D.A.M. van der Kuip, A. Hofman, et al., Prevalence, incidence and lifetime risk of atrial fibrillation: the Rotterdam study, Eur Heart J. 27 (2006) 949–953. [3] S. Chugh, J. Blackshear, W. Shen, S. Hammill, B. Gersh, Epidemiology and natural history of atrial fibrillation: clinical implications, J. Am. Coll. Cardiol. 37 (2001) 371–378. [4] A.A. Odutayo, C.X. Wong, A.J. Hsiao, S. Hopewell, D.G. Altman, C.A. Emdin, Atrial fibrillation and risks of cardiovascular disease, renal disease, and death: systematic review and meta-analysis, Br. Med. J. 354 (2016) i4482. [5] M. Proietti, C. Laroche, G. Opolski, A.P. Maggioni, G. Boriani, G.Y.H. Lip, ‘Real-world’ atrial fibrillation management in Europe: observations from the 2-year follow-up of the EURObservational Research Programme-Atrial Fibrillation General Registry Pilot Phase, Europace 19 (on behalf of the AF Gen Pilot Investigators) (2017) 722–733. [6] J. Olesen, G. Gislason, C. Torp-Pedersen, G.Y.H. Lip, Atrial fibrillation and vascular disease - a bad combination, Clin. Cardiol. 35 (2012) 15–20. [7] C. Ruff, D. Bhatt, P. Steg, et al., Long-term cardiovascular outcomes in patients with atrial fibrillation and atherothrombosis in the REACH Registry, Int. J. Cardiol. 170 (2014) 413–418. [8] A. Camm, G. Lip, R. De Caterina, et al., 2012 focused update on the ESC guidelines for the management of atrial fibrillation: an update of the 2010 ESC guidelines for the management of atrial fibrillation, Eur. Heart J. 33 (2012) 2719–2747. [9] C.T. January, L.S. Wann, J.S. Alpert, et al., 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation, J. Am. Coll. Cardiol. 64 (2014) e1–76. [10] Best Practice New Zealand, An update on antithrombotic medicines. What does primary care need to know? Best Practice J 67 (2015) 12–24. [11] M. Piepoli, A. Hoes, S. Agewall, et al., 2016 European Guidelines on cardiovascular disease prevention in clinical practice, Eur. Heart J. 37 (2016) 2315–2381. [12] D. Goff, D. Lloyd-Jones, G. Bennett, et al., 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines, Circulation 63 (2014) 2935–2959. [13] New Zealand Guidelines Group, New Zealand Primary Care Handbook 2012, 2012. [14] B. Freedman, J. Camm, H. Calkins, et al., Screening for atrial fibrillation. A report of the AF-SCREEN International Collaboration, Circulation 135 (2017) 1851–1867. [15] C.X. Wong, A.G. Brooks, Y.-H. Cheng, et al., Atrial fibrillation in indigenous and nonindigenous Australians: a cross-sectional study, BMJ Open 4 (2014) e006242. [16] R. Teh, A. Martin, N. Kerse, L. Dyall, R. Doughty, The burden of atrial fibrillation in octogenarians, Heart, Lung and Circulation. 22 (2013) 580–581. [17] M. Amponsah, E. Benjamin, J. Magnani, Atrial fibrillation and race - a contemporary review, Curr Cardiovasc Risk Rep. 7 (2013). [18] J.M. Katzenellenbogen, T.H.K. Teng, D. Lopez, et al., Initial hospitalisation for atrial fibrillation in aboriginal and non-aboriginal populations in Western Australia, Heart 101 (2015) 712–719. [19] S. Wells, T. Riddell, A. Kerr, et al., Cohort profile: the PREDICT cardiovascular disease cohort in New Zealand primary care (PREDICT CVD-19), Int. J. Epidemiol. 46 (1) (2017) 22-22j. [20] G. Lip, R. Nieuwlaat, R. Pisters, D. Lane, J. Crijns, Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk fact based approach. The Euro Heart Survey on Atrial Fibrillation, Chest 137 (2010) 263–272. [21] K. Anderson, P. Odell, P. Wilson, W. Kannel, Cardiovascular disease risk profiles, Am. Heart J. 121 (1991) 293–298. [22] A. Tomlin, H. Lloyd, M. Tilyard, Atrial fibrillation in New Zealand primary care: prevalence, risk factors for stroke and the management of thromboembolic risk, Eur. J. Prev. Cardiol. 24 (2016) 311–319. [23] A. Alonso, N. Roetker, E.Z. Soliman, L. Chen, P. Greenland, S. Heckbert, Prediction of atrial fibrillation in a racially diverse cohort: the Multi-Ethnic Study of Atherosclerosis (MESA), J. Am. Heart Assoc. 5 (2016) e003077. [24] R. Kabra, S. Girotra, M. Vaughan Sarrazin, Refining stroke prediction in atrial fibrillation patients by addition of African-American ethnicity to the CHA2DS2-VASc score, J. Am. Coll. Cardiol. 68 (2016) 461–470. [25] T.-F. Chao, G.Y.H. Lip, C.-J. Liu, et al., Validation of a modified CHA2DS2-VASc score for stroke risk stratification in Asian patients with atrial fibrillation, Stroke 47 (2016) 2462–2469. [26] Ministry of Health, Targeting Diabetes and Cardiovascular Disease, 2011.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Identification, risk assessment, and management of patients with atrial fibrillation in a large primary care cohort Katrina K. Poppe a,b,⁎,1, Robert N. Doughty b,1, Matire Harwood c,1, P. Alan Barber b,1, Jeff Harrison d,1, Rod Jackson a,1, Sue Wells a,1 a
School of Population Health, University of Auckland, New Zealand Department of Medicine, University of Auckland, New Zealand c Te Kupenga Hauora Māori, University of Auckland, New Zealand d School of Pharmacy, University of Auckland, New Zealand b
a r t i c l e
i n f o
Article history: Received 22 August 2017 Received in revised form 2 November 2017 Accepted 13 November 2017 Keywords: Atrial fibrillation Stroke risk Cardiovascular risk Primary care Electronic health record
