- Email: [email protected]
Trends in incidence and prevalence of hospitalization for atrial fibrillation and associated mortality in Western Australia, 1995–2010
International Journal of Cardiology 208 (2016) 19–25
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Trends in incidence and prevalence of hospitalization for atrial fibrillation and associated mortality in Western Australia, 1995–2010☆ Tom Briffa a,1, Joseph Hung b,⁎,1, Matthew Knuiman a, Brendan McQuillan b, Derek P. Chew c, John Eikelboom d, Graeme J. Hankey a, Tiew-Hwa K. Teng e, Lee Nedkoff a, Rukshen Weerasooriya f, Andrew Liu g, Paul Stobie h a
School of Population Health, The University of Western Australia, Crawley, Western Australia, Australia School of Medicine and Pharmacology, Sir Charles Gairdner Hospital Unit, The University of Western Australia, Crawley, Western Australia, Australia School of Medicine, Flinders University of South Australia, Adelaide, South Australia, Australia d Department of Medicine, McMaster University, Hamilton, Canada e Western Australian Centre for Rural Health, The University of Western Australia, Crawley, Western Australia, Australia f Department of Cardiology, Royal Perth Hospital, Perth, Western Australia, Australia g Department of Cardiology, Fiona Stanley Hospital, Murdoch, Western Australia, Australia h Department of Cardiovascular Medicine, Sir Charles Gairdner Hospital, Nedlands, Western Australia, Australia b c
a r t i c l e
i n f o
Article history: Received 5 June 2015 Received in revised form 18 December 2015 Accepted 15 January 2016 Available online 28 January 2016 Keywords: Atrial fibrillation Hospitalization Incidence Prevalence Mortality
a b s t r a c t Objective: Hospitalization for atrial fibrillation (AF) is a large and growing public health problem. We examined current trends in the incidence, prevalence, and associated mortality of first-ever hospitalization for AF. Methods: Linked hospital admission data were used to identify all Western Australia residents aged 35–84 years with prevalent AF and incident (first-ever) hospitalization for AF as a principal or secondary diagnosis during 1995–2010. Results: There were 57,552 incident hospitalizations, mean age 69.8 years, with 41.4% women. Over the calendar periods, age- and sex-standardized incidence of hospitalization for AF as any diagnosis declined annually by 1.1% (95% CI; 0.93, 1.29), while incident AF as a principal diagnosis increased annually by 1.2% (95% CI; 0.84, 1.50). Incident AF hospitalization was higher among men than women, and 15-fold higher in the 75–84 compared with 35–64 year age group. The age- and sex-standardized prevalence of AF increased annually by 2.0% (95% CI; 1.88, 2.03) over the same period. Comorbidity trends were mixed with diabetes and valvular heart disease increasing, and hypertension, coronary artery disease, heart failure, cerebrovascular disease, and chronic kidney disease decreasing. The 1-year all-cause mortality after incident AF hospitalization declined from 17.6% to 14.6% (trend P b 0.001), with an adjusted hazard ratio of 0.86 (95% CI; 0.81, 0.91). Conclusion: This contemporary study shows that incident AF hospitalization is not increasing except for AF as a principal diagnosis, while population prevalence of hospitalized AF has risen substantially. The high 1-year mortality following incident AF hospitalization has improved only modestly over the recent period. © 2016 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Atrial fibrillation (AF) is the most common chronic arrhythmia and is associated with substantial morbidity and mortality [1–3]. The increasing prevalence of AF has been attributed to a combination of ☆ Grant support: This work was supported by a project grant (#1020373) from the National Health and Medical Research Council. The funding body had no direct role in the study design, the collection, analysis, or interpretation of data, and the writing of the report. ⁎ Corresponding author at: School of Medicine & Pharmacology (SCGH Unit) M503, Harry Perkins Institute of Medical Research, QQ Block, QEII Medical Centre, Verdun Street, Nedlands, Western Australia 6009, Australia. E-mail address: [email protected] (J. Hung). 1 Denotes equal first co-authors.
http://dx.doi.org/10.1016/j.ijcard.2016.01.196 0167-5273/© 2016 Elsevier Ireland Ltd. All rights reserved.
population aging, changing pattern of risk factors, and improved survival from contributory cardiovascular and non-cardiovascular diseases [1–3]. Accurate estimates of trends in the incidence of AF, disease burden in the population and its effect on mortality are necessary to effectively manage this significant and growing public health problem. Notably hospitalizations account for most of the costs (over 50%) associated with AF [4–6]. Studies conducted from the 1980s to 1990s from North America and Western Europe have reported substantial increases in the prevalence and incidence of hospitalizations for AF either as a principal diagnosis or any diagnosis [7–12]. However, a recent retrospective cohort study of Medicare beneficiaries in North America, aged 65 years and older, reported that the incidence of AF has in fact remained steady between 1993 and 2007 [13]. Hence, it is unclear if the observed rise in AF
20
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
Table 1 Characteristics of patients hospitalized with incident atrial fibrillation as any discharge diagnosis stratified by calendar period from 1995 to 2010 in Western Australia. Characteristic, n (%) unless specified
Overall n = 57,552
Age, years, mean ± SD Age groups 35–64 years 65–74 75–84 Women Principal discharge diagnosis Emergency admission Comorbidities Coronary heart disease Cerebrovascular disease Peripheral arterial disease Heart failure Valvular heart disease Hypertension Diabetes Chronic kidney disease Cancer Chronic obstructive pulmonary disease CHA2DS2-VASc score, mean ± SD CHA2DS2-VASc score categories 0 points 1 2 or more
1995 to 1998 n = 13,205
1999 to 2002 n = 13,502
2003 to 2006 n = 14,352
2007 to 2010 n = 16,493
69.8 (10.8)
70.3 (10.3)
69.9 (10.7)
69.9 (10.8)
69.3 (11.1)
b.0001
15,823 (27.5) 18,078 (31.4) 23,651 (41.1) 23,808 (41.4) 17,660 (30.7) 37,258 (64.7)
3232 (24.5) 4552 (34.5) 5421 (41.1) 5605 (42.4) 3375 (25.6) 8213 (62.2)
3577 (26.5) 4366 (32.3) 5559 (41.2) 5547 (41.1) 4206 (31.2) 8755 (64.8)
3980 (27.7) 4308 (30.0) 6064 (42.3) 5959 (41.5) 4449 (31.0) 9462 (65.9)
5034 (30.5) 4852 (29.4) 6607 (40.1) 6697 (40.6) 5630 (34.1) 10,828 (65.6)
b.0001 0.012 b.0001 b.0001
22,143 (38.5) 7629 (13.3) 6712 (11.7) 14,560 (25.3) 12,031 (20.9) 28,739 (49.9) 11,195 (19.4) 5929 (10.3) 18,767 (32.6) 10,574 (18.4) 3.1 (1.9)
5722 (43.3) 2045 (15.5) 1789 (13.6) 4046 (30.6) 1819 (13.8) 6749 (51.1) 2139 (16.2) 1693 (12.8) 3411 (25.8) 3048 (23.1) 3.2 (1.9)
5416 (40.1) 1913(14.2) 1643 (12.2) 3592 (26.6) 2981 (22.1) 6695 (49.6) 2601 (19.3) 1336 (9.9) 4131 (30.6) 2759 (20.4) 3.1 (1.9)
5485 (38.2) 1839 (12.8) 1630 (11.4) 3521 (24.5) 3465 (24.1) 7219 (50.3) 2983 (20.8) 1378 (9.6) 4927 (34.3) 2433 (17.0) 3.1 (1.9)
5520 (33.5) 1832 (11.1) 1650 (10.0) 3401 (20.6) 3766 (22.8) 8076 (49.0) 3472 (21.1) 1522 (9.2) 6298 (38.2) 2334 (14.2) 2. 9 (1.9)
b.0001 b.0001 b.0001 b.0001 b.0001 0.002 b.0001 b.0001 b.0001 b.0001 b.0001
7055 (12.3) 5315 (9.2) 45,182 (78.5)
1236 (9.4) 1102 (8.3) 10,867 (82.3)
1522 (11.3) 1276 (9.4) 10,704 (79.3)
1775 (12.4) 1336 (9.3) 11,241 (78.3)
2522 (15.3) 1601 (9.7) 12,370 (75.0)
b .0001
prevalence is due to an increase in incident (de novo) AF, increasing survival after new AF diagnosis or can be largely attributed to population aging. Furthermore, there is significant uncertainty regarding projected increases in AF incidence and prevalence over the ensuing decades which undermines proper healthcare planning [14]. We therefore examined trends in the incidence of first-ever hospitalization for AF as a discharge diagnosis, and population prevalence of people with prior AF admission, among residents of Western Australia (WA), aged 35 to 84 years, from 1995 through 2010. We also assessed trends in antecedent risk factors, comorbid diseases, and all-cause mortality among incident AF cases.
Trend P value
2. Methods 2.1. Data sources Hospital admissions and mortality data were obtained from 1 January 1980 to 31 December 2012 from the WA Data Linkage System for residents in Western Australia [15]. An individual's administrative records are linked with a N 99% accuracy using probabilistic matching on key personal identifiers [15]. All records include dates of care, principal diagnosis, and up to 20 secondary discharge diagnosis codes from the International Classification of Diseases, versions nine (ICD-9), nine-
Table 2 Incidence of hospitalization for atrial fibrillation as a discharge diagnosis stratified by calendar periods from 1995 to 2010 in Western Australia. No. of persons with atrial fibrillation as any discharge diagnosis (Incidence rate per 1000 person-years in each perioda)
Mean annual % change (95% confidence interval)λ
P value
Characteristic, n (rate/1000) 1995 to 1998
1999 to 2002
2003 to 2006
2007 to 2010
Overall Gender Men Women Age-specific 35–64 years 65–74 75–84
n = 13,205 (4.39)
n = 13,502 (3.99)
n = 14,352 (3.73)
n = 16,493 (3.83)
−1.1 (−0.93, −1.29)
b.0001
7604 (5.56) 5606 (3.30)
7955 (5.10) 5547 (2.97)
8393 (4.66) 5959 (2.86)
9797 (4.79) 6696 (2.92)
−1.21 (−0.98, −1.44) −0.87 (−0. 60, −1.15)
b.0001 b.0001
3235 (1.46) 4554 (10.88) 5421 (26.39)
3577 (1.38) 4366 (9.85) 5559 (23.52 )
3980 (1.33) 4308 (8.95) 6064 (22.07)
5033 (1.46) 0.19 (−0.15, 0.54) 4851 (8.88) −1.66 (−1.35, −1.97) 6609 (22.08) −1.44 (−1.17, −1.72)
0.27 b.0001 b.0001
No. of persons with atrial fibrillation as a principal discharge diagnosis (Incidence rate per 1000 person-yearsa)
Mean annual % change (95% confidence interval)λ
P value
Characteristic, n (rate/1000)
1995 to 1998
1999 to 2002
2003 to 2006
2007 to 2010
Overall Gender Men Women Age-specific 35–64 years 65–74 75–84
n = 3375 (1.11)
n = 4206 (1.24)
n = 4449 (1.16)
n = 5630 (1.31)
1.17 (0.84, 1.50)
b.0001
1870 (1.31) 1506 (0.92)
2486 (1.55) 1720 (0.94)
2623 (1.43) 1826 (0.90)
3317 (1.59) 2313 (1.02)
1.33 (0.90, 1.57) 1.05 (0.55, 1.55)
b.0001 b.0001
1293 (0.57) 1122 (2.68) 961 (4.47)
1632 (0.62) 1332 (3.02) 1242 (5.16)
1811 (0.60) 1343 (2.79) 1295 (4.68)
2397 (0.69) 1641 (2.99) 1592 (5.30)
1.66 (1.14, 2.18) 0.84 (0.27, 1.43) 1.02 (0.41, 1.64)
b.0001 0.0042 0.0010
a Overall rates are age and sex standardized to the AF-free WA population aged 35 to 84 years in 2010; rates for men and women are age standardized; age-specific rates are sex standardized. λ Poisson models adjusted for 5-year age group and gender.
