Correspondence to: Martin Tobias, Ministry of Health, PO Box 5013, Wellington, New Zealand. Fax: (+64) 44962191; e-mail: email@example.com
Objective: To estimate coronary heart disease (CHD) incidence, prevalence, survival, case fatality and mortality for Māori, in order to support service planning and resource allocation.
Methods: Incidence was defined as first occurrence of a major coronary event, i.e. the sum of first CHD hospital admissions and out-of-hospital CHD deaths in people without a hospital admission for CHD in the preceding five years. Data for the years 2000-02 were sourced from the New Zealand Health Information Service and record linkage was carried out using a unique national identifier, the national health index.
Results: Compared to the non-Māori population, Māori had both elevated CHD incidence and higher case fatality. Median age at onset of CHD was younger for Māori, reflecting both higher age specific risks and younger population age structure. The lifetable risk of CHD for Māori was estimated at 37% (males) and 34% (females), only moderately higher than the corresponding estimates for the non-Māori population, despite higher Māori CHD incidence. This reflects the offsetting effect of the higher ‘other cause’ mortality experienced by Māori. Median duration of survival with CHD was similar to that of the non-Māori population for Māori males but longer for Māori females, which is most likely related to the earlier age of onset.
Conclusions: This study has generated consistent estimates of CHD incidence, prevalence, survival, case fatality and mortality for Māori in 2000-02. The inequality identified in CHD incidence calls for a renewed effort in primary prevention. The inequality in CHD case fatality calls for improvement in access for Māori to secondary care services.
Inequality in mortality from coronary heart disease (CHD) between Māori and non-Māori has long been recognised in New Zealand.1 This inequality has shown to be both large and increasing, at least on a relative scale.2 Less is known about inequality in CHD incidence. Incidence has typically been indirectly assessed by the rate of hospitalisation for major coronary events, so excluding out of hospital deaths. Such studies have often failed to demonstrate higher rates for Māori.3
We have previously developed a method for producing internally consistent estimates of CHD incidence, prevalence, survival, case fatality and mortality using routinely collected national data (via record linkage and multistate lifetable modelling), which we applied to the total New Zealand population.4 We now apply this methodology to the Māori and non-Māori populations.
We hypothesise that ethnic inequalities in CHD mortality are not caused solely by higher Māori case fatality, but also by higher incidence.
Unit record public hospital inpatient admissions data and mortality data by Māori/non-Māori ethnicity for 1996-2002 were provided by the New Zealand Health Information Service (NZHIS). The 2001 Census Māori and non-Māori populations, provided by Statistics New Zealand (SNZ), were used as the denominators. All cause mortality rates for Māori and non-Māori (by five year age group and sex) for 2000-02 were also provided by SNZ. For both numerators and denominators, the total response concept of ethnicity was used. That is, individuals self-identifying as Māori were counted as Māori, irrespective of whether or not they also identified with another ethnic group(s). All remaining individuals were classified as non-Māori.
Incidence was defined as first occurrence of a major coronary event, i.e. the sum of first CHD hospital admissions and out-of-hospital CHD deaths in people without a prior CHD admission. To identify first hospitalisations, admissions (by Māori and non-Māori ethnicity) in 2000-02 coded to ICD-10 I20-I22, I24, and I25 in the primary diagnosis field were linked by the national health index (NHI) number, a unique personal identifier used in the New Zealand health sector, to all admissions over the preceding five years. They were counted as first events only if no earlier CHD admissions could be identified in either primary or secondary diagnosis fields. Care was taken to avoid double counting due to interhospital transfers. Out-of-hospital deaths were identified similarly (i.e. by linking deaths to hospital admissions for CHD over the preceding five years through the NHI). The look-back time of five years is arbitrary, but was selected as a compromise between sensitivity (detecting all prior hospital admissions) and specificity (failure of linkage due to the individual having duplicate NHIs, which was more common in earlier years).
Deaths of Māori and non-Māori registered in 2000-02 with codes ICD-10 I20-I25 were extracted (this included both in-hospital and out-of-hospital deaths). These codes may, however, miss CHD deaths coded to sudden death, heart failure, cardiovascular disease not otherwise specified and type 2 diabetes mellitus. All deaths so coded over the study period were assessed using multiple cause of death coding, and were also linked by NHI to hospital discharges over the preceding four or five years for mention of CHD (in either primary or secondary diagnosis fields). This process yielded approximately an additional 10% of CHD deaths.
Multi-state lifetable modelling
Multi-state lifetables for Māori and non-Māori in 2000-02 (male and female) were constructed by conventional demographic methods.7 The mathematics of the multi-state lifetable and its use for descriptive epidemiology have been summarised elsewhere.8 In brief, construction of the lifetables required age and sex specific all-cause mortality rates for Māori and non-Māori (derived from data provided by Statistics New Zealand) and similar estimates for CHD incidence and mortality (derived as explained above). In addition, remission rates were assumed to be zero by definition (i.e. a person with CHD cannot transition back to the non-diseased state).
