Article Text

Download PDFPDF
Original article
Discordant age and sex-specific trends in the incidence of a first coronary heart disease event in Western Australia from 1996 to 2007
  1. Tom Briffa1,
  2. Lee Nedkoff1,
  3. Anna Peeters2,
  4. Andrew Tonkin2,
  5. Joseph Hung3,
  6. Stephen C Ridout1,
  7. Matthew Knuiman1
  1. 1School of Population Health, University of Western Australia, Crawley, Western Australia, Australia
  2. 2Department of Epidemiology & Preventive Medicine, Monash University, Melbourne, Australia
  3. 3Sir Charles Gairdner Hospital Unit and University of Western Australia, Perth, Australia
  1. Correspondence to Professor Tom Briffa, M431, School of Population Health, University of Western Australia, 35 Stirling Hwy, Crawley WA 6009, Australia; tom.briffa{at}


Objective To determine age- and sex-specific population trends in fatal and non-fatal first coronary heart disease (CHD) events in Western Australia from 1996 to 2007.

Design Longitudinal retrospective population study.

Setting State-wide population.

Patients All residents aged 35–84 years during 1996–2007 who died or were hospitalised with a principal diagnosis of acute CHD.

Data sources Person-linked file of mortality and morbidity records.

Main outcome measures Age-standardised (35–84 years) and age-specific (35–54, 55–69, 70–84 years) rates by gender for a first CHD event were calculated with a 10-year lead-in period to define first events.

Results From 1996 to 2007 there were 36 631 first CHD events, including 8518 (23%) fatal cases in those aged 35–84 years. Overall, age-adjusted rates for fatal first CHD declined 5.3%/year in men (95% CI −6.1% to −4.6%) and 6.5%/year in women (95% CI −7.5% to −5.5%). However, age-specific fatal first CHD rates were neutral in both men aged 35–54 years (0.1%/year; 95% CI −1.8% to 2.1%) and women of the same age, (−1.6%/year; 95% CI −5.6% to 2.5%). Age-specific trends in non-fatal CHD rates reflected the same trends in fatal CHD events in men and women, with rates reportedly increasing in women aged 35–54 years (2.5%/year (95% CI 1.1% to 3.9%).

Conclusion The age-specific decline in fatal and non-fatal first CHD rates in older men and women was not observed in those aged 35–54 years. These novel findings provide evidence for a levelling in the CHD incidence rates in younger adults and puts renewed importance on primary prevention in this group.

  • Coronary heart disease
  • incidence
  • mortalityage-specific
  • primary prevention
  • epidemiology

Statistics from


Coronary heart disease (CHD) is a leading cause of death world wide and accounts for tens of million of events annually, many prematurely.1 Importantly, age-standardised CHD mortality rates have declined in the USA, most Western European Countries and Australia since the 1960s.2–4 Modelling for determinants of declining CHD mortality suggests similar contributions from advances in medical care and risk factor control.2 5 Recent studies have demonstrated discordant trends in age-specific CHD mortality rates, with declines persisting in 55+ age groups but mostly neutral in younger groups.6–8 This has led to concern that increases in obesity and related cardiovascular risk factors (such as diabetes) could increasingly outweigh the positive effects gained from declining rates of other risk factors.5 9

However, it is unknown whether the previously described lack of improvement in CHD mortality at younger ages is due to true lack of improvement in disease incidence, a lack of improvement in treatment of CHD, or simply no room for substantial further improvements in CHD mortality at these ages. Incidence of a first CHD event (fatal and non-fatal) is the best indicator of the effect of primary prevention on CHD rates. It is not possible to analyse trends in incidence from total mortality rates as recurrent CHD cases cannot be excluded. To take these observations further there is a need for age- and sex-stratified population-based studies capable of defining more accurately the incidence of a first CHD event in a population free of disease (CHD-free) using medical record linkage over a lengthy period before a first event.

The aim of this study was to examine trends in the rates of hospitalised and non-hospitalised cases of fatal first CHD in Western Australia from 1996 to 2007 using the linked hospital admission and death data for the population starting in 1986. We also examined whether rates of hospitalised non-fatal first CHD events was concordant with trends in fatal first CHD during the same period. The study was approved by the Human Research Ethics Committee of the University of Western Australia and the Confidentiality of Health Information Committee of the Western Australian Department of Health.


