Article Text
Abstract
Objectives To examine temporal trends in the incidence and recurrence of hospitalised coronary heart disease (CHD), cerebrovascular disease (CeVD) and peripheral arterial disease (PAD) separately and combined, and by the history of all forms of atherothrombotic disease (ATD).
Design Population-based longitudinal data linkage study.
Setting Western Australia.
Participants All patients aged 35–84 years hospitalised in Western Australia for CHD, CeVD or PAD from 2000 to 2007.
Main Outcome Measures Age-standardised incidence and recurrence rates of CHD, CeVD and PAD stratified by ATD history, sex and age.
Results 107 576 events (65.9% men) were identified; 70% of all admissions were for CHD. In patients without a history of any ATD, incidence rates declined significantly in all groups, although the reduction in incident CHD in women was marginal (−0.7%/year, 95% CI −1.5 to +0.1%). The largest annual reductions in incidence rates were for PAD (men, −6.4%/year, 95% CI −7.7 to −5.0%; women, −5.4%/year, 95% CI −7.2 to −3.6%) and CeVD in women (−4.0%/year, 95% CI −5.0 to −3.0%). Falls in overall recurrence rates were greatest for CeVD (men, −3.2%/year, 95% CI −4.7 to −1.6%; women −4.6%/year, 95% CI −6.4 to −2.7%). Trends across all categories of polyvascular ATD were generally downward, although not all changes were statistically significant.
Conclusion The incidence and recurrence rates of hospitalised ATD have decreased over time, including in patients with disease involving multiple vascular territories. This implies that primary and secondary prevention strategies have been broadly effective. However, high absolute rates of recurrence and limited reduction in 35–54-year-old individuals highlight patient groups to target to reduce further the burden of ATD.
- Acute coronary syndrome
- cerebrovascular disorders
- coronary artery disease
- coronary heart disease
- epidemiology
- myocardial ischaemia and infarction (IHD)
- peripheral artery disease
- public health
- quality of care and outcomes
- risk factors
Statistics from Altmetric.com
- Acute coronary syndrome
- cerebrovascular disorders
- coronary artery disease
- coronary heart disease
- epidemiology
- myocardial ischaemia and infarction (IHD)
- peripheral artery disease
- public health
- quality of care and outcomes
- risk factors
Introduction
Coronary, cerebral and peripheral atherothrombotic disease (ATD) combined dominate morbidity and mortality worldwide.1 In middle and high income countries including Australia, this is exemplified by ATD leading hospitalisation statistics.2 There is a shared atherosclerotic pathology and heightened risk of atherothrombotic events across the vascular territories. Examining trends in the incidence and recurrence of all hospitalised ATD is therefore important for determining the overall burden of ATD, for public health planning, and for assessing the effectiveness of shared cardiovascular disease treatment and risk factor interventions across the vascular territories.
Contemporary studies have selectively investigated event rates and outcomes across vascular territories3 ,4 yet a paucity of data exists for temporal trends across the continuum of ATD measured concomitantly in the same population. Community and population-based studies demonstrate a declining incidence of coronary heart disease (CHD),5 ,6 cerebrovascular disease (CeVD)7 ,8 and peripheral arterial disease (PAD) separately.9 ,10 However, the risk of a new event in patients with established ATD remains high, particularly in those with polyvascular disease.11 ,12 Common risk factors and prevention approaches mean that our understanding of the effectiveness of both primary and secondary prevention is improved by describing trends across the whole spectrum of ATD. In addition, this would confirm whether any improvements extend to the highest-risk group of patients with polyvascular disease. Reporting trends in the incidence and recurrence of ATD by previous history will therefore benchmark the collective effectiveness of preventive strategies. The population-based data linkage system in Western Australia (WA) provides a unique opportunity for undertaking this study.
The aims of this study were to describe trends in incidence and recurrence rates of hospitalised CHD, CeVD and PAD in WA from 2000 to 2007, and to determine whether trends differed according to the history of ATD in other vascular territories.