a b s t r a c t Background: Atrial fibrillation (AF) is associated with increased risk of cardiovascular disease (CVD) complications including stroke. We investigated the assessment and management of cardiovascular risk among patients with AF aged 35–74 years, by ethnic group, in a large cohort of people receiving a CVD risk assessment in primary care (PREDICT). Methods: PREDICT was linked to national dispensing, hospitalisation and mortality records. AF was present if recorded in PREDICT or during a prior hospitalisation; medications were those dispensed ≤6 months before or after a PREDICT assessment; the CHA2DS2-VASc score and a New Zealand (NZ) adjusted Framingham CVD risk were calculated. Data were linked to outcomes of stroke or major adverse cardiovascular event (MACE). Results: 12,739 (2.8%) of 447,020 people aged 35–74 years had AF. Māori, the indigenous population of NZ, had the highest proportion of AF, which by age group, was similar to that among Europeans 10 years older. 77% were at high stroke risk, of whom 42% received anticoagulation; 54% were at high CVD risk, of whom 67% received both lipid- and blood pressure-lowering medication. Per category of predicted risk, stroke risk was overestimated and risk of MACE was underestimated. Conclusions: The burden of AF and risk factors differed by ethnic group thus recommendations to screen for AF above a universal age threshold may introduce inequity in the detection and management of associated risk. The high burden of comorbidities at younger ages among many ethnic groups contributes to the poor performance of available risk assessment tools, further compounding potential inequity. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Atrial fibrillation (AF) is the most common abnormal heart rhythm encountered in clinical practice. An estimated 33 million people worldwide have diagnosed AF although many more will have undiagnosed or subclinical AF [1,2] and the prevalence increases with age. Compared with the general population, AF is associated with at least a 5-fold increased risk of stroke [3,4], but also with a significantly greater risk of heart failure, admission to hospital, or death [4]. A focus of management for patients with AF is the assessment of thromboembolic risk, in particular that of stroke; however the assessment and management of all cardiovascular disease (CVD) risk is at least as important [5]. Atrial fibrillation and vascular disease share ⁎ Corresponding author at: Epidemiology and Biostatistics, University of Auckland, Building 730 Tamaki Campus, Morrin Road, Private Bag 92019, Auckland 1142, New Zealand. E-mail address: [email protected] (K.K. Poppe). 1 This author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.
https://doi.org/10.1016/j.ijcard.2017.11.045 0167-5273/© 2017 Elsevier B.V. All rights reserved.
many of the same risk factors and often co-exist [6]. Among patients with atherosclerosis in the REACH Registry, CVD risk was higher among patients with AF than those without [7]. Conventional risk assessment scores may not reliably incorporate AF in risk prediction. Clinical guidelines for the CVD risk management of patients with AF include strategies that can substantially decrease morbidity and mortality, particularly reducing the risk of stroke. Anticoagulation is advised for all people with AF unless they're at clearly low thromboembolic risk [8–10], and BP- and lipid-lowering medication is recommended for patients at high CVD risk [11–13]. The prevalence of AF is highest in the elderly, increasing from 0.7% among Europeans aged 55–59 years, to over 18% for those aged N85 years [2], and recommendations are to screen for AF in people aged ≥65 years or older [8,13,14]. Yet there is a lack of evidence about AF among younger adults, whether management of stroke and CVD risk is optimised in these patients, and about the accuracy of risk prediction. Similarly, much of our understanding about the prevalence and impact of AF comes from white or Caucasian populations. Data are limited for non-Caucasians; however Indigenous Australians develop
120
K.K. Poppe et al. / International Journal of Cardiology 254 (2018) 119–124
AF approximately 20 years earlier than non-Indigenous Australians [15], and among Māori, the indigenous population of New Zealand (NZ), 30% of people aged 80–90 years had AF compared with 21% of non-Māori aged 85 years [16]. As AF is a significant risk factor for adverse vascular outcomes, and the prevalence of AF appears to differ by age among some indigenous populations, AF is likely to contribute to disparities in clinical outcomes in these groups. In common with many countries in the world, NZ has a multi-ethnic population, with differing age distributions, co-morbidity burdens, and disparities in clinical outcomes by ethnicity. Internationally, a “racial paradox” between risk factors and AF has been described, where the non-Caucasian groups that have been studied tend to have a higher burden of risk factors for AF than Caucasian or Western populations, yet have a lower prevalence of AF [17]. Such discordance between risk factors and AF has not been observed in Indigenous Australians [18], and is not suggested among NZ Māori, who are identified at a population level as having a high burden of risk factors for CVD [13]. Using a large primary care cohort of routine CVD risk assessments which had been linked to national hospitalisation and dispensing records, we investigated the identification, vascular risk assessment and management of patients with known AF aged 35–74 years, by ethnic group.
