David Lillis: Ethnicity and Public Health

Abstract

Up to now ethnicity has been used as a marker of public health in New Zealand, both as a means of identifying high-risk communities and proportioning resources for health and medical services and intervention. In 2024 the Coalition Government has reduced, though not eliminated, the use of ethnicity as a marker, so that public services, including health services, are prioritized on need rather than race. 

Evidence for disparities in health outcomes across populations is seen in many statistical studies. However, attribution of those disparities to systemic bias and racism is questionable. Most probably, other factors account for those disparities, including socioeconomics, poor housing, levels of education, age distributions, gender, sexuality, genetics, disability, rurality and lifestyle choices.  

 

Identifying and Assessing Health and Medical Needs

An interesting piece was published in E-Tangata on 6 October, entitled ‘An attack on science and good medical practice’ (Loring et al., 2024). The six authors are highly qualified medical professionals and five of them appear to have Māori ancestry. They assert that the Coalition Government has launched an attack on science and good medical practice by directing agencies to downplay ethnicity as a marker of need. They go on to state that this particular directive suggests that the foundations of white superiority are still alive and well in New Zealand today, that the current policy is an affront to scientific and public health knowledge, that it requires rejection from health professionals and the scientific community and that the new policy threatens the efforts of the health and scientific communities in identifying and addressing inequities in health. 

 

I agree that many variables can assist in identifying and assessing need, of which self-reported ethnicity is one. Here, I discuss the assertions of Loring et al. and I provide further reactions to their article in Appendix 1. 

 

Loring et al. believe that ethnicity is an effective marker for identifying need in public health and that other “colour-blind” variables are not better proxies for health need. They are concerned that failure to utilize ethnicity will encourage weak analytic science and probably will lead to greater waste of public resources because of less effective targeting of resources towards groups of high need.

 

My response is that, because of lack of clarity about definitions, multiple ethnicities in many peoples’ ancestry and changing demographics, most probably other indices are at least as effective as, and possibly better than, ethnicity as markers, especially socioeconomic measures. Other useful markers include locality (i.e. the contextual difference between Remuera in Central Auckland and Māngere in South Auckland), local environment and associated hazards, age, gender, sexuality and disability. To clarify the notion of ethnicity in respect of Māori, at present standards are under development for Māori business, ethnicity, Māori descent and iwi affiliation (portal.zero.govt.nz, 2023). In any case, we will find significant overlaps in target populations that are categorized by ethnicity, locality and socioeconomic status. 

 

Used in many countries, including the United States, ethnicity can be a useful marker for public health analysis and policy but, like other markers, has inherent limitations. The validity and effectiveness of ethnicity are confounded to some extent by lack of consensus on definitions and disagreement on its measurement. Differences in terminology, self-perceptions of ethnic identity, methods and conduct of data collection, and evolving demographics make for challenges in using ethnicity as a marker for targeting and monitoring. Of course, self-reported ethnicity can be a matter of self-perceived membership of a particular cultural group, thus causing problems of consistency, especially where people report multiple ethnicities within their biological heritage. 

 

Ethnicity as a Marker of Public Health in New Zealand 

In New Zealand equity adjustor mechanisms used in recent years to prioritize surgical intervention involved parameters such as clinical speciality, clinical need, days already waited on the waiting list, ethnic group, socioeconomic deprivation and geographic area in order to determine the booking order, and final sign-off from doctors (Te Whatu Ora, 2024a). However, under Health Minister, Shane Reti, these mechanisms are undergoing revision following a review that found that they were legally and ethically justifiable but did not follow best practice (see Moir, 2024). 

 

A Government circular states that many variables can be used to identify and assess need and that all variables should be considered before ethnic identity is used automatically to determine need (DPMC, 2024). It states that Government’s expectation is that public services should be prioritized on the basis of need rather than race. 

 

“Government seeks to ensure that all New Zealanders, regardless of ethnicity or personal identity, have access to public services that are appropriate and effective for them, and that services are not arbitrarily allocated on the basis of ethnicity or any other aspect of identity.”  DPMC (2024). It is very important to note that neither ethnicity nor other aspects of identity are excluded from Government’s set of markers, but that they are not used automatically. I agree with this general approach. 