Methods
Setting and data source
The population of WA in 2010 was 2.3 million, with 75% residing in the capital city, Perth.13 WA is representative of the Australian population in major sociodemographic and health economic indicators.14 Data for all hospitalisations and deaths in WA are contained in the Hospital Morbidity Data Collection (HMDC) and Mortality Register, respectively, both of which are regularly audited.15 All hospitalisations and death records for an individual are linked within the WA Data Linkage System by probabilistic matching, which has greater than 99% accuracy.15 The dataset for this study contained all admissions and/or deaths for cardiovascular disease in patients aged 35–84 years from 1985 to 2007. The upper age limit was imposed due to lack of validation of ATD in the very elderly. Events were identified using the prevailing International Classification of Diseases (ICD) version and relevant modifications (ICD-9 from 1979, and ICD-10 from 1 July 1999). Approval for this study was obtained from the ethics committees of the University of Western Australia and the WA Department of Health.
Definition of atherothrombotic events
We identified hospital admissions with a principal diagnosis of CHD, CeVD or PAD, with ICD codes selected to ascertain events of atherothrombotic origin. CHD was defined as myocardial infarction (MI), unstable angina, stable angina or other ischaemic heart disease (ICD-9 410–414; ICD-10 I20–I25); CeVD as cerebral infarction, transient ischaemic attack, precerebral or cerebral artery disease without infarction, unspecified stroke or intracerebral haemorrhage (431, 433–436, 438; I61, I63, I64, I66, I69, G45); and PAD as atherosclerosis of the aorta, renal arteries or arteries of the extremities, unspecified peripheral vascular disease, Buergers disease or stricture of arteries (440, 443.1, 443.9, 447.1; I70, I73.1, I73.9, I77.1). Previous validation of HMDC coding showed a positive predictive value for MI16 and coronary death17 of 83% and over 90%, respectively, and positive predictive value for stroke and other CeVD of 85% and 69%, respectively.17 Review of coding of PAD in surgical admissions showed a tendency towards overestimation, which was insufficient to influence PAD trends.10 Interhospital transfers were identified when there was 1 day or less between hospitalisations, and counted as part of the same admission. A readmission for CHD within 28 days of an index CHD admission was considered part of the same event to reduce overcounting of recurrent events due to readmissions for revascularisation and diagnostic procedures17; for CeVD and PAD, readmissions within 1 day of an index hospitalisation were deemed part of the same event.
CHD, CeVD and PAD events with an admission date between 1 January 2000 and 31 December 2007 were included in the study, and each was classified as incident or recurrent using a 15-year lookback period, shown to differentiate first-time events.18 An event was deemed incident if there was no hospitalisation for the same type of event during the lookback period, and otherwise was considered recurrent. A 15-year lookback period was also used to identify history of CHD, CeVD or PAD for each incident and recurrent event, and events were categorised into four groups according to the number and type of other vascular territories involved (see tables 2 and 3).
ATD-related deaths were classified using the ICD codes described above for incident and recurrent events. A sensitivity analysis was undertaken to analyse the use of codes not included in our ATD definitions. For CHD, deaths coded to cardiac arrest, respiratory arrest or unspecified cardiovascular disease were also incorporated in the analyses (CHD + I46, R09.2, I51.6); and the broader coding groups for all CeVD (I60–I69, G45) and peripheral vascular disease (I70–I79) were also analysed.