Estimating prevalence, survival and case fatality
Given these inputs – incidence, mortality and (zero) remission – multi-state lifetables were constructed for Māori and non-Māori males and females, and a wide range of internally consistent outputs extracted, including incidence, lifetable risk, age of onset, prevalence, survival, duration, case fatality and mortality. Note that the output incidence and mortality rates may differ from those inputted, as the model forces these rates to be consistent with each other and the other lifetable outputs (e.g. prevalence). Further note that the case fatality outputted by the model estimates the proportion of prevalent cases at each age who are dead from CHD one year later. This differs from the conventional definition (which is based on incident cases), so estimates of case fatality reported here are not directly comparable with those reported elsewhere.
The burdens of CHD on Māori and non-Māori in 2000-02 were then estimated by applying the respective output rates from the multistate lifetable models to the 2001 Census Māori and non-Māori populations.
Estimated CHD incidence, case fatality, prevalence and mortality rates for Māori and non-Māori populations in 2000-02 are summarised by age and sex in Figure 1–4. The patterns are similar to those shown previously for the whole New Zealand population, albeit at a higher level for Māori (all variables) and with the exponential rise beginning at a younger age for Māori. As hypothesised, all age specific statistics – incidence rates, case fatality, prevalence and mortality rates – are higher for Māori than for non-Māori (both sexes). Moreover, the absolute ethnic differences in these measures increase with increasing age. Of note, case fatality is substantially higher in Māori males than non-Māori males, yet Māori females had only moderately higher case fatality than non-Māori females (or non-Māori males). So unlike non-Māori, Māori show a gender inequality in case fatality (higher in males).
Translating rates into counts (Table 1), we estimate that approximately 900 Māori experienced a first major coronary event in 2001. The median age at which this first event occurred was estimated to be approximately 56 years for Māori males and 59 years for females. This compares with 67 years for non-Māori males and 75 years for non-Māori females. The estimated lifetable risk of CHD for Māori in the early 2000s is 37% for males and 34% for females. This is moderately higher than the corresponding estimate for non-Māori males (34%) and females (27%). (Lifetable risk is the probability that an individual will have a major CHD event in his or her lifetime, taking into account risks of mortality from all causes).
Table 1. Burden of CHD in Māori and Non-Māori population, 2000-02.
Incidence count (n)
Prevalence count (n)
Mortality count (n)
Lifetable risk (%)
Median age of onset (yrs)
Median survival duration (yrs)
Including both in and out of hospital deaths, we estimate that approximately 460 deaths were attributable to CHD among Māori in 2001, 18.2% of all deaths in this ethnic group in that year. This is 8% less than the 500 deaths coded to CHD (for Māori) in the NZHIS mortality collection (annual average for 2000-02), a similar degree of under estimation to that found for non-Māori.
Approximately 6,250 Māori are estimated to have been living with CHD (as defined) in 2001. Median survival duration (time from first major CHD event to death from any cause) was estimated to be 9.7 years for Māori males and 10.9 years for females. This is similar to the corresponding estimate for non-Māori males (10.2 years), but longer than the corresponding estimate for non-Māori females (7.6 years).
This study has provided internally consistent estimates for CHD incidence, prevalence, survival, case fatality and mortality for Māori and non-Māori in 2001. Furthermore, these results are fully comparable to those reported earlier for the whole New Zealand population in 2002 (allowing for the one year difference in time period).4 Note, however, that our model captures only major coronary events and their consequences. The large numbers of people with subclinical CHD or angina only are excluded.
As hypothesised, Māori had elevated CHD incidence, not just elevated case fatality, relative to non-Māori. The higher incidence reflects both higher rates of hospitalisation for first major coronary event and higher rates of out of hospital cardiac death (without prior hospitalisation for major coronary event). For example, the age-standardised out-of-hospital CHD mortality rate (including only deaths without any prior hospital admission) in 2000-02 was 90.6 per 100,000 for Māori, approximately 2.4 times higher than the corresponding rate for non-Māori (38.2 per 100,000).The first major coronary event also occurs at a younger age among Māori, with median age of onset in the late 50s for both sexes, compared to late 60s and mid 70s for non-Māori males and females respectively. This reflects both increased age-specific risks and younger population age structure for Māori. The smaller gender difference in age of CHD onset observed among Māori compared to non-Māori may be explained in part by differences in risk factor distributions by gender between the two populations. For example, the prevalence of smoking is significantly higher in Māori females than males but slightly lower in non-Māori females than males.11 Moreover, the significantly shorter life expectancy of Māori compared to non-Māori truncates the potential age distribution of CHD onsets for Māori.