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
clinical modification (ICD-9-CM), and ten-Australian modification (ICD10-AM). We also identified death dates in the person-linked file containing complete mortality records through to 31 December 2012. 2.2. Incident and prevalent cases of AF We identified cases of atrial fibrillation/atrial flutter based on either a principal or secondary discharge diagnosis code (ICD-9 code 427.3; ICD-10-AM code I48) during the index admission period from 1 January 1995 to 31 December 2010. [16,17] Medical chart review of 184 randomly selected hospital records from 1998, 2003, and 2008 with a principal or secondary discharge diagnosis of AF established a positive predictive value of 84.6% when based on a confirmatory ECG in the medical record of the same admission and 98.9% when based on an ECG or written record of AF in the medical notes or discharge letter. AF cases were considered as incident (first-ever) if there was no hospitalization with AF coded in any discharge diagnosis field dating back a fixed 15-year period from the index date of admission. We also identified a subset of incident AF cases where AF was coded in the principal diagnosis field only. Conversely, an individual was considered to have prevalent AF if it was coded on any hospital discharge record in the 15 years prior to June 30 of a given calendar year and the person was still alive and residing in WA at that time. Mid-year estimates of the WA resident population by gender and 5-year age groups for each of the years, 1995 to 2010, were sourced from the Australian Bureau of Statistics, and represented the denominators for the calculation of prevalence. For incidence calculation in any calendar year, the denominator was the corresponding WA population count after prevalence correction, ie, after subtracting the number of prevalent AF cases. We calculated annual incidence per 1000 person-years and prevalence per 1000 persons as well as the average annual incidence and prevalence for consecutive 4-year calendar periods (1995–1998, 1999–2002, 2003– 2006 and 2007–2010).
21
period changes in categorical variables. Finally, we used Cox proportional hazards models to obtain mortality hazard ratios (HRs) with 95% confidence intervals (95% CIs) comparing the mortality risk in each calendar period versus the first after having satisfied that the assumption of proportional hazards was met. Covariates included were 5-year age group, gender, CHA2DS2-VASc score, ICD-9 period, admission year, and an admission year ∗ ICD-9 period interaction term.
2.3. Comorbidities For individuals with incident AF, we identified comorbidities dating back a fixed 15 years from the index admission or concurrent with the index AF admission. Comorbidities were identified from the relevant ICD codes for coronary artery disease, cerebrovascular disease, peripheral arterial disease, heart failure, valvular heart disease, hypertension, diabetes, chronic kidney disease, all cancers, and chronic obstructive pulmonary disease. Details of the ICD codes have been previously reported. [16,17]. A CHA2DS2-VASc score (congestive heart failure, hypertension, age ≥ 75 years, diabetes mellitus, stroke/transient ischemic attack, vascular disease, age 65–74 years, sex category) was calculated for each person, except that female gender received a score of 1 only if there was an additional risk factor. [18]. 2.4. Statistical analysis We calculated age- and sex-standardized incidence and prevalence of hospitalized AF overall, and by sex and age-specific groups (35–64, 65–74, and 75–84 years), with rates standardized to the age and sex distributions of the AF-free WA population in 2010 for incidence and the WA population in 2010 for prevalence. We estimated the average annual change in incidence rates and prevalence using age- and sex-adjusted Poisson and logistic regression models respectively and to determine the statistical significance of these trends. A calendar year ∗ ICD-9 interaction was tested to see if the trend changed from mid-1999 when the ICD9 coding system was replaced by the ICD10 system. For individuals with incident AF, we summarized demographic and clinical characteristics, comorbidities and discharge diagnosis category using means with standard deviations or frequency distribution with percentages. We used linear models to test for significant temporal trends in continuous variables, and the chi-squared statistic to test for
Fig. 1. Incidence of hospitalization with atrial fibrillation as any discharge diagnosis by calendar year, 1995 to 2010; (A) in the overall population, (B) by sex, and (C) by agespecific groups. The dashed lines in panel A represent 95% confidence limits.
22
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
We used SAS version 9.4 for all analyses (SAS Institute Inc., Cary, NC). Ethics committees of the Western Australian Department of Health, the University of Western Australia, Sir Charles Gairdner Hospital, Royal Perth Hospital, Fremantle Hospital, and Hollywood Private Hospital, Western Australia approved the study. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institution's human research committee.
period (P b 0.001) (Table 2). The increasing incidence was consistent across all subgroups with the 35–64 year olds showing the largest change (+1.7% annually) (Table 2). Modeling with inclusion of the calendar year ∗ ICD-9 interaction term showed that the estimated annual incidence trend was not significantly different in the ICD-9 and ICD-10 periods for AF as a principal or any discharge diagnosis. 3.3. Prevalence
3. Results 3.1. Patient characteristics Table 1 shows the characteristics of the 57,552 first-ever hospitalized AF cases, mean age 69.8 years, with 41.4% women, stratified by consecutive 4-year calendar periods from 1995 to 2010. Patients aged 65 years or older comprised more than two thirds of the incident AF cases. The age and sex distribution changed over the study period with an increasing proportion of cases in the youngest age group (35– 64 years) and a lower proportion of women. The proportion of incident cases with AF as a principal discharge diagnosis increased from 25.6% to 34.1% over the study period. Comorbidity trends were varied with diabetes, valvular heart disease, and cancer increasing, but coronary artery disease, heart failure, cerebrovascular disease, peripheral arterial disease, and chronic obstructive pulmonary disease decreasing. Although hypertension declined it remained highly prevalent (≈50%) over the study period. The mean CHA2DS2-VASc score decreased across calendar periods with the proportion of cases with a score ≥ 2 declining from 82.3% to 75.0% (p b 0.001). Compared to the whole cohort, the subset of people hospitalized with AF as a principal diagnosis (17,660; 30.7%), were younger (age mean 66.1, SD 11.8 years), with a similar proportion of women (41.7%), a lower prevalence of all comorbidities but with similar temporal trends, and a lower mean CHA2DS2-VASc score including proportion of people (62.1%) with a CHA2DS2-VASc score ≥ 2 (Supplemental Table 1).
The age- and sex-standardized prevalence of hospitalized AF cases increased overall from 20.2 to 25.6 per 1000 persons over the calendar periods, representing a 1.96% (95% CI: 1.88 to 2.03) annual increase (p b 0.001) (Table 3, Fig. 2A). The prevalence increased similarly in both genders and all age subgroups (Table 3, Fig. 2B and C). By the last calendar period, the prevalence of AF was still 1.8-fold higher in men than women and was 15-fold higher in the highest versus lowest age group (Table 3). 3.4. Mortality The overall 30-day, 1-year and 3-year all-cause mortality over the study period was 7.1%, 16.6%, and 27.1% respectively, and declined modestly across the calendar periods (all P b 0.001) (Table 4). Comparing the last to first calendar period, the age, sex, and CHA2DS2-VASc riskadjusted HR was 0.91 (95% CI; 0.83, 1.00) for death at 30 days, 0.86 (95% CI; 0.81, 0.91) for death at 1 year and 0.82 (95%CI; 0.78, 0.86) for death at 3 years (Table 4). The overall all-cause mortality in the subset with incident AF as a principal discharge diagnosis was considerably lower than the entire cohort being 0.9%, 4.7%, and 11.1% at 30 day, 1 year, and 3 years respectively. There was also a significant temporal reduction in crude 1-year and 3-year mortality in this subset and their reduced risk-adjusted HRs for death were more apparent and significant in the latter calendar periods than the overall cohort (Table 4). 4. Discussion
3.2. Incidence The age- and sex-standardized incidence of hospitalization for AF as any diagnosis declined overall from 4.39 to 3.83 per 1000 person-years over the calendar periods, representing a 1.1% (95%CI: 0.93, 1.29) annual rate decline (p b 0.001) (Table 2, Fig. 1A). Incidence was consistently higher among men than women and more than 15-fold higher in the 75–84 compared with 34–64 year age group (Fig. 1B and C). The incidence declined significantly in both genders and the 65–74 and 74– 84 year age groups (all P b 0.001), but was stable in the 35–64 year olds (Table 2). By contrast, the age- and sex-standardized incidence of hospitalization for AF as a principal diagnosis increased annually by 1.17% (95% CI: 0.84, 1.50) from 1.11 to 1.31 per 1000 person-years over the study
In this WA population-based study we found that incidence of hospitalization with AF did not increase from 1995 through 2010, except in people with AF admission as a principal diagnosis and particularly younger adults (aged b65 years). However, steadily increasing prevalence of people with hospitalized AF is adding to the overall population burden of this serious arrhythmia. Incident AF hospitalization is still associated with a high short and long-term mortality which has improved only modestly over the recent period. 4.1. Incidence Previous reports have highlighted a growing prevalence of AF although it is unclear if this is largely driven by an aging population or
Table 3 Prevalence of persons with prior hospital diagnosis of atrial fibrillation stratified by calendar periods from 1995 to 2010 in Western Australia. No. of persons (average annual age-standardized prevalence rate per 1000 Persons in each perioda)
Mean annual % change (95% confidence interval)λ
P value
Characteristic, n (rate/1000)
1995 to 1998
1999 to 2002
2003 to 2006
2007 to 2010
Overall Gender Men Women Age groups 35–64 years 65–74 75–84
n = 60,233 (20.2)
n = 79,688 (23.5)
n = 96,081 (24.7)
n = 112,255 (25.6)
1.96 (1.88, 2.03)
b.0001
36,677 (26.3) 23,556 (14.0)
48,591 (30.4) 31,097 (16.7)
59,010 (31.7) 37,071 (17.7)
69,953 (32.9) 42,302 (18.3)
1.76 (1.66, 1.85) 2.04 (1.92, 2.16)
b.0001 b.0001
15,537 (6.9) 19,950 (45.2) 24,746 (106.4)
20,993 (8.0) 25,153 (53.4) 33,542 (125.2)
25,690 (8.5) 29,041 (56.8) 41,350 (131.3)
31,208 (9.0) 34,105 (58.7) 46,942 (135.8)
2.10 (2.24, 1.95) 2.00 (1.87, 2.13) 1.83 (1.72, 1.95)
b.0001 b.0001 b.0001
a Overall prevalence is age and sex standardized to the whole WA population aged 35 to 84 years in 2010; rates for men and women are age standardized; age-specific rates are sex standardized. λ Logistic regression models adjusted for 5-year age group and gender.