The following comorbidities were identified from any diagnosis field of the HMDC using a 15-year hospitalisation history for each event: diabetes (250, E10–E14); hypertension (401–405; I10–I15); chronic kidney disease (based on the Australian Institute of Health and Welfare definition)2; atrial fibrillation (427.3; I48); heart failure (428; I50); chronic obstructive pulmonary disease (490–496; J40–J47); and cancer (140–239; C00–D48). Codes are consistent with those used in CHD patients in an Australian study, in which agreement between administrative data and medical records was greater than 88% for these comorbidities.19 Invasive diagnostic and interventional procedures were identified from any procedure field, and included coronary angiography, percutaneous coronary intervention, or coronary artery bypass surgery for CHD; cerebral angiography, embolectomy, thrombectomy or endarterectomy of the carotid or intracranial arteries for CeVD; and peripheral angiography, revascularisation or repair of peripheral or abdominal arteries, and lower limb amputation or revision procedure for PAD. An acute event was an emergency hospitalisation for acute coronary syndrome (for CHD), cerebral infarction or transient ischaemic attack (for CeVD), and atherosclerosis of the lower extremities (for PAD). There were fewer than 0.001% missing values for any of the variables used in this study.
Statistical analysis
Men and women were analysed separately. Characteristics of the study population are presented as mean (SD) for continuous variables and frequencies (%) for categorical variables, and are calculated separately for incident and recurrent events. Incidence and recurrence rates by calendar year were calculated for each ATD type overall and by ATD history group. Counts of incident or recurrent events in each group were the numerator, while the denominator was the population at risk for each type of event in each study year, including by ATD history. The denominator for incidence rates was calculated using the whole WA population20 minus the prevalent population for each ATD type and history group. The method for identifying the prevalent ATD population has been described previously.21 The denominator for recurrence rates was the prevalent population in WA for each ATD type and history group.21 Mortality rate numerators were the counts of deaths for each ATD type separately, and the denominators were the WA population in each calendar year.20 Annual age-standardised and age-specific rates were calculated by the direct method using 5-year age groups, with the WA population in 2007 as the standard population for incidence and ATD mortality rates,20 and the prevalent ATD population in WA in 2007 as the standard population for recurrence rates.21 Age-standardised comorbidity prevalence was calculated for each calendar year for all incident ATD with no ATD history, and for all ATD cases combined, using the age distribution of all ATD cases in 2007 as the standard population.
Trends in age-adjusted rates by calendar year were estimated using Poisson log-linear regression models that included 5-year age group and calendar year (continuous). Model fit was tested and no overdispersion was found. Annual changes in rates were calculated using the exponential of the beta-coefficient for calendar year, and are presented as estimated annual percentage changes. Interactions for ATD type and calendar year were tested to determine homogeneity of trends across ATD history groups. Age-adjusted trends in the prevalence of selected comorbidities were estimated using logistic regression models that included 5-year age group and calendar year. Statistical significance was set at p<0.05 for all analyses, which were undertaken using SAS statistical software (V.9.2).
Results
There was a total of 70 844 atherothrombotic events in men and 36 732 in women aged 35–84 years between 2000 and 2007 (table 1). Approximately half of these events were classified as incident for CHD and PAD, whereas 75% of CeVD cases were incident. The mean age of patients hospitalised with incident CHD was 62.2 and 66.5 years in men and women, respectively; this was younger than for CeVD (68.5 and 70.9 years, respectively) and PAD (69.0 and 71.5 years, respectively). A similar pattern was seen for recurrent cases (table 1). The percentage of cases occurring among the oldest age group (70–84 years) varied from 30% for incident CHD in men, to 73% for recurrent CeVD and PAD in women. The most common recorded comorbidity was hypertension. Diabetes, chronic kidney disease and atrial fibrillation were also prevalent in incident and recurrent cases (table 1). PAD patients were less likely to be admitted acutely and more likely to undergo angiography and/or invasive intervention than CHD or CeVD patients (table 1).
CHD dominated incident (66.3% in men, 59.3% in women) and recurrent hospitalisations (79.4% in men, 75.0% in women) (tables 2 and 3). Only 7% of incident CHD events had previous ATD recorded, compared with incident CeVD (26% men, 18% women) and incident PAD (39% men, 29% women). Recurrent cases were more likely than incident cases to have a history of ATD in other vascular territories (table 3).