Population and subjects

The state of Western Australia (WA) has a population of 2.3 million in 2010. Since 1996, the adult population aged 35–84 years has increased by about a third from 810 126 to 1 070 126 in 2007. All deaths are registered centrally and major acute care of CHD is concentrated in eight major public and private hospitals in Perth, the capital city, which contains approximately 75% of the state's total population (currently 1.65 million) and is situated 1000 miles from the nearest major city in Australia. Provision of acute care for CHD to residents by hospitals outside the state is rare and migration in those aged >50 years is estimated at <1%.10 The study focuses on the population of WA aged 35–84 years over the period 1996–2007.

Data sources

Linked hospital admission and death data were extracted from the Hospital Morbidity Data System and the Death Registry from 1 January 1986 to 31 December 2007. The morbidity collection includes all hospital admissions from public and private hospitals in WA (up to 21 discharge diagnoses). All death registrations, state and nationally from 2002 (including cause), among residents of WA are linked to the Hospital Morbidity Data System.

Definition and calculation of fatal and non-fatal first CHD event rate

Hospital admissions and deaths from CHD were identified using the International Classification of Diseases (ICD) codes incorporating ICD-9 revision, ICD-9 clinical modification and ICD-10 Australian Modification.11 Fatal and non-fatal CHD events were identified using codes 410 for myocardial infarction, 410–414 for CHD in ICD-9 versions and I21/I22 and I21-I25 in ICD-10, respectively. Diagnoses of myocardial infarction using hospital morbidity data, as well as the causes of death from CHD in the death register have been validated previously.12

The CHD-free population for each calendar year (from 1996 to 2007) was the total population at 30 June in that year less the population with established CHD for that year. The latter population for each year was defined as those who were alive at 30 June who had had a hospital admission for CHD (in any diagnosis field) in the preceding 10 years.

Fatal CHD events for each year were defined as a death from CHD (without hospitalisation) or a hospital admission for CHD (principal discharge diagnosis) with death within 28 days of admission. Non-fatal CHD events were all other hospital admissions for CHD. First CHD events were defined as those occurring in individuals not having been hospitalised for CHD in the previous 10 years. Transfers between hospitals within 28 days of a first event were counted once only.

The fatal and non-fatal first CHD event rates in a calendar year were calculated as the number of fatal and non-fatal first CHD events, respectively, in that year divided by the CHD-free population for that year.

Statistical methods

Men and women were analysed separately. Age-standardised rates for each year were calculated by the direct method using the census total population of WA in 2001 as the standard. Trends in non-fatal and fatal first CHD event rates were analysed overall (35–84 years) and by age stratum (35–54, 55–69, 70–84 years). Estimated age-adjusted annual changes in first CHD event rates were obtained using Poisson regression (Proc Genmod in SAS13).


Study population and first CHD events

A general description of the study population and counts of fatal and non-fatal first CHD events are shown in Table 1. The annual CHD-free populations for men and women aged 35–84 years averaged approximately 429 000 and 451 000, respectively. During the study there were 36 631 first CHD events (34% in women), including 8518 fatal first CHD events. Deaths occurring within 28 days of a hospitalisation for CHD accounted for 34% of all fatal events. Fatal and non-fatal first CHD events occurred in 55% and 80% respectively of those aged <75 years (data not shown).

Table 1

Characteristics of the study population

Crude trends in non-fatal and fatal first CHD events

From 1996 to 2007, overall crude fatal and non-fatal first CHD event rates declined in men and women. In men, first fatal events fell from 151 per 100 000 (95% CI 139 to 164) to 93 (95% CI 85 to 102) in this period, and in women it declined from 86 (95% CI 77 to 95) to 41 (95% CI 36 to 47). Corresponding rates of non-fatal first events declined to a lesser extent in men over the same period from 405 per 100 000 (95% CI 385 to 426) to 360 (95% CI 344 to 377) and in women from 188 (95% CI 175 to 202) to 175 (95% CI 164 to 187).