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
changing incidence [2,3,7–12]. There also remains a wide range of uncertainty around the magnitude of future trends in incidence of AF [14]. It is known that cardiovascular hospitalizations are frequent following initial AF diagnosis, [19,20] and therefore trends in incident hospitalization with AF may reflect underlying population incidence. Additionally, it is recognized that hospitalizations represent the major component of the healthcare costs related to AF [4–6]. Contrary to earlier studies, we actually found a small downward trend in age and sex-standardized incidence of admissions with AF, falling by nearly 1.1% annually from 1995 through 2010. Our findings are supported by the only other study which has reported on the incidence of AF in this contemporary era although this study was restricted to a representative 5% sample of Medicare beneficiaries in the United States aged 65 years and older [13]. They also found that the incidence rate of any AF diagnosis in both inpatient and outpatient settings was relatively stable from 1993 through to 2007 [13]. A novel finding from our study was that the incidence of admission with AF as a principal diagnosis increased by 1.2% annually over the same period, and incidence increased in both sexes and across all agespecific groups but most in the 35–64 year age group. In a national study, Wong et al. [21] examined all hospitalizations with a principal diagnosis of AF in the National Hospital Morbidity Data set from the Australian Institute of Health and Welfare, and reported that the number of hospitalizations increased by 7.9% annually from 1993 through 2007, of which AF related procedures constituted only a small percentage of total hospitalizations. However, the study could not distinguish between de novo and repeat hospitalizations. Since our local data for incident admissions are likely to be nationally representative, this suggests that repeat hospitalizations are the main driver for the substantial increases in total hospitalization numbers seen across Australia. This is concerning as recurrent admissions, particularly if unplanned, add to the cumulative healthcare costs and resource burden related to AF. Further studies addressing the causes of increasing AF admissions are therefore needed. Miysaka et al. [19] suggested that the marked increase in hospitalization after first AF diagnosis from 1980 to 2000 in Olmsted County was largely driven by a changing practice pattern in AF management, while Wong et al. [22] suggested that the main cause for expanding AF hospitalizations in Australia was increased hospitalizations in older people affected by AF rather than changing procedural trends.
23
suggested that higher obesity rates in urbanized populations may well underlie an increasing incidence of AF seen worldwide [2,16,26]. Coincidentally we observed an increasing prevalence of diabetes which may reflect a rising obesity rate in our population. 4.3. Prevalence We found that the prevalence of hospitalized AF was higher among men than women and increased exponentially with increasing age,
4.2. Effect of comorbidities Our incident AF hospitalized cases had a high prevalence of predisposing risk factors (such as hypertension and diabetes), and antecedent conditions (such as atherosclerotic vascular disease and heart failure). However, with the exception of diabetes and valvular heart disease, the prevalence of these comorbidities declined significantly over the study period which is also reflected in the falling individual CHA2DS2VASc scores. Similar to other studies, [11,13,23] coronary artery disease and heart failure are the two most common comorbidities associated with AF hospitalization but their incidence have declined substantially over the study period in the WA population [24,25]. It is very likely that this has had a significant impact on the incidence of AF hospitalization particularly as a secondary discharge diagnosis. Nevertheless, we observed an upward trend in the incidence and proportion of incident AF admissions as a principal diagnosis, particularly affecting younger adults (age b 65 years). Previous investigators [8] have speculated that the publication of clinical trials demonstrating the efficacy of antithrombotic therapy in the 1990s may have increased physician awareness of AF and resulted in an increase in hospital admission primarily for AF, as well as influencing coding practice especially secondary coding. We are however unable to ascertain if the increase in incident hospitalizations for AF as a principal diagnosis represents a lower admission threshold, an improved diagnosis rate, or a true increase in incidence of AF particularly in younger adults. It has been
Fig. 2. Prevalence of persons with any hospital diagnosis of atrial fibrillation by calendar year, 1995 to 2010; (A) in the overall population, (B) by sex, and (C) by age-specific groups. The dashed lines in panel A represent 95% confidence limits.
24
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
which is concordant with global studies of AF prevalence [2,3]. Given the slightly declining overall incidence rates, our observed ≈ 30% increase in age- and sex-standardized prevalence of people with hospitalized AF is likely to reflect increased long-term survival after incident AF hospitalization. Piccini et al. [13] reported in their Medicare 5% sample a similar growth (N 25%) in prevalent AF cases over a 15-year period when using a stricter definition of AF as the primary diagnosis. By the last calendar period, we estimated that the age-adjusted prevalence of hospitalized AF was around 3.3% in men and 1.8% in women (32.9 and 18.3 per 1000 persons respectively). This is in line with a recent review [3] which estimated a global pooled (age-adjusted) AF prevalence of around 3.3% in men and 2.4% in women, but higher than the prevalence estimate of 1.4% in men and 0.9% in women aged ≥ 35 years reported from a systematic review of populationbased studies in 2010 [2]. We suggest that while hospital administrative data will underestimate real AF prevalence, they are still valuable for monitoring the population burden of AF and its impact on healthcare systems. 4.4. Mortality Earlier community and hospital registry-based studies have established that a new diagnosis of AF is associated with a high short and long-term case-fatality [11,27,28]. In this contemporary study, the overall 1-year and 3-year mortality rate after incident AF hospitalization was 16.6% and 27.1% respectively. Similar to previous studies, [8,23] we found that mortality rates were considerably higher for people admitted with AF as a secondary compared with principal diagnosis, presumably because of older patient age and increased comorbidities such as stroke and myocardial infarction with a high case fatality. Although we established that mortality rates declined, a significant reduction in risk-adjusted mortality hazard ensued mainly during the latter calendar periods. By comparison, the Medicare 5% sample study of an older population (mean age 80 years) found no significant reduction in standardized mortality ratios from 1997 through 2007 [13]. Another contemporary registry study from Sweden from 1987 through 2006 confirmed the poor prognosis associated with AF, but also found that 3-year all-cause mortality after first-ever AF hospitalization had significantly declined throughout the period [23]. While the reasons for the improved survival are unclear, the investigators observed that many of the important comorbidities occurring in AF patients have also shown prognostic improvements during the same period [23]. Similarly
we have reported improved long-term survival of patients with coronary artery disease and heart failure in WA during the same era [29,30]. Oral anticoagulants are the only therapy that have been shown in randomized trials to decrease all-cause mortality in patients with nonvalvular AF [31]. However, data on the prescribing rates of oral anticoagulants for this condition are lacking in Australia although it is likely to reflect patterns seen in other western countries [32,33]. Patients admitted with a principal diagnosis of AF formed an increasing proportion of incident AF cases admitted during the more recent calendar periods and these younger lower risk cases could also contribute to an overall lower mortality in the cohort. The fact that their risk-adjusted hazard for death at 1 and 3 years is significantly reduced suggests that this may have occurred in part due to an improvement in clinical management of AF in this subset. 4.5. Limitations and strengths Our study has several limitations. We are unable to differentiate between subtypes of AF using administrative data. Incidence and prevalence estimates from hospital morbidity datasets will underestimate actual incidence and prevalence in the community, although temporal trends may still be revealing. Nonetheless, hospital data are crucial to assessment of AF burden and costs and for healthcare planning. We excluded people aged 85 years and older because hospital morbidity data are considered less reliable in the very elderly. The diagnosis of AF was not formally validated except in a small randomly selected sample which indicated a high positive predictive value based on medical record review. Coding practices change over time, including a change from ICD-9 to ICD-10 that occurred in mid-1999. However, we found that the estimated annual change in incidence rate was not significantly different in the ICD-9 and ICD-10 periods. There is likely underreporting of several important comorbidities, most notably hypertension but also diabetes and heart failure, and therefore CHA2DS2-VASc scores will also be underestimated. Our local hospital incident and prevalent trends may not be representative of trends in other populations especially from developing countries. We are also unable to ascertain the direct impact on incident hospitalization rates resulting from improved diagnostic methods or changing clinical practice or the direct impact of improved AF management on changing mortality. The major advantages of our study derive from the populationbased design in a state which is nationally representative [34], the high-quality person-based linkage [15], and the large sample size that
Table 4 Estimated all-cause mortality risk after incident hospitalization for atrial fibrillation as a discharge diagnosis stratified by calendar periods from 1995 to 2010 in Western Australia. Total mortality for AF as any diagnosis for each calendar period and hazard ratios compared to first calendar perioda All-cause deaths
1995 to 1998 (N = 13,205)
1999 to 2002 (N = 13,502)
30-Day, n (%) HR (95% CI) 1-Year, n (%) HR (95% CI) 3-Year, n (%) HR (95% CI)
935 (7.1) 1.00 2322 (17.6) 1.00 3879 (29.4) 1.00
1059 (7.8) 1.15 (1.05, 1.26) 2455 (18.2) 1.07 (1.01, 1.13) 3837 (28.4) 1.00 (0.95, 1.04)
2003 to 2006 (N = 14,352) P = 0.002 P = 0.023 P = 0.94
1064 (7.4) 1.08 (0.99, 1.18) 2378 (16.6) 0.96 (0.91, 1.02) 3772 (26.3) 0.92 (0.86, 0.94)
2007 to 2010 (N = 16,493) P = 0.08 P = 0.18 P b .0001
1007 (6.1) 0.91 (0.83, 1.00) 2401 (14.6) 0.86 (0.81, 0.91) 2925 (24.1)λ 0.82 (0.78, 0.86)
P b 0.001 P = 0.049 P b 0.001 P b .0001 P b 0.001 P b .0001
Total mortality for AF as a principal diagnosis for each calendar period and hazard ratios compared to first calendar perioda All-cause deaths
1995 to 1998 (N = 3375)
1999 to 2002 (N = 4206)
30-Day, n (%) HR (95% CI) 1-Year, n (%) HR (95% CI) 3-Year, n (%) HR (95% CI)
36 (1.1) 1.00 197 (5.8) 1.00 438 (13.0) 1.00
42 (1.0) 0.92 (0.59, 1.46) 207 (4.9) 0.83 (0.68, 1.01) 491 (11.7) 0.88 (0.78, 1.00)
2003 to 2006 (N = 4449) P = 0.74 P = 0.063 P = 0.06
45 (1.0) 0.98 (0.63, 1.52) 200 (4.5) 0.76 (0.63, 0.93) 469 (10.5) 0.80 (0.70, 0.92)
2007 to 2010 (N = 5630) P = 0.92 P = 0.008 P = 0.001
42 (0.8) 0.76 (0.48, 1.19) 231 (4.1) 0.72 (0.60, 0.88) 389 (9.6)λ 0.74 (0.65, 0.85)
P = 0.35 P = 0.23 P = 0.002 P = 0.001 P b .0001 P b .0001
HR; hazard ratio. a Models adjusted for age group, gender and CHA2DS2-VASc Score. λ 3-Year mortality results for the last calendar period was limited to incident AF cases during 2007 to 2009 with the possibility of 3-year survival ascertainment to 31 December 2012.