Trends in ATD incidence rates
Figure 1 shows the trends in overall age-standardised ATD incidence rates and table 2 the estimated annual change in age-adjusted incidence rates overall and by ATD history. Statistically significant declining overall trends in CHD, CeVD and PAD incidence were observed in men and women, with annual rates of decline greatest for PAD in men (−6.4%/year, 95% CI −7.7 to −5.0%) and women (−5.4%/year, 95% CI −7.2 to −3.6). A marked annual decline of 4.0%/year (95% CI −5.0 to 3.0) was also seen in women with CeVD. There were reductions in age-standardised incidence rates in all groups with no history of any ATD, although the trend was not significant in women with incident CHD (−0.7%/year, 95% CI −1.5 to +0.1%) (table 2). The rates of decline in incidence were generally greater in those groups with a history of other ATD but not all were statistically significant. There was increasing incidence in women aged 35–54 years in each of the vascular territories, and for men in this age group with CeVD, although trends were only significant in women with CHD (table 4).
Trends in ATD recurrence rates
Figure 2 shows the trends in overall age-standardised ATD recurrence rates and table 3 the estimated annual decline in age-adjusted recurrence rates overall and by ATD history. The annual rates of decline ranged from 1.6%/year (95% CI −3.6 to +0.05%) in women with recurrent PAD, to 4.6%/year (95% CI −6.4 to −2.7%) in women with recurrent CeVD. For recurrent PAD, the rate of decline was significantly greater in those with a history of CHD and CeVD compared with those without other ATD. Greater declines were also seen in men with recurrent CeVD with CHD and PAD history, and with previous CHD only, compared with no other ATD history. Trends were generally downward across all age groups, with the exception being men with CeVD and women with PAD in the 35–54-year age group, although these upward trends were not statistically significant (table 4).
Trends in ATD mortality rates
There were significant declines in age-standardised mortality rates for CHD and CeVD (figure 3). Age-standardised mortality rates for PAD were low, and there was no significant change in rates (men −2.1%/year, 95% CI −8.6 to +4.8%; women −4.3%/year, 95% CI −11.8 to +3.8%). Using a broader definition of CHD and CeVD deaths produced no significant effect on trends in CHD and CeVD mortality (data not shown). The broader definition for PAD deaths showed that mortality rates remained low in this group, ranging from 21.7/100 000 to 14.5/100 000 person-years in men, and 13.3/100 000 to 7.0/100 000 person-years in women. Trends persisted downwards in men and women, although became significant in men (−4.5%/year, 95% CI −7.8 to −1.2%) while the trend in women remained non-significant.
Trends in comorbidity prevalence
Trends and estimated annual percentage changes in age-standardised prevalence of hypertension, diabetes, atrial fibrillation and chronic kidney disease in incident ATD without ATD history, and in all ATD, are shown in figure 4. There were significant increases in the age-adjusted prevalence of hypertension in incident and in all ATD cases, and also significant increases in diabetes prevalence in incident and all ATD cases in men and all ATD cases in women.
Discussion
This state-wide Australian study of 107 576 atherothrombotic events is the first to report on temporal trends in incidence and recurrence concurrently across vascular territories and according to ATD history. Overall incidence and recurrence rates for hospitalised ATD declined between 2000 and 2007 in a whole population setting. Importantly, stratifying trends by ATD history showed that incidence rates also declined in those with no previous ATD of any type, although the reduction in women with incident CHD was limited. There was also limited reduction in incidence rates in the younger age group across all vascular territories, particularly in women. Our results also demonstrate that incidence and recurrence rates have generally fallen to a greater extent in those with polyvascular disease, suggesting the benefit of shared prevention treatments across multiple vascular territories. Although declining, recurrence rates for hospitalised ATD in our population remained many times higher than comparable incidence rates. This study also demonstrates that CHD is the major burden of hospitalised ATD (70% of cases).