loop diuretic on at least 3 occasions in the 5 years prior to the risk assessment, and hypertension was defined as dispensing of at least one BP lowering medication in the 6 months prior to the index risk assessment or a mean systolic BP of ≥150 mm Hg at the time of risk assessment. Vascular disease was defined as a history of CHD, PVD, or revascularisation procedure. A NZ-adjusted Framingham risk score was used to assess 5-year CVD risk (of CV death, non-fatal myocardial infarction (MI), stroke, or other vascular event) [13,21]. The NZ adjustments are described in Appendix A. A 5-year CVD risk ≥15% was used to indicate high CVD risk, 5–15% is intermediate risk and b5% is low risk. Assessment of CVD risk is recommended for people aged 35–74 years depending on sex and ethnicity [13]. Ethnicity is self-reported and defined according to a national prioritisation protocol in the order: Māori, Pacific, Indian, Chinese/other non-Indian Asian, European (Caucasians of European descent). In addition to lifestyle modification, clinical guidelines advise offering pharmacotherapy with blood pressure (BP) lowering and lipid lowering medications for patients at N15% 5-year CVD risk and consideration of aspirin for those at N20% risk (roughly equivalent to 40% 10-year CVD risk) [13]. Outcomes assessed for this study were non-fatal ischaemic or haemorrhagic stroke, and a composite of major adverse CVD events (MACE) defined as non-fatal MI, stroke, HF, or all-cause death. Analyses of outcomes were limited to patients who experienced an event or who had at least 1 year of follow-up since CVD risk assessment. Analyses were performed using R statistical software v3.2.3. Proportions were compared using the chi-squared test of proportions. The cohort study and research process was approved by the Northern Region Ethics Committee Y in 2003 (AKY/03/12/314) with subsequent annual approval by the National Multi Region Ethics Committee since 2007 (MEC/07/19/EXP).
2. Method
3. Results
The PREDICT cohort study contains information obtained during routine risk assessment for CVD and has been described previously [19]. Briefly, PREDICT is a web-based decision support programme that has been integrated with the most commonly used electronic practice management systems in NZ primary care. When PREDICT is used by a practitioner to estimate CVD risk for a patient, a risk profile is electronically stored both in the patient record and anonymously on a central database. With the permission of health providers, this profile is linked to an encrypted National Health Index number (eNHI) and made available to researchers at the University of Auckland. For the current study, PREDICT records were limited to patients with AF aged 35–74 years, between August 2002–October 2015, and linked to national pharmaceutical dispensing and ICD-10-AM coded national public hospital discharge records via the individual's eNHI. Only the first risk assessment for each patient was included. Atrial fibrillation was present if “ECG confirmed AF” had been recorded in PREDICT, or if AF or atrial flutter had been coded (ICD-10 I48) during a prior hospitalisation. Anticoagulation was defined as dispensing of warfarin or dabigatran, as other non-vitamin K antagonist oral anticoagulants were not widely available in NZ during the time of data collection. Similarly, the majority of anticoagulation was with warfarin (86%) as dabigatran became available for clinical use in July 2011. Medications were included if they were dispensed within 6 months before or after the PREDICT risk assessment. The CHA2DS2-VASc score [20] was calculated using information available in the database and used to assess thromboembolic risk. The score assigns points on the basis of age, sex, and medical history, specifically heart failure (HF), hypertension, diabetes, previous stroke, transient ischaemic attack (TIA) or systemic embolism, and vascular disease. A CHA2DS2-VASc score ≥ 2 was used to indicate a high thromboembolic risk. Cliniciandefined diagnoses of HF or hypertension were not available in the PREDICT database. Thus the presence of HF was defined as a prior hospitalisation for HF or dispensing of a
12,739 of 447,020 (2.8%) people aged 35–74 years who had received a routine CVD risk assessment in primary care had a recorded diagnosis of AF. Half (54%) of those with AF were aged b 65 years and one third were female, 78% had hypertension (as defined in the methods), 40% a history of vascular disease, 31% a history of HF, and 30% had diabetes (Table 1). 3.1. Proportion of AF and co-morbidities AF occurred more frequently among Māori compared with nonMāori at all ages, with the proportion of Māori with AF similar to that among Europeans 10 years older (Fig. 1, Supplementary Table 1). Māori, Pacific and Indian patients were generally younger than European and Chinese/Other Asian patients (consistent with national CVD risk screening recommendations [13]), with two thirds of these groups aged b65 years compared with half of Europeans (Table 1). They had a significantly greater burden of comorbidities: 48%, 42%, and 59% of Māori, Pacific and Indian patients had known vascular disease compared with 36% of Europeans and 33% of Chinese/Other Asians (all p-values b 0.001), and 44%, 48% and 56% had diabetes compared with 20–29% of Europeans and Chinese/Other Asians
Table 1 Characteristics of the cohort with atrial fibrillation.