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
allowed detailed analysis by calendar period, gender and age specific groups. A fixed 15-year look-back period improves discernment of prevalent from incident cases and increases detection of comorbidities. The unbiased outcome (ie, all-cause mortality) and the large number of events provide assurance that the mortality data are not spurious and accurately represent the longitudinal history of our AF patients. 4.6. Clinical implications The incidence of first-ever hospitalization for AF has declined overall in this contemporary population-based study associated with a declining prevalence of predisposing conditions such as coronary artery disease and heart failure. At the same time, the upward trend in admissions for AF as a principal diagnosis is concerning since it may potentially reflect a real increase in incidence of AF due to changing risk factors such as obesity. Regardless, the global burden of AF is increasing because of growing population prevalence. Given the high prevalent burden and mortality associated with AF, there is a need for better prevention and effective treatments that will reduce mortality. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijcard.2016.01.196. Conflict of interest The authors have no conflicts of interest to declare. Authorship statement Each author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. Acknowledgments The authors wish to thank the staff at the Western Australian Data Linkage Branch, the WA Department of Health Inpatient Data Collection and Epidemiology Branch, and the Registrar General of the WA Department of the Attorney General for the provision of data. This work was supported by a project grant (#1020373) from the National Health and Medical Research Council. LN is the recipient of a Postgraduate Scholarship from the National Health and Medical Research Council of Australia/National Heart Foundation of Australia. References [1] W.B. Kannel, E. Benjamin, Current perceptions of the epidemiology of atrial fibrillation, Cardiol Clin 27 (1) (2009) 13–24. [2] S.S. Chugh, R. Havmoeller, K. Narayanan, D. Singh, M. Rienstra, E.J. Benjamin, et al., Worldwide epidemiology of atrial fibrillation: a global burden of disease 2010 study, Circulation 129 (2014) 837–847. [3] J. Ball, M.J. Carrington, J.J.V. McMurray, S. Stewart, Atrial fibrillation: profile and burden of an evolving epidemic in the 21st century, Int. J. Cardiol. 167 (5) (2013) 1807–1824. [4] S. Stewart, N. Murphy, A. Walker, A. McGuire, J.J.V. McMurray, Cost of an emerging epidemic: an economic analysis of atrial fibrillation in the UK, Heart 90 (3) (2004) 286–292. [5] J.-Y. Le Heuzey, O. Paziaud, O. Piot, M. Ait Said, X. Copie, T. Lavergne, et al., Cost of care distribution in atrial fibrillation patients: the COCAF study, Am. Heart J. 147 (1) (2004) 121–126. [6] K.S. Coyne, C. Paramore, S. Grandy, M. Mercader, M. Reynolds, P. Zimetbaum, Assessing the direct costs of treating nonvalvular atrial fibrillation in the United States, Value Health 9 (5) (2006) 348–356. [7] P.A. Wolf, E.J. Benhamin, A.J. Belanger, W.B. Kannel, D. Levy, R.B. D'Agostino, Secular trends in the prevalence of atrial fibrillation: the Framingham study, Am Heart J 131 (1996) 790–795. [8] S. Stewart, K. MacIntyre, M.M.C. MacLeod, A.E.M. Bailey, S. Capewell, J.J.V. McMurray, Trends in hospital activity, morbidity and case fatality related to atrial fibrillation in Scotland, 1986–1996, Eur. Heart J. 22 (8) (2001) 693–701. [9] W.A. Wattigney, G.A. Mensah, J.B. Croft, Increasing trends in hospitalization for atrial fibrillation in the United States, 1985 through 1999, implications for primary prevention, Circulation 108 (2003) 711–716.
25
[10] F. Khairallah, R. Ezzedine, L.I. Ganz, B. London, S. Saba, Epidemiology and determinants of outcome of admissions for atrial fibrillation in the United States from 1996 to 2001, Am. J. Cardiol. 94 (4) (2004) 500–504. [11] L. Frost, P. Vestergaard, L. Mosekilde, L.S. Mortensen, Trends in incidence and mortality in the hospital diagnosis of atrial fibrillation or flutter in Denmark, 1980–1999, International Journal of Cardiology. 103 (1) (2005) 78–84. [12] K.H. Humphries, C. Jackevicius, Y. Gong, L. Svensen, J. Cox, J.V. Tu, et al., Population rates of hospitalization for atrial fibrillation/flutter in Canada, Can J Cardiol 20 (9) (2004) 869–876. [13] J.P. Piccini, B.G. Hammill, M.F. Sinner, P.N. Jensen, A.F. Hernandez, S.R. Heckbert, et al., Incidence and prevalence of atrial fibrillation and associated mortality among Medicare beneficiaries: 1993–2007, Circ. Cardiovasc. Qual. Outcomes 5 (1) (2012) 85–93. [14] S. Colilla, A. Crow, W. Petkun, D.E. Singer, T. Simon, X. Liu, Estimates of current and future incidence and prevalence of atrial fibrillation in the U.S. adult population, Am. J. Cardiol. 112 (8) (2013) 1142–1147. [15] C.D.J. Holman, A.J. Bass, I.L. Rouse, M.S.T. Hiobbs, Population-based linkage of health records in Western Australia: development of a health services research linked database, Aust. NZ J. Publ. Health 23 (1999) 453–459. [16] M. Knuiman, T. Briffa, M. Divitini, D. Chew, J. Eikelboom, B. McQuillan, et al., A cohort study examination of established and emerging risk factors for atrial fibrillation: the Busselton Health Study, Eur. J. Epidemiol. 29 (3) (2014) 181–190. [17] J.M. Katzenellenbogen, T.H.K. Teng, D. Lopez, J. Hung, M.W. Knuiman, F.M. Sanfilippo, et al., Initial hospitalisation for atrial fibrillation in Aboriginal and nonAboriginal populations in Western Australia, Heart 101 (9) (2015) 712–719. [18] G.Y.H. Lip, R. Nieuwlaat, R. Pisters, D.A. Lane, H.J.G.M. Crijns, Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation, Chest 137 (2) (2010) 263–272. [19] Y. Miyasaka, M.E. Barnes, B.J. Gersh, S.S. Cha, K.R. Bailey, J.B. Seward, et al., Changing trends of hospital utilization in patients after their first episode of atrial fibrillation, Am. J. Cardiol. 102 (5) (2008) 568–572. [20] M.P. Turakhia, M.D. Solomon, M. Jhaveri, P. Davis, M.R. Eber, R. Conrad, et al., Burden, timing, and relationship of cardiovascular hospitalization to mortality among Medicare beneficiaries with newly diagnosed atrial fibrillation, Am. Heart J. 166 (3) (2013) 573–580. [21] C.X. Wong, A.G. Brooks, D.P. Leong, K.C. Roberts-Thomson, P. Sanders, The increasing burden of atrial fibrillation compared with heart failure and myocardial infarction: a 15-year study of all hospitalizations in Australia, Arch. Intern. Med. 172 (2012) 739–741. [22] C.X. Wong, A.G. Brooks, D.H. Lau, D.P. Leong, M.T. Sun, T. Sullivan, et al., Factors associated with the epidemic of hospitalizations due to atrial fibrillation, Am. J. Cardiol. 110 (10) (2012) 1496–1499. [23] L.G. Olsson, K. Swedberg, G. Lappas, S. Stewart, A. Rosengren, Trends in mortality after first hospitalization with atrial fibrillation diagnosis in Sweden 1987 to 2006, Int. J Cardiol. 170 (1) (2014) 75–80. [24] L.J. Nedkoff, T.G. Briffa, D.B. Preen, F.M. Sanfilippo, J. Hung, S.C. Ridout, et al., Age- and sex-specific trends in the incidence of hospitalized acute coronary syndromes in Western Australia, Circulation: Cardiovascular Quality and Outcomes. 4 (5) (2011) 557–564. [25] T.-H.K. Teng, J. Finn, M. Hobbs, J. Hung, Heart failure: incidence, case-fatality and hospitalization rates in Western Australia between 1990 and 2005, Circ Heart Fail 3 (2010) 236–243. [26] N. Wanahita, F.H. Messerli, S. Bangalore, A.S. Gami, V.K. Somers, J.S. Steinberg, Atrial fibrillation and obesity—results of a meta-analysis, Am. Heart J. 155 (2) (2008) 310–315. [27] Y. Miyasaka, M.E. Barnes, K.R. Bailey, S.S. Cha, B.J. Gersh, J.B. Seward, et al., Mortality trends in patients diagnosed with first atrial fibrillation: a 21-year communitybased study, J. Am. Coll. Cardiol. 49 (9) (2007) 986–992. [28] S. Stewart, K. MacIntyre, J.W.T. Chalmers, J. Boyd, A. Finlayson, A. Redpath, et al., Trends in case-fatality in 22 968 patients admitted for the first time with atrial fibrillation in Scotland, 1986–1995, Int. J. Cardiol. 82 (3) (2002) 229–236. [29] T. Briffa, S. Hickling, M. Knuiman, M. Hobbs, J. Hung, F.M. Sanfilippo, et al., Long term survival after evidence based treatment of acute myocardial infarction and revascularisation: follow-up of population based Perth MONICA cohort, 1984– 2005, BMJ 338 (jan26_2) (2009) (b36-.). [30] T.-H.K. Teng, J. Hung, M. Knuiman, S. Stewart, L. Arnolda, I. Jacobs, et al., Trends in long-term cardiovascular mortality and morbidity in men and women with heart failure of ischemic versus non-ischemic aetiology in Western Australia between 1990 and 2005, Int J Cardiol 158 (3) (2012) 405–410. [31] R.G. Hart, O. Benavente, R. McBride, L.A. Pearce, Antithrombotic therapy to prevent stroke in patients with atrial fibrillation: a meta-analysis, Ann. Intern. Med. 131 (7) (1999) 492–501. [32] Piccini JP, Fraulo ES, Ansell JE, Fonarow GC, Gersh BJ, Go AS, et al. Outcomes registry for better informed treatment of atrial fibrillation: rationale and design of ORBIT-AF. Am. Heart J. 162(4):(606–12.e1). [33] A.K. Kakkar, I. Mueller, J.-P. Bassand, D.A. Fitzmaurice, S.Z. Goldhaber, S. Goto, et al., International longitudinal registry of patients with atrial fibrillation at risk of stroke: Global Anticoagulant Registry in the FIELD (GARFIELD), Am. Heart J. 163 (1) (2012) (13–9.e1). [34] A. Clark, D.B. Preen, J.Q. Ng, J.B. Semmens, C.D.J. Holman, Is Western Australia representative of other Australian States and Territories in terms of key sociodemographic and health economic indicators? Aust. Health Rev. 34 (2) (2010) 210–215.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Trends in incidence and prevalence of hospitalization for atrial fibrillation and associated mortality in Western Australia, 1995–2010☆ Tom Briffa a,1, Joseph Hung b,⁎,1, Matthew Knuiman a, Brendan McQuillan b, Derek P. Chew c, John Eikelboom d, Graeme J. Hankey a, Tiew-Hwa K. Teng e, Lee Nedkoff a, Rukshen Weerasooriya f, Andrew Liu g, Paul Stobie h a