 

We can think of ethnicity as a social-political construct, including the notions of shared origin, language, cultural traditions, diet, ancestry and physical features. Indeed, much of the differentiation in certain diseases across populations can be explained by social-environmental factors that influence both susceptibility to certain illnesses and access to medical care. Particular minority groups may have dietary or lifestyle habits and socioeconomic status that give rise to predispositions towards conditions such as cardiovascular disease and diabetes, and that are associated with reduced access to preventive and emergency medical care and advice. Genetics may also play a part. 

Loring et al. believe that ethnic health inequities in New Zealand are unjust and avoidable, and that it is the job of health professionals to use all available tools to intervene. I subscribe to the view that disparities in health outcomes across communities are very strongly associated with the Social Determinants of Health (see World Health Organization, 2024). Systemic bias and racism within the health sector may or may not be real and may or may not contribute to those disparities. If they are real, their contribution is unknown but most probably small. In a discussion of discrimination on the basis of gender (or “sex”):

 

“By "bias" we mean here a pattern of association between a particular decision and a particular sex of applicant, of sufficient strength to make us confident that it is unlikely to be the result of chance alone. By "discrimination" we mean the exercise of decision influenced by the sex of the applicant when that is immaterial to the qualifications for entry.” Bickel et al. (1975) 

 

Biases can manifest itself for all sorts of reasons, but “institutional racism” implies active discrimination; i.e. bigotry. Do we have evidence of bigotry within our health sector? 

 

Social Determinants that impact on health and wellbeing include education, access to health care, nutrition, culture, income, housing, genetics and family issues. These factors can affect health directly (for example, through damp, cold and overcrowded conditions, which increase the transmission of infectious diseases) and indirectly; for example, by limiting opportunities to engage in health-promoting behaviors. Lifestyle choices in relation to regular physical exercise, use of recreational drugs, smoking and consumption of alcohol are other factors, as are genetic differences across populations, awareness of provision of healthcare, accessibility to that provision, and cost and affordability. Many such factors lie largely outside the jurisdiction of the health sector. 

 

Analytic Approaches

While statistical and other analytic approaches may not provide the last word in investigations of health across groups and cannot rule out systemic bias completely, nevertheless such approaches are highly suggestive of economic circumstances as the primary, though not exclusive, agent of disparity. This finding is also true of education, and relative underachievement of Māori in education can be explained by disparities in socio-economic status during childhood (Marie et al., 2008).

 

Given clear limitations of self-reported ethnicity as a marker, targeting by combinations of other indices is to be preferred, including family income, measures of socioeconomic deprivation such as NZDep (Health Geography and Deprivation, 2018), rurality, age, gender, sexuality and disability.  

 

While Loring et al. focus on one ethnic and cultural group only, for several years I have attempted to convey the message that New Zealand’s Pacific people are amongst those most affected by disparities in the socioeconomic determinants of health, including living in areas of high socioeconomic deprivation, being unemployed and having low weekly earnings (Lillis, 2023). Unfortunately, Loring et al. make no mention of New Zealand’s Pacific community. 

 

Inequity or Disparity?

I agree with the view of Loring et al. that persistent disparities for Māori in life expectancy, exposure to risk factors, access to care and health outcomes indicate that interventions have not met Māori needs. Surely the same is true of Pacific people. However, we must identify the reasons for relative failure objectively, some of which may lie outside the health sector. 

 

Apart from physical conditions, mental disorders present the sixth most significant cause of health loss globally, and anxiety and depression are among the most prevalent (Institute for Health Metrics and Evaluation, 2024). In New Zealand, between 50% and 80% of the population will experience mental distress or addiction challenges in their lifetime (Patterson et al., 2018), and we see increasing rates of poor mental health among youth, particularly Māori, Pacific people and those experiencing deprivation. In New Zealand, youth mental health appears to have deteriorated from 2012 onwards, but the deterioration was particularly severe for Māori, Pacific people, females and those experiencing high deprivation (Sutcliffe et al., 2023). Again, socioeconomic status emerges as a predictor of disparity. 