Māori experience only moderately higher lifetable risks of CHD than non-Māori, despite higher CHD incidence. This counter-intuitive finding reflects the impact of the higher ‘other cause’ mortality rates experienced by Māori (at all ages). Even so, more than one-third of Māori will experience one or more major coronary events over their lifetime. Māori who develop CHD will spend on average about a decade of their lives with prevalent CHD. This is similar to the corresponding estimate for non-Māori males but longer for females. In males, the similar time with prevalent disease reflects the higher age-specific case-fatality in Māori. In contrast, the longer period that Māori females live with prevalent disease most likely reflects their younger median age of CHD onset relative to non-Māori females, along with their moderately higher age-specific case-fatality. Reasons for the gender gap in CHD case fatality among Māori (which is not mirrored in the non-Māori population) are unclear and represent an important topic for further research.
Our study has some limitations. Conceptually, an important limitation is the definition of incidence as the occurrence of the first major coronary event. This definition is restrictive and will lead to under estimation of incidence (and hence prevalence and mortality). Thus, Māori with diagnosed coronary disease who have not (yet) had a coded admission to hospital are excluded. Another conceptual limitation is the inability of the model to output a measure of inpatient case fatality, which would be more clinically relevant. The case fatality measure reported here differs from that conventionally defined and cannot be directly compared with estimates derived from clinical studies. The model is also silent on the health status of people with prevalent CHD – for example, it cannot distinguish between those with and those without heart failure.
Technical limitations include data quality issues, in particular use of an arbitrary ‘five year rule’ to identify first admissions and out-of-hospital deaths without prior admission; misclassification of ethnicity data; linkage bias introduced via duplicate NHI assignment; and inability to fully adjust cause of death data for miscoding and misclassification. Relatively sparse incidence and mortality data for Māori in very old age groups (75+ years) results in the rate estimates for these age groups being strongly influenced by data smoothing.
The lifetable method is also limited in that it is based on a steady state. The model generates estimates for prevalence in 2001 by assuming that the age specific incidence and mortality estimates for that year applied also to previous years. Given declining trends in CHD mortality1 and possibly incidence,9 we will have under estimated CHD prevalence. Our modelled estimate for CHD mortality among Māori is approximately 8% less than recorded in the NZHIS mortality collection for 2001. This may reflect under estimation by the model or inaccuracies in cause of death assignment or coding of the empirical mortality data. Also, being deterministic, the model is unable to output confidence intervals. Instead, sensitivity analysis can be used to quantify uncertainty, although we do not report results of such sensitivity analyses here. Finally, while the model itself is equally robust for both ethnic groups, the poorer quality of the data for Māori, and the greater impact of misclassification on the smaller ethnic group, mean that the epidemiological estimates for Māori are likely to be less robust than are those for non-Māori.
Despite these limitations, our results provide information that can be used to guide cardiovascular health service planning. Internally consistent CHD incidence, prevalence, survival, case fatality and mortality estimates will enable health service funders and purchasers to assess the level of met and unmet need for preventive, therapeutic and rehabilitative cardiology services, and will also contribute to the assessment of service access barriers for Māori.5,6 Indeed, our findings reinforce the ‘call to action’ on Māori cardiovascular health issued by a health sector working group in 2004,5 and echoed more recently in He Korowai Oranga, the Māori Health Strategy.6 The major inequality in incidence we have demonstrated calls for realignment of the distribution of resources to primary preventive services in order to better address the issue of ethnic cardiovascular inequalities and enable a greater focus of interventions on key risk factors such as cigarette smoking and obesity. Similarly, the equally significant inequality in CHD case fatality (demonstrated here and elsewhere) calls for system changes that ensure increased responsiveness of specialist care for Māori patients post-coronary event (including improved access to reperfusion surgery, secondary prevention and cardiac rehabilitation). The scope for elimination of inequalities (whether socio-economic or ethnic) in coronary disease, at least in an absolute sense, has been reviewed elsewhere.12
The methods outlined in this study can also be used to monitor health system responsiveness, by providing a useful benchmark against which to measure Māori/non-Māori cardiovascular inequalities. The multistate lifetable is a pure period indicator, so by constructing such lifetables for each period, a time series of the descriptive epidemiology of CHD by ethnicity will be built up. Such monitoring will be of increasing importance as the obesity and type 2 diabetes epidemics evolve, possibly reversing the long term downward trend in CHD mortality – with earlier and potentially greater impact on Māori.10,11
This paper is published with the approval of the Deputy Director-General (Health and Disability Systems Strategy Directorate), New Zealand Ministry of Health. However, opinions are the authors' own and do not necessarily reflect the Ministry's policy advice.