School of Population Health, The University of Western Australia, Crawley, Western Australia, Australia School of Medicine and Pharmacology, Sir Charles Gairdner Hospital Unit, The University of Western Australia, Crawley, Western Australia, Australia School of Medicine, Flinders University of South Australia, Adelaide, South Australia, Australia d Department of Medicine, McMaster University, Hamilton, Canada e Western Australian Centre for Rural Health, The University of Western Australia, Crawley, Western Australia, Australia f Department of Cardiology, Royal Perth Hospital, Perth, Western Australia, Australia g Department of Cardiology, Fiona Stanley Hospital, Murdoch, Western Australia, Australia h Department of Cardiovascular Medicine, Sir Charles Gairdner Hospital, Nedlands, Western Australia, Australia b c
a r t i c l e
i n f o
Article history: Received 5 June 2015 Received in revised form 18 December 2015 Accepted 15 January 2016 Available online 28 January 2016 Keywords: Atrial fibrillation Hospitalization Incidence Prevalence Mortality
a b s t r a c t Objective: Hospitalization for atrial fibrillation (AF) is a large and growing public health problem. We examined current trends in the incidence, prevalence, and associated mortality of first-ever hospitalization for AF. Methods: Linked hospital admission data were used to identify all Western Australia residents aged 35–84 years with prevalent AF and incident (first-ever) hospitalization for AF as a principal or secondary diagnosis during 1995–2010. Results: There were 57,552 incident hospitalizations, mean age 69.8 years, with 41.4% women. Over the calendar periods, age- and sex-standardized incidence of hospitalization for AF as any diagnosis declined annually by 1.1% (95% CI; 0.93, 1.29), while incident AF as a principal diagnosis increased annually by 1.2% (95% CI; 0.84, 1.50). Incident AF hospitalization was higher among men than women, and 15-fold higher in the 75–84 compared with 35–64 year age group. The age- and sex-standardized prevalence of AF increased annually by 2.0% (95% CI; 1.88, 2.03) over the same period. Comorbidity trends were mixed with diabetes and valvular heart disease increasing, and hypertension, coronary artery disease, heart failure, cerebrovascular disease, and chronic kidney disease decreasing. The 1-year all-cause mortality after incident AF hospitalization declined from 17.6% to 14.6% (trend P b 0.001), with an adjusted hazard ratio of 0.86 (95% CI; 0.81, 0.91). Conclusion: This contemporary study shows that incident AF hospitalization is not increasing except for AF as a principal diagnosis, while population prevalence of hospitalized AF has risen substantially. The high 1-year mortality following incident AF hospitalization has improved only modestly over the recent period. © 2016 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Atrial fibrillation (AF) is the most common chronic arrhythmia and is associated with substantial morbidity and mortality [1–3]. The increasing prevalence of AF has been attributed to a combination of ☆ Grant support: This work was supported by a project grant (#1020373) from the National Health and Medical Research Council. The funding body had no direct role in the study design, the collection, analysis, or interpretation of data, and the writing of the report. ⁎ Corresponding author at: School of Medicine & Pharmacology (SCGH Unit) M503, Harry Perkins Institute of Medical Research, QQ Block, QEII Medical Centre, Verdun Street, Nedlands, Western Australia 6009, Australia. E-mail address: [email protected] (J. Hung). 1 Denotes equal first co-authors.
http://dx.doi.org/10.1016/j.ijcard.2016.01.196 0167-5273/© 2016 Elsevier Ireland Ltd. All rights reserved.
population aging, changing pattern of risk factors, and improved survival from contributory cardiovascular and non-cardiovascular diseases [1–3]. Accurate estimates of trends in the incidence of AF, disease burden in the population and its effect on mortality are necessary to effectively manage this significant and growing public health problem. Notably hospitalizations account for most of the costs (over 50%) associated with AF [4–6]. Studies conducted from the 1980s to 1990s from North America and Western Europe have reported substantial increases in the prevalence and incidence of hospitalizations for AF either as a principal diagnosis or any diagnosis [7–12]. However, a recent retrospective cohort study of Medicare beneficiaries in North America, aged 65 years and older, reported that the incidence of AF has in fact remained steady between 1993 and 2007 [13]. Hence, it is unclear if the observed rise in AF
20
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
Table 1 Characteristics of patients hospitalized with incident atrial fibrillation as any discharge diagnosis stratified by calendar period from 1995 to 2010 in Western Australia. Characteristic, n (%) unless specified
Overall n = 57,552
Age, years, mean ± SD Age groups 35–64 years 65–74 75–84 Women Principal discharge diagnosis Emergency admission Comorbidities Coronary heart disease Cerebrovascular disease Peripheral arterial disease Heart failure Valvular heart disease Hypertension Diabetes Chronic kidney disease Cancer Chronic obstructive pulmonary disease CHA2DS2-VASc score, mean ± SD CHA2DS2-VASc score categories 0 points 1 2 or more
1995 to 1998 n = 13,205
1999 to 2002 n = 13,502
2003 to 2006 n = 14,352
2007 to 2010 n = 16,493
69.8 (10.8)
70.3 (10.3)
69.9 (10.7)
69.9 (10.8)
69.3 (11.1)
b.0001
15,823 (27.5) 18,078 (31.4) 23,651 (41.1) 23,808 (41.4) 17,660 (30.7) 37,258 (64.7)
3232 (24.5) 4552 (34.5) 5421 (41.1) 5605 (42.4) 3375 (25.6) 8213 (62.2)
3577 (26.5) 4366 (32.3) 5559 (41.2) 5547 (41.1) 4206 (31.2) 8755 (64.8)
3980 (27.7) 4308 (30.0) 6064 (42.3) 5959 (41.5) 4449 (31.0) 9462 (65.9)
5034 (30.5) 4852 (29.4) 6607 (40.1) 6697 (40.6) 5630 (34.1) 10,828 (65.6)
b.0001 0.012 b.0001 b.0001
22,143 (38.5) 7629 (13.3) 6712 (11.7) 14,560 (25.3) 12,031 (20.9) 28,739 (49.9) 11,195 (19.4) 5929 (10.3) 18,767 (32.6) 10,574 (18.4) 3.1 (1.9)
5722 (43.3) 2045 (15.5) 1789 (13.6) 4046 (30.6) 1819 (13.8) 6749 (51.1) 2139 (16.2) 1693 (12.8) 3411 (25.8) 3048 (23.1) 3.2 (1.9)
5416 (40.1) 1913(14.2) 1643 (12.2) 3592 (26.6) 2981 (22.1) 6695 (49.6) 2601 (19.3) 1336 (9.9) 4131 (30.6) 2759 (20.4) 3.1 (1.9)
5485 (38.2) 1839 (12.8) 1630 (11.4) 3521 (24.5) 3465 (24.1) 7219 (50.3) 2983 (20.8) 1378 (9.6) 4927 (34.3) 2433 (17.0) 3.1 (1.9)
5520 (33.5) 1832 (11.1) 1650 (10.0) 3401 (20.6) 3766 (22.8) 8076 (49.0) 3472 (21.1) 1522 (9.2) 6298 (38.2) 2334 (14.2) 2. 9 (1.9)
b.0001 b.0001 b.0001 b.0001 b.0001 0.002 b.0001 b.0001 b.0001 b.0001 b.0001
7055 (12.3) 5315 (9.2) 45,182 (78.5)
1236 (9.4) 1102 (8.3) 10,867 (82.3)
1522 (11.3) 1276 (9.4) 10,704 (79.3)
1775 (12.4) 1336 (9.3) 11,241 (78.3)
2522 (15.3) 1601 (9.7) 12,370 (75.0)
b .0001
prevalence is due to an increase in incident (de novo) AF, increasing survival after new AF diagnosis or can be largely attributed to population aging. Furthermore, there is significant uncertainty regarding projected increases in AF incidence and prevalence over the ensuing decades which undermines proper healthcare planning [14]. We therefore examined trends in the incidence of first-ever hospitalization for AF as a discharge diagnosis, and population prevalence of people with prior AF admission, among residents of Western Australia (WA), aged 35 to 84 years, from 1995 through 2010. We also assessed trends in antecedent risk factors, comorbid diseases, and all-cause mortality among incident AF cases.
Trend P value
2. Methods 2.1. Data sources Hospital admissions and mortality data were obtained from 1 January 1980 to 31 December 2012 from the WA Data Linkage System for residents in Western Australia [15]. An individual's administrative records are linked with a N 99% accuracy using probabilistic matching on key personal identifiers [15]. All records include dates of care, principal diagnosis, and up to 20 secondary discharge diagnosis codes from the International Classification of Diseases, versions nine (ICD-9), nine-
Table 2 Incidence of hospitalization for atrial fibrillation as a discharge diagnosis stratified by calendar periods from 1995 to 2010 in Western Australia. No. of persons with atrial fibrillation as any discharge diagnosis (Incidence rate per 1000 person-years in each perioda)
Mean annual % change (95% confidence interval)λ
P value
Characteristic, n (rate/1000) 1995 to 1998
1999 to 2002
2003 to 2006
2007 to 2010
Overall Gender Men Women Age-specific 35–64 years 65–74 75–84
n = 13,205 (4.39)
n = 13,502 (3.99)
n = 14,352 (3.73)
n = 16,493 (3.83)
−1.1 (−0.93, −1.29)
b.0001
7604 (5.56) 5606 (3.30)
7955 (5.10) 5547 (2.97)
8393 (4.66) 5959 (2.86)
9797 (4.79) 6696 (2.92)
−1.21 (−0.98, −1.44) −0.87 (−0. 60, −1.15)
b.0001 b.0001
3235 (1.46) 4554 (10.88) 5421 (26.39)
3577 (1.38) 4366 (9.85) 5559 (23.52 )
3980 (1.33) 4308 (8.95) 6064 (22.07)
5033 (1.46) 0.19 (−0.15, 0.54) 4851 (8.88) −1.66 (−1.35, −1.97) 6609 (22.08) −1.44 (−1.17, −1.72)
0.27 b.0001 b.0001
No. of persons with atrial fibrillation as a principal discharge diagnosis (Incidence rate per 1000 person-yearsa)
Mean annual % change (95% confidence interval)λ
P value
Characteristic, n (rate/1000)
1995 to 1998
1999 to 2002
2003 to 2006
2007 to 2010
Overall Gender Men Women Age-specific 35–64 years 65–74 75–84
n = 3375 (1.11)
n = 4206 (1.24)
n = 4449 (1.16)
n = 5630 (1.31)
1.17 (0.84, 1.50)
b.0001
1870 (1.31) 1506 (0.92)
2486 (1.55) 1720 (0.94)
2623 (1.43) 1826 (0.90)
3317 (1.59) 2313 (1.02)
1.33 (0.90, 1.57) 1.05 (0.55, 1.55)
b.0001 b.0001
1293 (0.57) 1122 (2.68) 961 (4.47)
1632 (0.62) 1332 (3.02) 1242 (5.16)
1811 (0.60) 1343 (2.79) 1295 (4.68)
2397 (0.69) 1641 (2.99) 1592 (5.30)
1.66 (1.14, 2.18) 0.84 (0.27, 1.43) 1.02 (0.41, 1.64)
b.0001 0.0042 0.0010
a Overall rates are age and sex standardized to the AF-free WA population aged 35 to 84 years in 2010; rates for men and women are age standardized; age-specific rates are sex standardized. λ Poisson models adjusted for 5-year age group and gender.