 

Loring et al. use the term “inequity”, which implies systemic unfairness. However, it is not clear how pervasive systemic unfairness, or indeed bias, may be in either health policy or delivery. Perhaps it is more correct to say that disparities persist today but that the reasons for those disparities are numerous and complex. They include socioeconomics, poor housing and discrepancies in educational attainment and, in my view, may also include low levels of awareness of existing health services within certain communities and reduced accessibility to those services for some people. 

 

Naturally, a highly-remunerated European professional can meet the cost of a consult to a General Practitioner, or even more expensive treatment, very easily whereas a minority person on a low wage and supporting a family may find the cost almost impossible to meet. 

 

Outcomes from Socioeconomics or System Issues?

If health outcomes constitute the aggregated result of socioeconomic determinants and health system issues, then Loring et al. raise valid concerns about the sources of disparities that are worthy of further investigation, and other studies may support their case. For example, after conducting statistical modelling on patient mortality data, Gurney et al. (2021) found disparities in post-operative mortality between Māori and Europeans in acute cardiovascular, digestive system and musculoskeletal system procedures in publicly-funded hospitalizations and those privately-funded hospitalizations at larger private facilities that report to the National Minimum Dataset (see Te Whatu Ora, 2024b). 

 

Disparities for Pacific patients were broadly similar to those of Māori, but display reduced statistical precision because of smaller numbers of patients and fewer deaths. Nevertheless, their findings for Pacific patients deserved further discussion, as did their findings for Middle Eastern/Latin American/African/Other patients (MELAA/Other). 

 

I note the apparent absence of Pacific authors on this paper whereas, of the 16 authors (on the basis of information provided on relevant web pages), roughly half appear to have Māori affiliation. They assert that disparities in outcomes for the Indigenous Māori population can be considered a breach of the Treaty of Waitangi. My reaction is that if the Treaty, which was given legal effect in Section 6 of the Pae Ora (Healthy Futures Act 2022) Act (Parliamentary Council Office, 2022) is to be invoked in the reporting of medical research and in the development of policy, then all demographic groups, especially Pacific, should be included in the discussion. I note that the original version of Section 6 mentions the word “Māori” fourteen times but does not mention any other ethnicity.

 

In passing, I have performed a more comprehensive word count on the Pae Ora Act and note that, as at 30 June 2024, the entire Act mentions the word “Māori” 473 times across 84 sections. By comparison, the word “Pacific” is mentioned 18 times in seven sections, while the words “Asian”, “Pakeha” and “European” do not appear at all. 

 

Modelling Disparities in Post-Operation Mortality

Of the 876,976 acute surgical procedures evaluated in the study of Gurney et al., 22% were for Māori. Thus, Māori were somewhat over-represented by comparison with their presence in the total population whereas, at 60%, Europeans were somewhat under-represented. Pacific people were somewhat under-represented (6%), while Asians were markedly under-represented (2%, by comparison with more than 15% in the total population). 

 

Of the 2,990,726 elective/waiting list surgeries, both Māori (16%) and Europeans (69%) were represented at approximately their presence in the total population. Pacific people were somewhat under-represented (6%) while Asians were under-represented (7%).  

 

Gurney et al. developed statistical models in order to explain the observed disparities. After accounting for age, sex, deprivation, rurality, comorbidity, ASA score (a score of physical status, as follows - healthy or mild/moderate disease; severe but stable disease; severe disease with disease with immediate threat to life or unknown); anaesthetic type; procedure risk and procedure specialty, their fully-adjusted model for all acute surgical procedures suggested that Māori are 14% more likely to die within 30 days. In addition, Māori appear to be 35% more likely to die within 30 days for all elective (non-emergency surgery that can be scheduled in advance) and waiting list procedures combined.