Trends in age-standardised rates of fatal and non-fatal first CHD events

Figure 1 shows age-standardised trends by gender in rates of fatal and non-fatal first CHD events. Annually, rates of fatal first events fell 5.3% per year (95% CI −6.1% to −4.6%) in men and 6.5% per year (95% CI −7.5% to −5.5%) in women. Non-fatal first events fell 1.7% per year (95% CI −2.1% to −1.2%) in men and 1.3% per year (95% CI −1.9% to −0.7%) in women.

Figure 1

Age-standardised trends by gender in first fatal and non-fatal coronary heart disease.

Trends in age-specific rates of fatal and non-fatal first CHD events

Figure 2 shows age-specific trends by gender in rates of fatal and non-fatal first CHD events. In the 35–54 year age group, annual rates of fatal first events were essentially unchanged over the period with CIs including zero, whereas non-fatal event rates increased 2.5% per year (95% CI 1.1% to 3.9%) in women and decreased marginally by 0.9% per year (95% CI –1.7% to −0.2%) in men. In older age strata, age-specific fatal first events in men and women declined 6% or more per year (all p<0.001). Annual non-fatal first event rates also fell significantly in the older age strata (all p<0.001), although to a lesser extent than for fatal first events.

Figure 2

Age-specific trends by gender in first fatal and non-fatal coronary heart disease. (A) 35–54 years; (B) 55–69 years; (C) 70–84 years.


This population-based study included 36 631 individuals diagnosed with a first CHD event, of which 8518 (23%) were fatal. While age-standardised rates for both fatal and non-fatal first CHD declined from 1996 to 2007, age-specific fatal trends were significantly discordant. Fatal first CHD rates in both men and women aged 35–54 years showed no fall over the study period, whereas rates in 55–69 and 70–84 year age groups in both sexes fell 6% or more per annum. Concomitant changes in age-specific trends in non-fatal first CHD, including an increase in event rates in women aged 35–54 years strengthen the primary fatal CHD findings.

International comparisons

The unique aspect of this study is the population-based measurement of the actual incidence of fatal and non-fatal first CHD events in men and women aged 35–84 years having excluded prevalent CHD cases. The two other analogous population studies are the Monitoring Trends and Determinants in Cardiovascular Disease centres in New Zealand and Australia, and the Finnish acute myocardial infarction Registry. However, these studies were limited to people aged 35–64 years and trends before 1997.12 14 Recent studies in the USA, and UK have described age- and sex-specific CHD mortality rates in the total population but include people with established disease.7 8 15 These studies probably mask any deceleration in the decline of actual incident CHD mortality, as patients with prior CHD have 5-year risks of another event approximately 20% higher, in absolute terms than patients without prior CHD, after accounting for standard risk factors.16 Notwithstanding, all four studies confirm the ongoing steady decline in age-standardised CHD mortality rates, while variously contributing to our understanding of age- and sex-specific rates. Ford and Capewell7 reported falling average annual mortality rates in US adults aged 35–54 years decelerated in the 1990s, at 1.3% (95% CI −1.6% to −0.8%) in women and 2.3% (95% CI −2.6% to −2.1%) in men, and levelled off between 2000 and 2002 with wide CIs that included 0%. By contrast, in men and women aged ≥55 years, the expected annual average percentage decline in CHD mortality accelerated by a substantially greater magnitude over the same period. Both these findings are consistent with the results of our study, in particular the neutral CHD mortality rate trend in those aged 35–54 years, but now in a population not previously hospitalised with CHD. In socioeconomic disadvantaged Scottish men and women declining CHD mortality rates levelled out from 1994 in 35–44 year olds and from 2003 in 45–54 year olds.15 Data from England and Wales also show the rate of change in CHD mortality has reversed in the past 20 years, while indicating a future plateau and possible reversal of previous improvement in CHD mortality rates.8

There is indirect support for the current study's age-standardised and age-specific trends in non-fatal first CHD events. Previous population and community-based studies in Finland,14 17 the United Kingdom,18 19 US centres of Olmstead County20 21 and Northern California,22 examining trends in the incidence of first myocardial infarction (fatal and non-fatal) that overlap 1996 to 2007, suggest mostly a decline or stabilising (including younger women20) in the composite rates.