Strengths and limitations
The use of high-quality linked administrative data with an extensive lookback period allowed analysis of rates by vascular territory and atherothrombotic history, and differentiation between first and recurrent events. We were also able to calculate events based on episodes of care for an individual due to person-based linkage of records in WA.15 Although there is the potential for the exclusion of true recurrent CHD events when imposing 28-day episodes, a sensitivity analysis showed that CHD patients were twice as likely to be readmitted within 28 days of an index admission as patients with CeVD or PAD, highlighting that admission practices are likely to drive this difference. Although out-of-hospital deaths are not included in our definition of incident and recurrent events, declines in these rates are unlikely to be due to a shift from non-fatal to fatal cases. This is because fatal incident CHD rates, including out-of-hospital deaths, have fallen in WA22 and the current study shows declining ATD mortality rates. This is consistent with concurrent falls in event rates and mortality rates for MI and stroke elsewhere.23–25 Changing thresholds for diagnostic tests and hospital admissions over time could impact on trends. Troponin testing has attenuated downward trends in MI incidence26; however, the inclusion of all CHD hospitalisations means that any apparent increase in MI counts is likely to be offset by a shift in diagnosis from other variants of CHD. There may be a greater propensity to treat less severe PAD in community or outpatient settings relative to comparable CHD or CeVD cases, which could underestimate the total relative burden of PAD. The decline in PAD rates is unlikely to be due to a shift of diagnostic testing to the outpatient setting, as preliminary analysis showed that there was a limited change in the proportion of PAD patients undergoing peripheral angiography as inpatients. Identification of events was based on clinical advice and previous validation studies locally.10 ,16 ,17 A limitation is the potential misclassification of diagnostic coding, which could contribute to over or underestimation of absolute event numbers, but is unlikely to have varied during the study period and consequently would be unlikely to bias trend estimates. Events were identified only from the principal diagnosis field as validity of coding is higher than for secondary diagnoses.17
Comparisons with other studies
The declines in overall incidence in our study are consistent with historical trends in WA for coronary events,27 stroke28 and PAD-related admissions,10 and with reductions seen elsewhere in community and hospital settings,7 ,25 ,29 ,30 including reductions in first-time hospitalisations for acute MI in Denmark.23 Temporal trend data on recurrence rates are limited, but recurrence rates have fallen concurrently with declines in MI incidence in Finland and Sweden6 ,31 and stroke incidence in Scotland.7 ,32 There are limited data on recurrence rates for hospitalised PAD, and on incidence and recurrence according to atherothrombotic history across vascular territories. The age-standardised trends in our study mask possible attenuation of declines in rates in 35–54-year–old individuals. Other population-based studies have reported limited reductions in MI and stroke rates in this age group7 ,24; however, of note in our study was the significant increase in CHD incidence in younger women, a trend also seen in this group for acute coronary syndromes in WA.33
Notably, CHD dominated incident and recurrent events in our population. This was similar to the REACH study,4 which investigated stable outpatients, but in contrast to the findings of the Oxford Vascular Study,3 in which stroke dominated and which included acute hospitalised and non-hospitalised cases with no upper age limit. The ratio of acute incident cerebrovascular to coronary events in Oxfordshire was higher than in our study population (1.19 vs 0.6). This difference was primarily driven by a higher ratio in women in the former study, and was also associated with a higher mean age. Although our upper age limit of 84 years to improve case validity would have contributed to this difference, there is also the possibility of differing population demographics and hospital admission thresholds, particularly in women. Despite these apparent variances, downward trends in incident stroke have also been reported in Oxfordshire.25
Implications of results
Trends in incidence when there is no history of any ATD are an important indicator of the effectiveness of primary prevention in the community. Therefore, declines in incidence that occurred in men and women without previous ATD implied positive effects of primary prevention in the WA population. Also, the rates of decline in incidence were observed to be generally greater in groups with a history of other ATD (table 2), although in PAD, incidence declined markedly irrespective of other atherothrombotic history, suggesting a shared benefit of cardiovascular treatments and prevention measures across vascular territories and possibly a greater impact of primary prevention in this group compared with CHD or CeVD. In Australia, rapid uptake of antihypertensive and cholesterol-lowering medication and long-term anti-smoking campaigns are likely contributors to these declines.34 A greater uptake of hypertension treatment in women compared with men has been shown in the UK,35 a pattern which if duplicated in WA could account for the greater falls in CeVD incidence seen in women in our study. Although hypertension prevalence increased in incident ATD cases in our study, this may represent improved targeting and admission practices for high-risk ATD patients.