 

I agree that their models suggest disparities in outcomes but the true magnitude of those disparities is uncertain. They use a commonly-used method, known as “age standardization”, to correct for dissimilar age distributions across the ethnic groups, and then a second method involving “hazard ratios”, to estimate the ratio of the chances of mortality occurring within the Māori groups and within the European and other groups etc. Appendix 2 outlines the statistical procedures adopted by Gurney et al. and gives my brief comments on those procedures. Appendix 3 discusses the strengths and limitations of age-standardization. 

 

Their statistical modelling is entirely in accordance with internationally-accepted approaches, but involves a variety of unavoidable assumptions and considerable uncertainties. For example, the “crude” (prior to modelling) mortality among Europeans in acute surgery is actually higher (2.3 per 100 patients), rather than lower, than that of the other groups (1.4 per 100 for Māori) and at least as high for Elective/Waiting list procedures. Following age standardization, the mortalities for Europeans drop below those of other groups; in other words, we see reversal of the order of mortalities. Reversal occurs for both categories of surgery, presumably because the European group is older than the others but, if so, this point is not explained. 

 

Gurney et al. used 95% confidence intervals, fully in keeping with approaches used internationally. However, the estimated differences between groups are not large and, had wider confidence intervals been used (e.g. 99%), then the evidence for disparities would have been much less clear. This is especially true when not all causes have been accounted for completely (e.g. severity of illness).

 

Reversal of the relative magnitudes of mortality following age standardization is well-known, but the reversal of the European crude mortality is severe and considerable statistical uncertainties are involved. In any case it is the age standardization step that reduces the very high crude European mortality for acute surgery to below that of the other groups. Thus, I believe that Gurney et al. have provided evidence of disparities, but the stated magnitudes of those disparities are partly an artefact of their modelling procedures and the true or “real world” magnitudes are unknown.  

 

Accounting for Relevant Causes

Their models suggest disparities that remain even after accounting for factors such as age, comorbidity and socioeconomic deprivation, and they propose that the remaining disparities may, at least partly, reflect differences in the quality of surgical care received by Māori patients compared to European patients. They say that Māori may wait longer for treatment and that the treatment they receive could be of poorer quality than that delivered to non-Māori ethnic groups. For example, on the basis of prior research, they report that Māori patients were less likely to have their resection performed in a main treatment hub and less likely to have a specialist upper-gastrointestinal surgeon perform their gastric resection, even when the surgery was carried out in a major urban centre (Signal et al., 2015). 

 

That particular study involved stomach cancer patients who were diagnosed between 2006 and 2008. I have read that paper and agree with their assertion of differential presentation and access to specialized surgical services, as well as differential survival, for Māori stomach cancer patients compared to non-Māori. The caveats here are that at that time Māori patients tended to present with higher comorbidities and that comorbidity may have played a role in the decision to treat. It would be interesting to know the equivalent findings for Pacific people during those years and we hope that matters have improved since then. 

 

Signal et al. suggest that the observed disparities are “likely driven by structural factors including institutional racism acting through a combination of healthcare system, process and clinical team factors”.  They do note that they may not have adjusted completely for the impact of severity of illness at time of surgery. My opinion is that severity of illness must be included fully within any such analysis. Further, the greater average capacity of Europeans and Asians to meet the cost of private care that offers shorter waiting times, and differences in access to post-operative care and indeed self-care following operations, may also be significant factors. While I do not believe that the authors are engaging in motivated reasoning, nevertheless, their suggestion of institutional racism without accounting in full for all possible causal factors is a matter of concern. 

 

In relation to self-care, a financially-well off European is in a better position to take time off work to recover following an operation, whereas a Māori or Pacific person who is supporting a family on limited income may not have the means to take time off. Perhaps differences in self-care are also significant. Thus, it is not clear that discrepancies in quality of surgical procedures across ethnic groups, including bias in surgical prioritization and selection, were significant reasons for the observed disparities and it seems that further work on this issue is necessary. 

 

My opinion of the statistical work of Gurney et al. is positive but I recommend that future studies take greater account of severity of illness and discriminate between public and private procedures. 