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
clinical modification (ICD-9-CM), and ten-Australian modification (ICD10-AM). We also identified death dates in the person-linked file containing complete mortality records through to 31 December 2012. 2.2. Incident and prevalent cases of AF We identified cases of atrial fibrillation/atrial flutter based on either a principal or secondary discharge diagnosis code (ICD-9 code 427.3; ICD-10-AM code I48) during the index admission period from 1 January 1995 to 31 December 2010. [16,17] Medical chart review of 184 randomly selected hospital records from 1998, 2003, and 2008 with a principal or secondary discharge diagnosis of AF established a positive predictive value of 84.6% when based on a confirmatory ECG in the medical record of the same admission and 98.9% when based on an ECG or written record of AF in the medical notes or discharge letter. AF cases were considered as incident (first-ever) if there was no hospitalization with AF coded in any discharge diagnosis field dating back a fixed 15-year period from the index date of admission. We also identified a subset of incident AF cases where AF was coded in the principal diagnosis field only. Conversely, an individual was considered to have prevalent AF if it was coded on any hospital discharge record in the 15 years prior to June 30 of a given calendar year and the person was still alive and residing in WA at that time. Mid-year estimates of the WA resident population by gender and 5-year age groups for each of the years, 1995 to 2010, were sourced from the Australian Bureau of Statistics, and represented the denominators for the calculation of prevalence. For incidence calculation in any calendar year, the denominator was the corresponding WA population count after prevalence correction, ie, after subtracting the number of prevalent AF cases. We calculated annual incidence per 1000 person-years and prevalence per 1000 persons as well as the average annual incidence and prevalence for consecutive 4-year calendar periods (1995–1998, 1999–2002, 2003– 2006 and 2007–2010).
21
period changes in categorical variables. Finally, we used Cox proportional hazards models to obtain mortality hazard ratios (HRs) with 95% confidence intervals (95% CIs) comparing the mortality risk in each calendar period versus the first after having satisfied that the assumption of proportional hazards was met. Covariates included were 5-year age group, gender, CHA2DS2-VASc score, ICD-9 period, admission year, and an admission year ∗ ICD-9 period interaction term.
2.3. Comorbidities For individuals with incident AF, we identified comorbidities dating back a fixed 15 years from the index admission or concurrent with the index AF admission. Comorbidities were identified from the relevant ICD codes for coronary artery disease, cerebrovascular disease, peripheral arterial disease, heart failure, valvular heart disease, hypertension, diabetes, chronic kidney disease, all cancers, and chronic obstructive pulmonary disease. Details of the ICD codes have been previously reported. [16,17]. A CHA2DS2-VASc score (congestive heart failure, hypertension, age ≥ 75 years, diabetes mellitus, stroke/transient ischemic attack, vascular disease, age 65–74 years, sex category) was calculated for each person, except that female gender received a score of 1 only if there was an additional risk factor. [18]. 2.4. Statistical analysis We calculated age- and sex-standardized incidence and prevalence of hospitalized AF overall, and by sex and age-specific groups (35–64, 65–74, and 75–84 years), with rates standardized to the age and sex distributions of the AF-free WA population in 2010 for incidence and the WA population in 2010 for prevalence. We estimated the average annual change in incidence rates and prevalence using age- and sex-adjusted Poisson and logistic regression models respectively and to determine the statistical significance of these trends. A calendar year ∗ ICD-9 interaction was tested to see if the trend changed from mid-1999 when the ICD9 coding system was replaced by the ICD10 system. For individuals with incident AF, we summarized demographic and clinical characteristics, comorbidities and discharge diagnosis category using means with standard deviations or frequency distribution with percentages. We used linear models to test for significant temporal trends in continuous variables, and the chi-squared statistic to test for
Fig. 1. Incidence of hospitalization with atrial fibrillation as any discharge diagnosis by calendar year, 1995 to 2010; (A) in the overall population, (B) by sex, and (C) by agespecific groups. The dashed lines in panel A represent 95% confidence limits.
22
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
We used SAS version 9.4 for all analyses (SAS Institute Inc., Cary, NC). Ethics committees of the Western Australian Department of Health, the University of Western Australia, Sir Charles Gairdner Hospital, Royal Perth Hospital, Fremantle Hospital, and Hollywood Private Hospital, Western Australia approved the study. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institution's human research committee.
period (P b 0.001) (Table 2). The increasing incidence was consistent across all subgroups with the 35–64 year olds showing the largest change (+1.7% annually) (Table 2). Modeling with inclusion of the calendar year ∗ ICD-9 interaction term showed that the estimated annual incidence trend was not significantly different in the ICD-9 and ICD-10 periods for AF as a principal or any discharge diagnosis. 3.3. Prevalence
3. Results 3.1. Patient characteristics Table 1 shows the characteristics of the 57,552 first-ever hospitalized AF cases, mean age 69.8 years, with 41.4% women, stratified by consecutive 4-year calendar periods from 1995 to 2010. Patients aged 65 years or older comprised more than two thirds of the incident AF cases. The age and sex distribution changed over the study period with an increasing proportion of cases in the youngest age group (35– 64 years) and a lower proportion of women. The proportion of incident cases with AF as a principal discharge diagnosis increased from 25.6% to 34.1% over the study period. Comorbidity trends were varied with diabetes, valvular heart disease, and cancer increasing, but coronary artery disease, heart failure, cerebrovascular disease, peripheral arterial disease, and chronic obstructive pulmonary disease decreasing. Although hypertension declined it remained highly prevalent (≈50%) over the study period. The mean CHA2DS2-VASc score decreased across calendar periods with the proportion of cases with a score ≥ 2 declining from 82.3% to 75.0% (p b 0.001). Compared to the whole cohort, the subset of people hospitalized with AF as a principal diagnosis (17,660; 30.7%), were younger (age mean 66.1, SD 11.8 years), with a similar proportion of women (41.7%), a lower prevalence of all comorbidities but with similar temporal trends, and a lower mean CHA2DS2-VASc score including proportion of people (62.1%) with a CHA2DS2-VASc score ≥ 2 (Supplemental Table 1).
The age- and sex-standardized prevalence of hospitalized AF cases increased overall from 20.2 to 25.6 per 1000 persons over the calendar periods, representing a 1.96% (95% CI: 1.88 to 2.03) annual increase (p b 0.001) (Table 3, Fig. 2A). The prevalence increased similarly in both genders and all age subgroups (Table 3, Fig. 2B and C). By the last calendar period, the prevalence of AF was still 1.8-fold higher in men than women and was 15-fold higher in the highest versus lowest age group (Table 3). 3.4. Mortality The overall 30-day, 1-year and 3-year all-cause mortality over the study period was 7.1%, 16.6%, and 27.1% respectively, and declined modestly across the calendar periods (all P b 0.001) (Table 4). Comparing the last to first calendar period, the age, sex, and CHA2DS2-VASc riskadjusted HR was 0.91 (95% CI; 0.83, 1.00) for death at 30 days, 0.86 (95% CI; 0.81, 0.91) for death at 1 year and 0.82 (95%CI; 0.78, 0.86) for death at 3 years (Table 4). The overall all-cause mortality in the subset with incident AF as a principal discharge diagnosis was considerably lower than the entire cohort being 0.9%, 4.7%, and 11.1% at 30 day, 1 year, and 3 years respectively. There was also a significant temporal reduction in crude 1-year and 3-year mortality in this subset and their reduced risk-adjusted HRs for death were more apparent and significant in the latter calendar periods than the overall cohort (Table 4). 4. Discussion
3.2. Incidence The age- and sex-standardized incidence of hospitalization for AF as any diagnosis declined overall from 4.39 to 3.83 per 1000 person-years over the calendar periods, representing a 1.1% (95%CI: 0.93, 1.29) annual rate decline (p b 0.001) (Table 2, Fig. 1A). Incidence was consistently higher among men than women and more than 15-fold higher in the 75–84 compared with 34–64 year age group (Fig. 1B and C). The incidence declined significantly in both genders and the 65–74 and 74– 84 year age groups (all P b 0.001), but was stable in the 35–64 year olds (Table 2). By contrast, the age- and sex-standardized incidence of hospitalization for AF as a principal diagnosis increased annually by 1.17% (95% CI: 0.84, 1.50) from 1.11 to 1.31 per 1000 person-years over the study
In this WA population-based study we found that incidence of hospitalization with AF did not increase from 1995 through 2010, except in people with AF admission as a principal diagnosis and particularly younger adults (aged b65 years). However, steadily increasing prevalence of people with hospitalized AF is adding to the overall population burden of this serious arrhythmia. Incident AF hospitalization is still associated with a high short and long-term mortality which has improved only modestly over the recent period. 4.1. Incidence Previous reports have highlighted a growing prevalence of AF although it is unclear if this is largely driven by an aging population or
Table 3 Prevalence of persons with prior hospital diagnosis of atrial fibrillation stratified by calendar periods from 1995 to 2010 in Western Australia. No. of persons (average annual age-standardized prevalence rate per 1000 Persons in each perioda)
Mean annual % change (95% confidence interval)λ
P value
Characteristic, n (rate/1000)
1995 to 1998
1999 to 2002
2003 to 2006
2007 to 2010
Overall Gender Men Women Age groups 35–64 years 65–74 75–84
n = 60,233 (20.2)
n = 79,688 (23.5)
n = 96,081 (24.7)
n = 112,255 (25.6)
1.96 (1.88, 2.03)
b.0001
36,677 (26.3) 23,556 (14.0)
48,591 (30.4) 31,097 (16.7)
59,010 (31.7) 37,071 (17.7)
69,953 (32.9) 42,302 (18.3)
1.76 (1.66, 1.85) 2.04 (1.92, 2.16)
b.0001 b.0001
15,537 (6.9) 19,950 (45.2) 24,746 (106.4)
20,993 (8.0) 25,153 (53.4) 33,542 (125.2)
25,690 (8.5) 29,041 (56.8) 41,350 (131.3)
31,208 (9.0) 34,105 (58.7) 46,942 (135.8)
2.10 (2.24, 1.95) 2.00 (1.87, 2.13) 1.83 (1.72, 1.95)
b.0001 b.0001 b.0001
a Overall prevalence is age and sex standardized to the whole WA population aged 35 to 84 years in 2010; rates for men and women are age standardized; age-specific rates are sex standardized. λ Logistic regression models adjusted for 5-year age group and gender.