The flat or increasing trends in incidence in 35–54-year–old individuals in our study, particularly in women, may indicate the effect of adverse risk factor trends. National data show that men and women aged 30–34 years in 1980 gained on average 8 kg and 12 kg, respectively, over the subsequent 20-year period.34 The prevalence of diabetes in Australia has increased during the same period and is higher than in many other developed countries.36 The concurrent attenuation of downward trends in CHD mortality in younger people22 ,37–39 further supports the impact of unfavourable trends in these risk factors.
Declining recurrence rates for CHD, CeVD and PAD reflect the effectiveness of well-established secondary prevention. In particular, reductions in incidence and recurrence rates in those with polyvascular disease may indicate that evidence-based cardiovascular treatments and prevention measures aimed at secondary prevention in one vascular territory (predominantly CHD) may have reduced the progression of atherothrombosis to other vascular territories. This is supported by large falls in recurrence rates in the highest risk groups, particularly recurrent PAD with concomitant CHD and CeVD in men (−9.5%/year) and women (−12.0%/year).
Importantly, this study identifies targets for further reduction of the ATD burden. The proportion of incident admissions with no previous ATD history is high (86%). Many asymptomatic persons at high cardiovascular risk are not targeted for prevention nor have their risk factors optimally controlled,40 therefore targeting of this group in the primary care setting would decrease the atherothrombotic burden. Despite falling recurrence rates, the current study highlights that absolute rates of recurrence are 15–20 times higher for CHD and CeVD and over 100 times higher for PAD than equivalent incidence rates. This supports evidence that the risk of recurrence and rehospitalisation for cardiovascular events remains high,4 and reinforces the need for optimising the application of evidence-based treatments in patients with established disease.
Conclusions
This study has shown that incidence and recurrence rates of hospitalised atherothrombotic CHD, CeVD and PAD declined between 2000 and 2007 in a state-wide Australian setting. Reductions were generally greater in those with polyvascular disease for both incident and recurrent cases, implying that primary and secondary prevention strategies have been broadly effective in our state. However, there was evidence of attenuation of downward trends in 35–54-year-old individuals, particularly in women, and increasing prevalence of hypertension and diabetes in hospitalised ATD patients. Population-based monitoring is therefore an essential means of identifying target groups such as these to reduce further the burden of ATD.
What is already known on the subject
Incidence rates of CHD, CeVD and PAD have declined in developed nations over four decades.
There is limited evidence for temporal trends on polyvascular or ATD recurrence.
What this study adds
Age-standardised incidence and recurrence rates of hospitalised CHD, CeVD and PAD declined between 2000 and 2007 in 35–84-year-olds in WA.
Age-standardised incidence rates for CHD, CeVD and PAD declined in those without previous ATD, but marginally so in women with CHD, and declines were limited in those aged 35–54 years.
Rates of decline were generally greater in those with polyvascular disease, particularly for recurrent cases, suggesting positive effects of secondary prevention.
Acknowledgments
The authors acknowledge the Data Linkage Branch (WA Department of Health) for extraction and provision of the data used in this study.
References
Footnotes
Funding LN is funded by the National Health and Medical Research Council of Australia (NHMRC); the NHMRC funded the Real and Changing Atherothrombotic Disease Burden and Secondary Prevention project (#572558); the funding body played no role in the study design, analysis or interpretation of data, writing of the report, or of the decision to submit the article for publication.
Competing interests None.
Ethics approval This study was approved by the ethics committees of the University of Western Australia (RA/4/1/1491) and the WA Department of Health (#2009/18).
Provenance and peer review Not commissioned; externally peer reviewed.