 

Government’s Vision for Health

This year, the Māori Health Authority was disestablished. The relevant Cabinet Paper, entitled ‘Vision and priorities to address Māori health need’, proposes to shift decision-making and accountability closer to homes and communities, enabling local leadership, collaboration and innovation to meet needs (Ministry of Health, 2024). It promises to continue to focus on monitoring Māori health at all levels of the system. Further, it asserts that by reducing central bureaucracy and integrating Māori health expertise within the mainstream health system, New Zealand will have a clearer pathway towards decentralization, enabling greater community leadership in meeting local needs. 

 

It says that the causes of poor Māori health are complex, driven largely by the broader social, economic and behavioral determinants of health and further perpetuated by difficulties

in access to, and quality of, health services. It commits to a continued role of Iwi-Māori Partnership Boards in determining local priorities for the health system and it says that other population groups with high health needs will also benefit from moving decision-making and resourcing closer to communities. 

 

The Cabinet Paper states that good progress has been made across some indicators, including a substantial increase in funding to kaupapa Māori providers, an increase in the number of rongoā providers, and increases in the proportion of Māori healthcare professionals and appointment of Māori to statutory heath roles. Thus, the Cabinet Paper conveys commitment to all ethnic groups, including Māori. 

 

Conclusion

The World Health Organization provides a list of Social Determinants of Health (World Health Organization, 2024). Their list is as follows: Income and social protection; Education; Unemployment and job insecurity; Working life conditions; Food insecurity; Housing, basic amenities and the environment; Early childhood development; Social inclusion and non-discrimination; Structural conflict; Access to affordable health services of decent quality.

 

The World Health Organization suggests that the Social Determinants can influence health more greatly than healthcare or lifestyle choices and that the contribution of sectors outside the health sector to population health exceeds the contribution from the health sector itself. It is my belief that in New Zealand health and social outcomes for disadvantaged minorities could be improved by building partnerships between community groups, health providers and other relevant agencies. Such partnerships could enhance the support that is already available to vulnerable communities and help to address the present disparities.  

 

Conferring special privilege to any one ethnic or cultural group will not repair inequality; nor will consuming scarce resources to address structural racism and bias if these factors are small, or in practice no longer present, and if the core structural and systemic problems lie elsewhere. Thus, in New Zealand the true agents of disparity, mainly socioeconomic in nature, may lie largely outside the jurisdictions of education, health and even science, and we have a duty of care to address those causes if we are to enhance the health and wellbeing of all New Zealanders. Nevertheless, I recommend that further work is undertaken on aspects of public health where disparities are evident, including waiting times, quality of surgery, fairness in decision-making on selection and prioritization of patients, and quality of post-surgical medical care for all patients. 

 

Other studies similar to that of Gurney et al., should be conducted in order to provide reliable estimates of disparities in outcomes, taking full account of comorbidity, socioeconomic status, post-surgery care, post-surgery self-care, the variable capacities across communities to meet the cost of private care, accessibility and, finally, severity of illness. 

 

The stated view of Gurney et al. is that the observed disparities are “likely driven by structural factors including institutional racism acting through a combination of healthcare system, process and clinical team factors”. While acknowledging their view, I treat this assertion with great caution. 

 

Disparities in the present often are ascribed unfairly to systemic bias and racism. This is not to assert that bias, unconscious or otherwise, and racism absolutely do not exist. Indeed, unconscious bias may play a part in creating disparities but the extent of the contribution is unknown and so I recommend that further work is done on the issues identified by Gurney et al. and other studies. Otherwise, apart from anecdote, at present the evidence for systemic bias within the health sector today is ambiguous. 

 

APPENDIX 1

Reponses to Selected Quotes from Loring et al. (2024)

The term “race” originates from a long-discredited presumption of a biological hierarchy of human beings from white to black, and for decades the New Zealand health system has instead used “ethnicity”. This return to discredited terminology suggests that the foundations of white superiority are still alive and well in New Zealand today.