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
changing incidence [2,3,7–12]. There also remains a wide range of uncertainty around the magnitude of future trends in incidence of AF [14]. It is known that cardiovascular hospitalizations are frequent following initial AF diagnosis, [19,20] and therefore trends in incident hospitalization with AF may reflect underlying population incidence. Additionally, it is recognized that hospitalizations represent the major component of the healthcare costs related to AF [4–6]. Contrary to earlier studies, we actually found a small downward trend in age and sex-standardized incidence of admissions with AF, falling by nearly 1.1% annually from 1995 through 2010. Our findings are supported by the only other study which has reported on the incidence of AF in this contemporary era although this study was restricted to a representative 5% sample of Medicare beneficiaries in the United States aged 65 years and older [13]. They also found that the incidence rate of any AF diagnosis in both inpatient and outpatient settings was relatively stable from 1993 through to 2007 [13]. A novel finding from our study was that the incidence of admission with AF as a principal diagnosis increased by 1.2% annually over the same period, and incidence increased in both sexes and across all agespecific groups but most in the 35–64 year age group. In a national study, Wong et al. [21] examined all hospitalizations with a principal diagnosis of AF in the National Hospital Morbidity Data set from the Australian Institute of Health and Welfare, and reported that the number of hospitalizations increased by 7.9% annually from 1993 through 2007, of which AF related procedures constituted only a small percentage of total hospitalizations. However, the study could not distinguish between de novo and repeat hospitalizations. Since our local data for incident admissions are likely to be nationally representative, this suggests that repeat hospitalizations are the main driver for the substantial increases in total hospitalization numbers seen across Australia. This is concerning as recurrent admissions, particularly if unplanned, add to the cumulative healthcare costs and resource burden related to AF. Further studies addressing the causes of increasing AF admissions are therefore needed. Miysaka et al. [19] suggested that the marked increase in hospitalization after first AF diagnosis from 1980 to 2000 in Olmsted County was largely driven by a changing practice pattern in AF management, while Wong et al. [22] suggested that the main cause for expanding AF hospitalizations in Australia was increased hospitalizations in older people affected by AF rather than changing procedural trends.
23
suggested that higher obesity rates in urbanized populations may well underlie an increasing incidence of AF seen worldwide [2,16,26]. Coincidentally we observed an increasing prevalence of diabetes which may reflect a rising obesity rate in our population. 4.3. Prevalence We found that the prevalence of hospitalized AF was higher among men than women and increased exponentially with increasing age,
4.2. Effect of comorbidities Our incident AF hospitalized cases had a high prevalence of predisposing risk factors (such as hypertension and diabetes), and antecedent conditions (such as atherosclerotic vascular disease and heart failure). However, with the exception of diabetes and valvular heart disease, the prevalence of these comorbidities declined significantly over the study period which is also reflected in the falling individual CHA2DS2VASc scores. Similar to other studies, [11,13,23] coronary artery disease and heart failure are the two most common comorbidities associated with AF hospitalization but their incidence have declined substantially over the study period in the WA population [24,25]. It is very likely that this has had a significant impact on the incidence of AF hospitalization particularly as a secondary discharge diagnosis. Nevertheless, we observed an upward trend in the incidence and proportion of incident AF admissions as a principal diagnosis, particularly affecting younger adults (age b 65 years). Previous investigators [8] have speculated that the publication of clinical trials demonstrating the efficacy of antithrombotic therapy in the 1990s may have increased physician awareness of AF and resulted in an increase in hospital admission primarily for AF, as well as influencing coding practice especially secondary coding. We are however unable to ascertain if the increase in incident hospitalizations for AF as a principal diagnosis represents a lower admission threshold, an improved diagnosis rate, or a true increase in incidence of AF particularly in younger adults. It has been
Fig. 2. Prevalence of persons with any hospital diagnosis of atrial fibrillation by calendar year, 1995 to 2010; (A) in the overall population, (B) by sex, and (C) by age-specific groups. The dashed lines in panel A represent 95% confidence limits.
24
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
which is concordant with global studies of AF prevalence [2,3]. Given the slightly declining overall incidence rates, our observed ≈ 30% increase in age- and sex-standardized prevalence of people with hospitalized AF is likely to reflect increased long-term survival after incident AF hospitalization. Piccini et al. [13] reported in their Medicare 5% sample a similar growth (N 25%) in prevalent AF cases over a 15-year period when using a stricter definition of AF as the primary diagnosis. By the last calendar period, we estimated that the age-adjusted prevalence of hospitalized AF was around 3.3% in men and 1.8% in women (32.9 and 18.3 per 1000 persons respectively). This is in line with a recent review [3] which estimated a global pooled (age-adjusted) AF prevalence of around 3.3% in men and 2.4% in women, but higher than the prevalence estimate of 1.4% in men and 0.9% in women aged ≥ 35 years reported from a systematic review of populationbased studies in 2010 [2]. We suggest that while hospital administrative data will underestimate real AF prevalence, they are still valuable for monitoring the population burden of AF and its impact on healthcare systems. 4.4. Mortality Earlier community and hospital registry-based studies have established that a new diagnosis of AF is associated with a high short and long-term case-fatality [11,27,28]. In this contemporary study, the overall 1-year and 3-year mortality rate after incident AF hospitalization was 16.6% and 27.1% respectively. Similar to previous studies, [8,23] we found that mortality rates were considerably higher for people admitted with AF as a secondary compared with principal diagnosis, presumably because of older patient age and increased comorbidities such as stroke and myocardial infarction with a high case fatality. Although we established that mortality rates declined, a significant reduction in risk-adjusted mortality hazard ensued mainly during the latter calendar periods. By comparison, the Medicare 5% sample study of an older population (mean age 80 years) found no significant reduction in standardized mortality ratios from 1997 through 2007 [13]. Another contemporary registry study from Sweden from 1987 through 2006 confirmed the poor prognosis associated with AF, but also found that 3-year all-cause mortality after first-ever AF hospitalization had significantly declined throughout the period [23]. While the reasons for the improved survival are unclear, the investigators observed that many of the important comorbidities occurring in AF patients have also shown prognostic improvements during the same period [23]. Similarly
we have reported improved long-term survival of patients with coronary artery disease and heart failure in WA during the same era [29,30]. Oral anticoagulants are the only therapy that have been shown in randomized trials to decrease all-cause mortality in patients with nonvalvular AF [31]. However, data on the prescribing rates of oral anticoagulants for this condition are lacking in Australia although it is likely to reflect patterns seen in other western countries [32,33]. Patients admitted with a principal diagnosis of AF formed an increasing proportion of incident AF cases admitted during the more recent calendar periods and these younger lower risk cases could also contribute to an overall lower mortality in the cohort. The fact that their risk-adjusted hazard for death at 1 and 3 years is significantly reduced suggests that this may have occurred in part due to an improvement in clinical management of AF in this subset. 4.5. Limitations and strengths Our study has several limitations. We are unable to differentiate between subtypes of AF using administrative data. Incidence and prevalence estimates from hospital morbidity datasets will underestimate actual incidence and prevalence in the community, although temporal trends may still be revealing. Nonetheless, hospital data are crucial to assessment of AF burden and costs and for healthcare planning. We excluded people aged 85 years and older because hospital morbidity data are considered less reliable in the very elderly. The diagnosis of AF was not formally validated except in a small randomly selected sample which indicated a high positive predictive value based on medical record review. Coding practices change over time, including a change from ICD-9 to ICD-10 that occurred in mid-1999. However, we found that the estimated annual change in incidence rate was not significantly different in the ICD-9 and ICD-10 periods. There is likely underreporting of several important comorbidities, most notably hypertension but also diabetes and heart failure, and therefore CHA2DS2-VASc scores will also be underestimated. Our local hospital incident and prevalent trends may not be representative of trends in other populations especially from developing countries. We are also unable to ascertain the direct impact on incident hospitalization rates resulting from improved diagnostic methods or changing clinical practice or the direct impact of improved AF management on changing mortality. The major advantages of our study derive from the populationbased design in a state which is nationally representative [34], the high-quality person-based linkage [15], and the large sample size that
Table 4 Estimated all-cause mortality risk after incident hospitalization for atrial fibrillation as a discharge diagnosis stratified by calendar periods from 1995 to 2010 in Western Australia. Total mortality for AF as any diagnosis for each calendar period and hazard ratios compared to first calendar perioda All-cause deaths
1995 to 1998 (N = 13,205)
1999 to 2002 (N = 13,502)
30-Day, n (%) HR (95% CI) 1-Year, n (%) HR (95% CI) 3-Year, n (%) HR (95% CI)
935 (7.1) 1.00 2322 (17.6) 1.00 3879 (29.4) 1.00
1059 (7.8) 1.15 (1.05, 1.26) 2455 (18.2) 1.07 (1.01, 1.13) 3837 (28.4) 1.00 (0.95, 1.04)
2003 to 2006 (N = 14,352) P = 0.002 P = 0.023 P = 0.94
1064 (7.4) 1.08 (0.99, 1.18) 2378 (16.6) 0.96 (0.91, 1.02) 3772 (26.3) 0.92 (0.86, 0.94)
2007 to 2010 (N = 16,493) P = 0.08 P = 0.18 P b .0001
1007 (6.1) 0.91 (0.83, 1.00) 2401 (14.6) 0.86 (0.81, 0.91) 2925 (24.1)λ 0.82 (0.78, 0.86)
P b 0.001 P = 0.049 P b 0.001 P b .0001 P b 0.001 P b .0001
Total mortality for AF as a principal diagnosis for each calendar period and hazard ratios compared to first calendar perioda All-cause deaths
1995 to 1998 (N = 3375)
1999 to 2002 (N = 4206)
30-Day, n (%) HR (95% CI) 1-Year, n (%) HR (95% CI) 3-Year, n (%) HR (95% CI)
36 (1.1) 1.00 197 (5.8) 1.00 438 (13.0) 1.00
42 (1.0) 0.92 (0.59, 1.46) 207 (4.9) 0.83 (0.68, 1.01) 491 (11.7) 0.88 (0.78, 1.00)
2003 to 2006 (N = 4449) P = 0.74 P = 0.063 P = 0.06
45 (1.0) 0.98 (0.63, 1.52) 200 (4.5) 0.76 (0.63, 0.93) 469 (10.5) 0.80 (0.70, 0.92)
2007 to 2010 (N = 5630) P = 0.92 P = 0.008 P = 0.001
42 (0.8) 0.76 (0.48, 1.19) 231 (4.1) 0.72 (0.60, 0.88) 389 (9.6)λ 0.74 (0.65, 0.85)
P = 0.35 P = 0.23 P = 0.002 P = 0.001 P b .0001 P b .0001
HR; hazard ratio. a Models adjusted for age group, gender and CHA2DS2-VASc Score. λ 3-Year mortality results for the last calendar period was limited to incident AF cases during 2007 to 2009 with the possibility of 3-year survival ascertainment to 31 December 2012.