n Age, years, mean (SD) b65 years Female Current smoker Hypertension Prior vascular disease Prior stroke or TIA Heart failure Diabetes HbA1c ≥ 64 mmol/mol Duration ≥ 10 years With nephropathy SBP, mm Hg, mean (SD) Extreme BP* TC:HDL, mean (SD) Extreme cholesterol**
Total
Māori
Pacific
Indian
Chinese/other Asian
European
12,739 62 (8.9) 6833 (54) 4361 (34) 1446 (11) 9986 (78) 5073 (40) 1670 (13) 3943 (31) 3759 (30) 990 (26) 850 (23) 338 (9) 129 (15.9) 407 (3) 3.9 (1.20) 149 (1)
2669 60 (9.3) 1721 (64) 1110 (42) 602 (23) 2227 (84) 1276 (48) 504 (19) 1309 (49) 1169 (44) 359 (31) 270 (23) 162 (14) 129 (17.1) 138 (5) 4.0 (1.26) 46 (2)
1436 58 (10.0) 986 (69) 631 (44) 181 (13) 1169 (81) 608 (42) 256 (18) 678 (47) 682 (48) 229 (34) 149 (22) 93 (14) 128 (16.8) 68 (5) 3.9 (1.17) 11 (1)
302 60 (9.4) 186 (62) 112 (37) 16 (5) 265 (88) 179 (59) 41 (14) 117 (39) 169 (56) 57 (34) 52 (31) 16 (10) 130 (18.0) 92 (3) 3.8 (1.14) 3 (1)
474 63 (8.8) 249 (52) 185 (39) 28 (6) 365 (77) 154 (33) 48 (10) 74 (16) 136 (29) 19 (14) 20 (15) 2 (2) 125 (15.9) 5 (1) 3.7 (1.04) 2 (0.4)
7858 64 (8.1) 3691 (47) 2323 (30) 619 (8) 5960 (76) 2856 (36) 821 (10) 1765 (23) 1603 (20) 326 (20) 359 (22) 65 (4) 130 (15.2) 187 (2) 3.9 (1.20) 87 (1)
Values are n (%) unless otherwise stated. BP = blood pressure, SBP = systolic BP, TC:HDL = ratio of total cholesterol to high density lipoprotein, TIA = transient ischaemic attack. *BP consistently ≥ 170/100; **TC ≥ 8 mmol/L or TC:HDL ≥ 8.
K.K. Poppe et al. / International Journal of Cardiology 254 (2018) 119–124
20
Men
20
Women
121
Maori
15
European Indian
35-44
10 0
0
5
10
Asian
5
Atrial fibrillation, %
15
Pacific
45-54
55-64
65-74
35-44
45-54
Age, years
55-64
65-74
Age, years
Fig. 1. Proportion of people in the PREDICT cohort with atrial fibrillation, by sex, ethnicity and age group.
respectively (all p-values b0.001). Diabetes was less well controlled (HbA1c ≥64 mmol/mol among 31–34% of Māori, Pacific and Indian vs 14–20% of Europeans and Chinese/Other Asians, all p-values b 0.001) and renal complications of diabetes more prevalent (nephropathy in 10–14% vs 2–4%, all p-values b0.003). These differences were present in all age groups (Supplementary Table 2). 3.2. Thromboembolic risk High thromboembolic risk (CHA2DS2-VASc score ≥ 2) was seen in 77% of the cohort. A higher proportion of Māori, Pacific, and Indian people were at high risk compared with Chinese/Other Asian and
European patients, consistent with the higher burden of comorbidities seen in these groups (Table 2, Supplementary Fig. 1). This difference was demonstrated at all ages, although Chinese/Other Asian had a higher proportion of each age group at high risk than Europeans. Anticoagulation therapy had been dispensed to less than half of those with AF, with only 49% of those at high thromboembolic risk and 19% of the 2997 patients who were at low risk (CHA2DS2-VASc b2) dispensed warfarin or dabigatran. Ethnic differences in those treated with anticoagulation therapy were also seen, with 34% of Indian people and 50% of Māori treated (Table 2). 3508 (70%) of the high stroke risk patients not treated with anticoagulation were prescribed aspirin.
Table 2 Vascular risk and management in patients with atrial fibrillation.
n Thromboembolic risk At high risk Women at high risk At high risk by age group 35–44 years 45–54 years 55–64 years 65–74 years Anticoagulation In those at high risk Cardiovascular risk At high risk Women at high risk At high risk by age group 35–44 years 45–54 years 55–64 years 65–74 years Blood pressure lowering In those at high risk Lipid lowering In those at high risk Aspirin In those at high risk
Total
Māori
Pacific
Indian
Chinese/other Asian
European
12,739
2669
1436
302
474
7858
9742 (77) 4010 (41)
2279 (85) 1050 (46)
1203 (84) 599 (50)
264 (87) 105 (40)
357 (75) 162 (45)
5639 (72) 2094 (37)
240 (52) 1100 (52) 2811 (66) 5591 (95) 5350 (42) 4766 (49)
99 (58) 429 (72) 822 (86) 929 (98) 1325 (50) 1225 (54)
92 (64) 282 (77) 386 (81) 443 (98) 673 (47) 620 (52)
8 (44) 52 (84) 89 (84) 115 (99) 104 (34) 98 (37)
9 (45) 27 (43) 105 (63) 216 (96) 180 (38) 159 (45)
32 (30) 310 (30) 1409 (55) 3888 (93) 3068 (39) 2664 (47)
6838 (54) 2113 (31)
1807 (68) 691 (38)
854 (60) 337 (40)
208 (69) 66 (32)
206 (44) 69 (34)
3763 (48) 950 (25)
152 (33) 788 (37) 2108 (50) 3790 (64) 10,252 (81) 6117 (90) 6922 (54) 4945 (72) 6033 (47) 3933 (58)
74 (43) 307 (52) 676 (71) 750 (79) 2295 (86) 1644 (91) 1563 (59) 1240 (69) 1294 (49) 1006 (56)
47 (33) 162 (44) 297 (63) 348 (77) 1199 (84) 771 (90) 821 (57) 599 (70) 673 (47) 466 (55)
2 (11) 34 (55) 76 (72) 96 (83) 274 (91) 201 (97) 229 (76) 182 (88) 184 (61) 145 (70)
6 (30) 15 (24) 58 (35) 127 (56) 381 (80) 187 (91) 264 (56) 161 (78) 220 (46) 123 (60)
23 (21) 270 (26) 1001 (39) 2469 (59) 6103 (78) 3314 (88) 4045 (52) 2763 (73) 3662 (47) 2193 (58)
Values are n (%) unless otherwise specified.