Possible explanations for discordant age-specific trends

Levelling of CHD mortality rates in younger age groups has been attributed to an unchanged incidence in the disease, underuse of evidence-based treatment for CHD, or that no room remains for substantial further improvements in CHD mortality at these ages. This study which reports a flattening of both fatal and non-fatal CHD incident rates in the younger initially CHD-free population suggests that preventable lifestyle factors are operative. We have previously reported that over 1980 to 1999, Perth residents aged 35–64 years had unfavourable risk factor trends.23 There was a marked and rapid increase in obesity as measured by body mass index and waist:hip ratio, in men and women during this period; but particularly in women aged 35–44 years.23 National data for the same period showed that the average measured weight increased by around 6.5 kg for men and 7.1 kg for women in each 5-year age group over 30 years.4 Disquietingly, recent reports show that younger cohorts are gaining weight more rapidly than those born before 1960,24 and that the future healthcare costs for CVD for these generations could be substantial.25 Concomitant with obesity trends, diabetes prevalence has more than doubled during the same period to 8.0% in men and 6.8% in women, and an additional 17.4% of men and 15.4% of women had impaired glucose regulation.4 There is also evidence that rates of smoking in young adults are not declining as fast as in older groups and that its prevalence is highest in 25–34 year olds.26 27 Other international studies in which substantial deceleration in CHD mortality has occurred in the past decade postulate similar changes in risk factors as a likely explanation for the slowing in decline of mortality rates.5 8 19 28 The latter modelling study28 contends that additional deaths owing to higher obesity and diabetes rates and a rise in systolic blood pressure in women would more than halve the events prevented from improvements in total blood cholesterol, smoking, physical activity and blood pressure in men.

The reasons for a smaller decline in non-fatal versus fatal CHD incident rates are uncertain but potential causes are a fall in acute myocardial infarction case fatality, less disease severity on presentation, the troponin biomarker effect and better acute treatment.20 22 27 Another likely reason is better primary prevention and the increased use of cardioprotective medicines (eg, antiplatelets, and lipid-lowering and anti-hypertensive agents) in those asymptomatic subjects at higher absolute CHD risk (eg, elderly) attending their primary care doctor.23 29


The point estimate for average annual percentage change in CHD mortality suggests neutrality in men aged 35–54 years and a decrease in women of the same age. However, the wide CI in women, encompassing zero simply means a flat line is also possible. The increase in the low non-fatal event rate in young women may also be a chance finding. Thus, a deceleration of the fatal rate approaching zero might simply reflect the natural low-point in rates in younger women. However, this could not explain the flattening in the rate in young men, in whom rates were considerably higher than in young women.

CHD event records are subject to misclassification and miscoding in administrative datasets. However, we have previously shown that trends of fatal (independent of autopsy) and non-fatal myocardial infarction based on record linkage are valid.12 Our methods will not have captured CHD which has been either undiagnosed or diagnosed and treated in primary care, or lost to follow-up with emigration overseas or linkage failure. However, we have previously shown that a negligible (<1%) proportion of patients with recognised non-fatal CHD in Perth are managed solely at home30 and WA has the advantage of a lower rate of out-of-state migration than the national average (2.0% vs 3.7% per annum in 2003). Further, an audit of the state record linkage system found that failures in the linkage of hospital records to death records are rare (0.11%).31


From 1996 to 2007, average annual age and sex-specific fatal and non-fatal first CHD rates were mostly neutral in 35–54 year olds, yet continued to decline in 55–84 year olds. These diverging rates coincide with adverse changes in the prevalence of obesity and diabetes since the 1980s and slower rates of smoking cessation in younger adults. Such unfavourable trends in key risk factors for CHD will do little to reverse this levelling of first CHD event rates. Effective public health policies that can achieve a net benefit in risk factor trends particularly in younger men and women are of the upmost priority if the gains in declining CHD mortality over four decades are to be preserved.


We thank the staff of the Data Linkage Branch, Western Australian Department of Health, for extraction of the Linked data and cardiologist Michael Nguyen for reviewing the penultimate version of this manuscript and for his helpful suggestions for improvement.



  • Funding This work was supported by University of Western Australia and the National Health and Medical Research Council.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Human Research Ethics Committee of the University of Western Australia and the Confidentiality of Health Information Committee of the Western Australian Department of Health.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.