T. Briffa et al. / International Journal of Cardiology 208 (2016) 19–25
allowed detailed analysis by calendar period, gender and age specific groups. A fixed 15-year look-back period improves discernment of prevalent from incident cases and increases detection of comorbidities. The unbiased outcome (ie, all-cause mortality) and the large number of events provide assurance that the mortality data are not spurious and accurately represent the longitudinal history of our AF patients. 4.6. Clinical implications The incidence of first-ever hospitalization for AF has declined overall in this contemporary population-based study associated with a declining prevalence of predisposing conditions such as coronary artery disease and heart failure. At the same time, the upward trend in admissions for AF as a principal diagnosis is concerning since it may potentially reflect a real increase in incidence of AF due to changing risk factors such as obesity. Regardless, the global burden of AF is increasing because of growing population prevalence. Given the high prevalent burden and mortality associated with AF, there is a need for better prevention and effective treatments that will reduce mortality. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijcard.2016.01.196. Conflict of interest The authors have no conflicts of interest to declare. Authorship statement Each author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. Acknowledgments The authors wish to thank the staff at the Western Australian Data Linkage Branch, the WA Department of Health Inpatient Data Collection and Epidemiology Branch, and the Registrar General of the WA Department of the Attorney General for the provision of data. This work was supported by a project grant (#1020373) from the National Health and Medical Research Council. LN is the recipient of a Postgraduate Scholarship from the National Health and Medical Research Council of Australia/National Heart Foundation of Australia. References [1] W.B. Kannel, E. Benjamin, Current perceptions of the epidemiology of atrial fibrillation, Cardiol Clin 27 (1) (2009) 13–24. [2] S.S. Chugh, R. Havmoeller, K. Narayanan, D. Singh, M. Rienstra, E.J. Benjamin, et al., Worldwide epidemiology of atrial fibrillation: a global burden of disease 2010 study, Circulation 129 (2014) 837–847. [3] J. Ball, M.J. Carrington, J.J.V. McMurray, S. Stewart, Atrial fibrillation: profile and burden of an evolving epidemic in the 21st century, Int. J. Cardiol. 167 (5) (2013) 1807–1824. [4] S. Stewart, N. Murphy, A. Walker, A. McGuire, J.J.V. McMurray, Cost of an emerging epidemic: an economic analysis of atrial fibrillation in the UK, Heart 90 (3) (2004) 286–292. [5] J.-Y. Le Heuzey, O. Paziaud, O. Piot, M. Ait Said, X. Copie, T. Lavergne, et al., Cost of care distribution in atrial fibrillation patients: the COCAF study, Am. Heart J. 147 (1) (2004) 121–126. [6] K.S. Coyne, C. Paramore, S. Grandy, M. Mercader, M. Reynolds, P. Zimetbaum, Assessing the direct costs of treating nonvalvular atrial fibrillation in the United States, Value Health 9 (5) (2006) 348–356. [7] P.A. Wolf, E.J. Benhamin, A.J. Belanger, W.B. Kannel, D. Levy, R.B. D'Agostino, Secular trends in the prevalence of atrial fibrillation: the Framingham study, Am Heart J 131 (1996) 790–795. [8] S. Stewart, K. MacIntyre, M.M.C. MacLeod, A.E.M. Bailey, S. Capewell, J.J.V. McMurray, Trends in hospital activity, morbidity and case fatality related to atrial fibrillation in Scotland, 1986–1996, Eur. Heart J. 22 (8) (2001) 693–701. [9] W.A. Wattigney, G.A. Mensah, J.B. Croft, Increasing trends in hospitalization for atrial fibrillation in the United States, 1985 through 1999, implications for primary prevention, Circulation 108 (2003) 711–716.
25
[10] F. Khairallah, R. Ezzedine, L.I. Ganz, B. London, S. Saba, Epidemiology and determinants of outcome of admissions for atrial fibrillation in the United States from 1996 to 2001, Am. J. Cardiol. 94 (4) (2004) 500–504. [11] L. Frost, P. Vestergaard, L. Mosekilde, L.S. Mortensen, Trends in incidence and mortality in the hospital diagnosis of atrial fibrillation or flutter in Denmark, 1980–1999, International Journal of Cardiology. 103 (1) (2005) 78–84. [12] K.H. Humphries, C. Jackevicius, Y. Gong, L. Svensen, J. Cox, J.V. Tu, et al., Population rates of hospitalization for atrial fibrillation/flutter in Canada, Can J Cardiol 20 (9) (2004) 869–876. [13] J.P. Piccini, B.G. Hammill, M.F. Sinner, P.N. Jensen, A.F. Hernandez, S.R. Heckbert, et al., Incidence and prevalence of atrial fibrillation and associated mortality among Medicare beneficiaries: 1993–2007, Circ. Cardiovasc. Qual. Outcomes 5 (1) (2012) 85–93. [14] S. Colilla, A. Crow, W. Petkun, D.E. Singer, T. Simon, X. Liu, Estimates of current and future incidence and prevalence of atrial fibrillation in the U.S. adult population, Am. J. Cardiol. 112 (8) (2013) 1142–1147. [15] C.D.J. Holman, A.J. Bass, I.L. Rouse, M.S.T. Hiobbs, Population-based linkage of health records in Western Australia: development of a health services research linked database, Aust. NZ J. Publ. Health 23 (1999) 453–459. [16] M. Knuiman, T. Briffa, M. Divitini, D. Chew, J. Eikelboom, B. McQuillan, et al., A cohort study examination of established and emerging risk factors for atrial fibrillation: the Busselton Health Study, Eur. J. Epidemiol. 29 (3) (2014) 181–190. [17] J.M. Katzenellenbogen, T.H.K. Teng, D. Lopez, J. Hung, M.W. Knuiman, F.M. Sanfilippo, et al., Initial hospitalisation for atrial fibrillation in Aboriginal and nonAboriginal populations in Western Australia, Heart 101 (9) (2015) 712–719. [18] G.Y.H. Lip, R. Nieuwlaat, R. Pisters, D.A. Lane, H.J.G.M. Crijns, Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation, Chest 137 (2) (2010) 263–272. [19] Y. Miyasaka, M.E. Barnes, B.J. Gersh, S.S. Cha, K.R. Bailey, J.B. Seward, et al., Changing trends of hospital utilization in patients after their first episode of atrial fibrillation, Am. J. Cardiol. 102 (5) (2008) 568–572. [20] M.P. Turakhia, M.D. Solomon, M. Jhaveri, P. Davis, M.R. Eber, R. Conrad, et al., Burden, timing, and relationship of cardiovascular hospitalization to mortality among Medicare beneficiaries with newly diagnosed atrial fibrillation, Am. Heart J. 166 (3) (2013) 573–580. [21] C.X. Wong, A.G. Brooks, D.P. Leong, K.C. Roberts-Thomson, P. Sanders, The increasing burden of atrial fibrillation compared with heart failure and myocardial infarction: a 15-year study of all hospitalizations in Australia, Arch. Intern. Med. 172 (2012) 739–741. [22] C.X. Wong, A.G. Brooks, D.H. Lau, D.P. Leong, M.T. Sun, T. Sullivan, et al., Factors associated with the epidemic of hospitalizations due to atrial fibrillation, Am. J. Cardiol. 110 (10) (2012) 1496–1499. [23] L.G. Olsson, K. Swedberg, G. Lappas, S. Stewart, A. Rosengren, Trends in mortality after first hospitalization with atrial fibrillation diagnosis in Sweden 1987 to 2006, Int. J Cardiol. 170 (1) (2014) 75–80. [24] L.J. Nedkoff, T.G. Briffa, D.B. Preen, F.M. Sanfilippo, J. Hung, S.C. Ridout, et al., Age- and sex-specific trends in the incidence of hospitalized acute coronary syndromes in Western Australia, Circulation: Cardiovascular Quality and Outcomes. 4 (5) (2011) 557–564. [25] T.-H.K. Teng, J. Finn, M. Hobbs, J. Hung, Heart failure: incidence, case-fatality and hospitalization rates in Western Australia between 1990 and 2005, Circ Heart Fail 3 (2010) 236–243. [26] N. Wanahita, F.H. Messerli, S. Bangalore, A.S. Gami, V.K. Somers, J.S. Steinberg, Atrial fibrillation and obesity—results of a meta-analysis, Am. Heart J. 155 (2) (2008) 310–315. [27] Y. Miyasaka, M.E. Barnes, K.R. Bailey, S.S. Cha, B.J. Gersh, J.B. Seward, et al., Mortality trends in patients diagnosed with first atrial fibrillation: a 21-year communitybased study, J. Am. Coll. Cardiol. 49 (9) (2007) 986–992. [28] S. Stewart, K. MacIntyre, J.W.T. Chalmers, J. Boyd, A. Finlayson, A. Redpath, et al., Trends in case-fatality in 22 968 patients admitted for the first time with atrial fibrillation in Scotland, 1986–1995, Int. J. Cardiol. 82 (3) (2002) 229–236. [29] T. Briffa, S. Hickling, M. Knuiman, M. Hobbs, J. Hung, F.M. Sanfilippo, et al., Long term survival after evidence based treatment of acute myocardial infarction and revascularisation: follow-up of population based Perth MONICA cohort, 1984– 2005, BMJ 338 (jan26_2) (2009) (b36-.). [30] T.-H.K. Teng, J. Hung, M. Knuiman, S. Stewart, L. Arnolda, I. Jacobs, et al., Trends in long-term cardiovascular mortality and morbidity in men and women with heart failure of ischemic versus non-ischemic aetiology in Western Australia between 1990 and 2005, Int J Cardiol 158 (3) (2012) 405–410. [31] R.G. Hart, O. Benavente, R. McBride, L.A. Pearce, Antithrombotic therapy to prevent stroke in patients with atrial fibrillation: a meta-analysis, Ann. Intern. Med. 131 (7) (1999) 492–501. [32] Piccini JP, Fraulo ES, Ansell JE, Fonarow GC, Gersh BJ, Go AS, et al. Outcomes registry for better informed treatment of atrial fibrillation: rationale and design of ORBIT-AF. Am. Heart J. 162(4):(606–12.e1). [33] A.K. Kakkar, I. Mueller, J.-P. Bassand, D.A. Fitzmaurice, S.Z. Goldhaber, S. Goto, et al., International longitudinal registry of patients with atrial fibrillation at risk of stroke: Global Anticoagulant Registry in the FIELD (GARFIELD), Am. Heart J. 163 (1) (2012) (13–9.e1). [34] A. Clark, D.B. Preen, J.Q. Ng, J.B. Semmens, C.D.J. Holman, Is Western Australia representative of other Australian States and Territories in terms of key sociodemographic and health economic indicators? Aust. Health Rev. 34 (2) (2010) 210–215.