122
K.K. Poppe et al. / International Journal of Cardiology 254 (2018) 119–124 Table 3 Event rates among patients with AF and at least one year of follow-up (n = 12,326).
Stroke rates within 1 year Overall event rate, % By risk groupa, % 0.8 1.8 2.7 ≥3.2 MACE rates within 5 years Overall event rate, % By risk group, % b5 5–10 10–15 ≥15 Known N20% risk
Whole group n = 12,326
Māori n = 2573
Pacific n = 1384
Indian n = 291
Chinese/other Asian n = 458
European n = 7620
1.2
1.9
1.4
1.0
2.0
0.9
0.2 0.3 0.8 1.7
1.8 0.8 1.3 2.3
0 0 0 2.1
0 0 0 1.5
0 1.3 2.5 2.2
0 0.2 0.7 1.4
25
38
36
24
12
19
8 14 20 26 36
0 20 31 35 48
0 27 26 34 45
0 10 5 17 33
8 11 3 22 16
8 10 15 20 30
a Stroke risk groups are represented by the mean risk where 0.8% risk correlates to a CHADS-VASc score of 0, 1.8% risk to a score of 1, 2.7% risk to a score of 2, and ≥3.2 to a score of ≥3 [8].
Stroke occurred in 709 of 12,326 (6%) patients with AF who either experienced an event or had at least one year of follow-up within 1 to 4122 days of baseline risk assessment. For the entire cohort, 150 (1.2%) had a stroke in the first year of follow-up, with Māori and Chinese/Other Asian having the highest proportion of stroke events (1.9% and 2.0% respectively; Table 3). Results differed slightly for the sub-group of patients who were not receiving anticoagulant therapy (n = 7160): 63 (0.9%) had a stroke however the highest rates occurred among Māori and Pacific (1.4% and 1.6% respectively; Supplementary Table 3a). The CHA2DS2-VASc score overestimated stroke risk in all ethnic groups, with the exception of Māori with a CHA2DS2-VASc score of 0 (approximately 0.8% risk), where the actual event rate was 1.8% (Table 3). The degree of overestimation was least for Chinese/ Other Asian patients.
3.3. Cardiovascular risk Estimated 5-year CVD risk was 15% or more, indicating high CVD risk, in 54% of the cohort (Table 2). As seen for thromboembolic risk, and consistent with their higher burden of comorbidities, a higher proportion of Māori, Pacific, and Indian patients were at high risk compared with Chinese/Other Asian and European patients (Table 2). This difference was demonstrated at all ages, particularly among Māori and Indian patients aged ≥45 years. Over half of the patients with AF were dispensed at least one BP lowering or lipid lowering medication, and 47% received aspirin. Of those at high CVD risk, 90% received BP lowering and 72% received lipid lowering medication (Table 2); 4561 (67%) received both lipidand BP-lowering however 451 (7%) received neither. The highest rates of medication use, including for those at high CVD risk, occurred among Indian patients. A major adverse cardiovascular event (MACE) occurred in 3575 (29%) patients, within 0–4167 days of their baseline risk assessment. The 5-year rate of a MACE was 25%, with the highest rates occurring among Māori and Pacific (38 and 36% respectively) (Table 3). In contrast to stroke risk, 5-year CVD risk was underestimated: 125 (8%) of patients at low risk (b5%) experienced a MACE within 5 years, as did 672 (16%) of patients at intermediate risk (5–15%). Patients at high risk, whether estimated by the CVD risk equation or categorised as high risk due to existing CVD or other high risk conditions, had MACE rates of 26% and 36% respectively. Risk was similarly underestimated among patients who did not receive BP- and lipidlowering medication, particularly among Māori and Pacific patients (Supplementary Table 3b).
4. Discussion This study has shown that the burden of AF in a primary care cohort differs not only by age, but by age within ethnic group. Contrary to the racial paradox that has been observed in some non-European populations [17], Māori, the indigenous population of NZ, and Pacific peoples, had both a higher burden of risk factors for AF and a higher burden of AF than Europeans/Caucasians at all ages. In contrast, Indians had a higher burden of risk factors and a lower burden of AF than Europeans. Our cohort had a generally higher prevalence of risk factors than was described in a study of AF in NZ practices not included in our cohort [22]; however that work used history as recorded in the primary care record only (which are not connected to hospitalisation records) and had significantly lower representation of non-European ethnic groups. For all patients with AF, there are clear disparities between estimated vascular risk and medication use. Over three quarters of the patients in this study were estimated to be at high risk of stroke within one year, and half were at high risk of a CVD event in the next five years. The majority of these high risk patients received a BP lowering medication, three quarters received a statin but only half of those at estimated high stroke risk received anticoagulation therapy (only one third of Indian patients at high risk). The associations between AF and vascular disease, including but not limited to stroke, reinforce the importance of aggressive management of common risk factors [4]. These findings show that there is a clear opportunity to improve evidence-based practice. Current risk assessment tools were inaccurate in this population: three quarters were predicted to be at high risk of stroke in the next year yet only 1.2% experienced a stroke; conversely, a MACE occurred in 34% of those estimated at high 5-year CVD risk but also in 16% at intermediate (5–15%) risk and 8% at low (b5%) CVD risk. Similar risk factors are important in predicting each outcome and are included in each risk score, yet one score overestimates risk (of stroke) and the other underestimates risk (of MACE). In a recent meta-analysis of over 580,000 subjects with AF, Odutayo and colleagues [4] noted that evidencebased strategies to reduce the burden of events in people with AF should include discussion of predicted CVD risk. We agree, however our data demonstrate that the current risk assessment tools need to be optimised for patients with AF in order to more accurately identify those at higher risk. Thromboembolic risk for people with AF is typically assessed using the CHA2DS2-VASc score [20]. While this scoring system is used in many countries, it is important to note that it is based on data from European populations. Yet as we have shown, age, risk, and risk profiles
K.K. Poppe et al. / International Journal of Cardiology 254 (2018) 119–124
differ for different ethnic groups. Compared with Caucasians, AfricanAmericans are less likely to develop AF [23], yet once they have it, have a higher risk of stroke [24] and adding African-American ethnicity to the CHA2DS2-VASc score more accurately predicted their higher risk in a multi-ethnic cohort of people aged N 65 [24]. Risk scores can be modified in a number of ways to suit a population that is different from that used to derive it. Recalibration applies a global factor to the score to increase or decrease results in everyone, and adding new risk factors, such as ethnicity, improves discrimination of the group affected by that new variable. Although useful, these are crude ways to improve risk prediction at the level of the individual. Among Europeans, the proportion with AF tends to increase above the age of 65 years; among Asians, an annual stroke risk of approximately 1.8% for younger AF patients led to the development of a modified CHA2DS2-VASc score that assigned 1 point for age between 50 and 74 years rather than 65–74 years [25]. Older people are more likely to have had a longer exposure to comorbidities than younger people, thus the risk associated with each comorbidity is greater in a score developed for people presenting with AF later in life. The high stroke risk seen in our non-European patients aged b 65, where no points are assigned for age in the standard CHA2DS2-VASc score, reflect the high burden of comorbidities in this group. A high comorbidity burden may indeed increase risk of an adverse event, but the overestimation of 1-year stroke risk seen in our cohort suggests that the amount of risk assigned to each comorbidity by the CHA2DS2-VASc score was greater than needed. If the score was developed in a cohort that spanned the age range of adults with AF, the differing contributions of comorbidities by age may be better represented. Almost half of the cohort with AF had existing vascular disease, placing them at high risk of a future event. For those without CVD, the NZ adjusted Framingham score was used to estimate the 5-year risk of a CVD event; however this score does not include an adjustment for AF. Examining the rates of MACE by ethnicity showed that CVD risk among Indian people with AF was only marginally underestimated. The NZ adjustments to the Framingham score add 5% to the absolute risk for people of Indian descent in an effort to capture the known higher CVD risk among this group [13]. This 5% adjustment is also made for Māori and Pacific peoples, yet CVD risk was still underestimated for these patients in our cohort. The burden of comorbidities was similarly high across all three ethnic groups but a key difference is that a smaller proportion of Indian people had AF than Māori or Pacific (or Europeans) at each age group. As age is a strong driver of CVD risk, the relatively older age of Indian patients with AF in this cohort is accommodated by the current risk score. New risk equations for the primary prevention of CVD in NZ are being developed that include AF as a predictor, thus CVD risk assessment may improve in this patient group. Health policy and practice needs to shift away from its focus on silos of disease-specific risk and instead develop a more holistic, individualised approach to risk assessment and management that is relevant to the life-course of risk and disease. AF occurs for individuals in the setting of a number of CV risk factors and associated clinical manifestations of CV disease. Thus a process of systematic detection [14] and appropriate management in the context of overall CVD risk must be considered for people with AF. After analysing outcomes on almost 10,000,000 people, 6% of whom had AF, Odutayo and colleagues recommended that “the burden of non-stroke events in adults with AF would benefit from a focus on primary prevention and management of CV risk factors” [4]. Clinical practice guidelines need to reflect this life-course approach, and be augmented by continuous refinement of risk assessment in parallel to practice change.
123
eligible population to have had a CVD risk assessment in the past 5 years [26]. In addition, software integrated with the electronic health record has enabled the collection of relevant data on a large and representative sample of adults in contact with the health system, which can be accurately linked to national databases of dispensing and outcome using a unique patient identification number. As a result, these data represent routine clinical practice, include large numbers of non-European patients, and have complete data on many factors, including outcomes. It remains, though, that the cohort represents people who are eligible for a CVD risk assessment (based on age criteria) and in whom AF has been diagnosed. Although representative of the ethnic mix undergoing CVD risk assessment, Indian and Chinese/Other Asian groups are represented by a relatively small number of patients in this cohort. These groups also had a low event rate, limiting the robustness of the comparisons between these and the other ethnic groups. All outcomes are defined from national administrative databases of mortality and public hospital admissions, so if a person experienced an event when out of the country, or if it resulted in admission only to a private hospital, it will not be recorded. Fewer than 10% of all women were defined as being at low thromboembolic risk. The 2016 ESC guidelines for the management of AF effectively remove the additional risk ascribed to women by the CHA2DS2-VASc score by defining high risk among women as a score of 3 or more (and among men as a score of 2 or more) [1]. The current study defined high stroke risk as a score of 2 or more, regardless of sex, as that was the recommendation during the period of data collection [8,10]. 5. Conclusions These data, from a contemporary multi-ethnic population, demonstrate substantial variability by ethnic group in the age at which AF occurs and in the associated risk factors for vascular events. Overestimation of stroke risk and underestimation of CVD risk, contributed to by the high burden of comorbidities at younger ages in many ethnic groups, signals a need to improve the accuracy of risk prediction in this patient group. Overall clinical event rates remain high, and while the majority of patients at high estimated vascular risk received BPlowering medication, use of anticoagulant and lipid-lowering medications for these patients was suboptimal. Systematic approaches to improving cardiovascular risk assessment and management for patients with AF in the community are urgently required. Author contributions KP and RD conceived and designed this work. SW, RJ, MH and JH were involved in the data collection process; KP designed and performed all analyses. KP and RD interpreted the results and drafted the manuscript. All authors critically revised the manuscript. All gave final approval and agree to be accountable for all aspects of work ensuring integrity and accuracy. Funding The PREDICT CVD study has been supported by funding from the Health Research Council of New Zealand via two project grants (HRC 03/183 and 08/121) and under a programme grant (HRC VIEW 11/800 programme). No funders had a role in the design of the study; the collection, analysis, or interpretation of the data; or the decision to approve publication of the finished manuscript.
4.1. Strengths and limitations Disclosures Over the last 5–10 years, NZ has developed a systematic approach to CVD risk assessment in primary care. This has been supported by a Ministry of Health-led national health target requiring 90% of the
Outside this study, SW has received a Fellowship from the Stevenson Foundation and a research grant from Roche Diagnostics Ltd. All other
124
K.K. Poppe et al. / International Journal of Cardiology 254 (2018) 119–124
authors declare no conflicts of interest. All authors have approved the final article. Appendix A. Cardiovascular risk assessment in New Zealand National clinical guidelines advise the following adjustments are made to the 1991 Framingham 5-year CVD risk estimate: A single 5% increase in absolute risk is added for one of: family history of premature CHD or ischaemic stroke; Māori, Pacific or Indo-Asian ethnicity; or diabetes with: microalbuminuria, duration N10 years, or HbA1c consistently N 64 mmol/mol. CVD risk is considered to be ≥ 15% if total or total: HDL cholesterol are ≥ 8 or if BP is consistently ≥ 170/100 mm Hg. The risk score is not applied to patients with: prior CVD, familial hypercholesterolaemia, defective ApoB or combined dyslipidaemia, or diabetes with overt nephropathy or severe renal impairment. These patients are considered as having a CVD-equivalent risk (i.e. at least N20% risk). The PREDICT cohort represents people who have undergone routine CVD risk assessment in accordance with national guidelines. As people of Māori, Pacific or Indian descent are known to be at high CVD risk (at a population level), guidelines advise that risk should start to be assessed in people identifying with these ethnicities 10 years earlier than for other ethnic groups, i.e. Maori, Pacific, Indian men from 35 years old; Maori, Pacific, Indian women from 45 years old; European, Asian men from 45 years old; European, Asian women from 55 years old. As a result, the PREDICT cohort over-represents high-risk ethnic groups compared with their distribution in the general population. Representation is also influenced by the geographic distribution of where people live, as seen in the relatively high proportions of Pacific, Indian and Asian people in PREDICT. The majority of people in these groups live in the upper half of the North Island of New Zealand, where PREDICT software is used. Aged 35–74 years 2013 NZ populationa PREDICT cohort PREDICT as % of NZ n (%) of total n (%) of total population All Māori Pacific Indian Asian European/Caucasian a
2,090,970 235,860 (11%) 99,645 (5%) 66,152 (3%) 143,303 (7%) 1,546,010 (74%)
446,961 59,185 (13%) 56,147 (13%) 37,170 (8%) 42,934 (10%) 251,525 (56%)
21 25 56 56 30 16
NZ Ministry of Health population using a single prioritised ethnicity per person.
Appendix B. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.ijcard.2017.11.045. References [1] P. Kirchhof, S. Benussi, D. Kotecha, et al., 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS, Eur. Heart J. 37 (2016) 2893–2962.
[2] J. Heeringa, D.A.M. van der Kuip, A. Hofman, et al., Prevalence, incidence and lifetime risk of atrial fibrillation: the Rotterdam study, Eur Heart J. 27 (2006) 949–953. [3] S. Chugh, J. Blackshear, W. Shen, S. Hammill, B. Gersh, Epidemiology and natural history of atrial fibrillation: clinical implications, J. Am. Coll. Cardiol. 37 (2001) 371–378. [4] A.A. Odutayo, C.X. Wong, A.J. Hsiao, S. Hopewell, D.G. Altman, C.A. Emdin, Atrial fibrillation and risks of cardiovascular disease, renal disease, and death: systematic review and meta-analysis, Br. Med. J. 354 (2016) i4482. [5] M. Proietti, C. Laroche, G. Opolski, A.P. Maggioni, G. Boriani, G.Y.H. Lip, ‘Real-world’ atrial fibrillation management in Europe: observations from the 2-year follow-up of the EURObservational Research Programme-Atrial Fibrillation General Registry Pilot Phase, Europace 19 (on behalf of the AF Gen Pilot Investigators) (2017) 722–733. [6] J. Olesen, G. Gislason, C. Torp-Pedersen, G.Y.H. Lip, Atrial fibrillation and vascular disease - a bad combination, Clin. Cardiol. 35 (2012) 15–20. [7] C. Ruff, D. Bhatt, P. Steg, et al., Long-term cardiovascular outcomes in patients with atrial fibrillation and atherothrombosis in the REACH Registry, Int. J. Cardiol. 170 (2014) 413–418. [8] A. Camm, G. Lip, R. De Caterina, et al., 2012 focused update on the ESC guidelines for the management of atrial fibrillation: an update of the 2010 ESC guidelines for the management of atrial fibrillation, Eur. Heart J. 33 (2012) 2719–2747. [9] C.T. January, L.S. Wann, J.S. Alpert, et al., 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation, J. Am. Coll. Cardiol. 64 (2014) e1–76. [10] Best Practice New Zealand, An update on antithrombotic medicines. What does primary care need to know? Best Practice J 67 (2015) 12–24. [11] M. Piepoli, A. Hoes, S. Agewall, et al., 2016 European Guidelines on cardiovascular disease prevention in clinical practice, Eur. Heart J. 37 (2016) 2315–2381. [12] D. Goff, D. Lloyd-Jones, G. Bennett, et al., 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines, Circulation 63 (2014) 2935–2959. [13] New Zealand Guidelines Group, New Zealand Primary Care Handbook 2012, 2012. [14] B. Freedman, J. Camm, H. Calkins, et al., Screening for atrial fibrillation. A report of the AF-SCREEN International Collaboration, Circulation 135 (2017) 1851–1867. [15] C.X. Wong, A.G. Brooks, Y.-H. Cheng, et al., Atrial fibrillation in indigenous and nonindigenous Australians: a cross-sectional study, BMJ Open 4 (2014) e006242. [16] R. Teh, A. Martin, N. Kerse, L. Dyall, R. Doughty, The burden of atrial fibrillation in octogenarians, Heart, Lung and Circulation. 22 (2013) 580–581. [17] M. Amponsah, E. Benjamin, J. Magnani, Atrial fibrillation and race - a contemporary review, Curr Cardiovasc Risk Rep. 7 (2013). [18] J.M. Katzenellenbogen, T.H.K. Teng, D. Lopez, et al., Initial hospitalisation for atrial fibrillation in aboriginal and non-aboriginal populations in Western Australia, Heart 101 (2015) 712–719. [19] S. Wells, T. Riddell, A. Kerr, et al., Cohort profile: the PREDICT cardiovascular disease cohort in New Zealand primary care (PREDICT CVD-19), Int. J. Epidemiol. 46 (1) (2017) 22-22j. [20] G. Lip, R. Nieuwlaat, R. Pisters, D. Lane, J. Crijns, Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk fact based approach. The Euro Heart Survey on Atrial Fibrillation, Chest 137 (2010) 263–272. [21] K. Anderson, P. Odell, P. Wilson, W. Kannel, Cardiovascular disease risk profiles, Am. Heart J. 121 (1991) 293–298. [22] A. Tomlin, H. Lloyd, M. Tilyard, Atrial fibrillation in New Zealand primary care: prevalence, risk factors for stroke and the management of thromboembolic risk, Eur. J. Prev. Cardiol. 24 (2016) 311–319. [23] A. Alonso, N. Roetker, E.Z. Soliman, L. Chen, P. Greenland, S. Heckbert, Prediction of atrial fibrillation in a racially diverse cohort: the Multi-Ethnic Study of Atherosclerosis (MESA), J. Am. Heart Assoc. 5 (2016) e003077. [24] R. Kabra, S. Girotra, M. Vaughan Sarrazin, Refining stroke prediction in atrial fibrillation patients by addition of African-American ethnicity to the CHA2DS2-VASc score, J. Am. Coll. Cardiol. 68 (2016) 461–470. [25] T.-F. Chao, G.Y.H. Lip, C.-J. Liu, et al., Validation of a modified CHA2DS2-VASc score for stroke risk stratification in Asian patients with atrial fibrillation, Stroke 47 (2016) 2462–2469. [26] Ministry of Health, Targeting Diabetes and Cardiovascular Disease, 2011.