Why does the use of particular terminology suggest the presence of white superiority? 

 

[Government policy] says that agencies must include an assessment of any opportunity costs for all New Zealanders, and “where culturally specific models are used, eligibility should not be restricted to the specific population group unless there is a strong rationale (e.g. value for money).” This directive, and the political discourse surrounding it, is an affront to scientific and public health knowledge. It requires explicit rejection from health professionals and the scientific community.

I agree with Government that eligibility for analysis of needs or, for that matter, for provision of healthcare should not be restricted to any specific population group. If resources are to be allocated on the basis of need as identified by multiple markers, then on what basis is Government’s policy an affront to scientific and public health knowledge and why does it require explicit rejection from health professionals and the scientific community?

 

While not forgetting or diminishing that Māori have inalienable rights to health, and rights-based arguments for addressing health inequities, there is a strong connection between current Māori health needs and the denial of these rights.

Each and every New Zealander has rights to health and wellbeing, including Māori and Pacific people, Asians, Europeans and other populations. Do those strong connections between current Māori health needs and the denial of these rights apply to other ethnic groups; for example, to Pacific people? Given the concerns of Gurney et al., I accept the possibility that denial of rights does exist and, without judgement and in good faith, I invite Loring et al. to provide the evidence for such denial.

 

A few years ago I was somewhat against the establishment of the Māori Health Authority because singling out one particular community is essentially undemocratic and because other communities also suffer shortfalls in health and wellbeing. However, once established, it would have been an interesting experiment to see whether the Authority led to improved outcomes for Māori. For that reason I was mildly against disestablishing it, though the argument for devolving decision-making to local agencies and communities remains very compelling. 

 

Inequities in health need, access and outcomes persist for Māori at all levels of socio-economic deprivation, where-ever they happen to live.

Regrettably, this assertion is absolutely correct, as it is for Pacific People, provided that we substitute the term “disparity” for “inequity”. Therefore, New Zealand must address the principal causes that most probably include: 

 

1.         Household Overcrowding and Home Ownership

2.         Employment, Income and Deprivation 

3.         Genetic Factors

4.         Lifestyle Choices

5.         Accessibility

6.         Affordability

7.         Awareness of Provision of Healthcare.

 

However, we must also consider the possibility of system-wide issues such as disparities in elective and waiting list procedures of the kind identified by Gurney et al. Only objective research and evaluation will identify any system-wide issues that could lead to disparity. 

 

In requesting that other variables be considered before ethnicity, the government erroneously singles out ethnicity to require a higher standard of proof than allocations based on any other population risk characteristic (for example, rurality, sex or age).

Most probably socioeconomics and other markers are much more strongly associated with health outcomes across populations than self-reported ethnicity. We see the same effect in education (Marie et al., 2008).

 

Comprehensive, consistent and long-standing evidence demonstrates that ethnicity is a stronger marker of need than other commonly accessible variables such as rurality and the New Zealand Index of Deprivation (NZDep).

The authors may be correct in citing the existence of comprehensive evidence of the effectiveness of ethnicity as a marker and I invite them to provide that evidence. 

 

Our most widespread marker for socio-economic deprivation, NZDep, does not assess individual characteristics, but is based on a collective neighbourhood score.

Certainly, this is a limitation of the NZDep, but other variables could be used or are used too. It is important to note that the data zones within the NZDep can go to a relatively fine level of aggregation. The 2018 version of New Zealand data zones comprises 6,181 data and aims to maintain a target population range of between 500 and 1200 people (Health Geography and Deprivation, 2018).

 

Data zones are not intended to reflect the true extent of actual communities, but instead to facilitate small-area analyses of health and social data at a scale small enough to be statistically robust and convey a sense of neighborhood. In suburban areas they are a few streets long and a few streets wide. Thus, while not perfect, the NZDep is nevertheless is a reasonably effective marker for public health. 

 

Racism distributes the determinants of health along ethnic lines and impacts health directly, so until racism is eliminated, ethnicity will be a valid marker of need.

It is quite possible that racist thinking, prejudice and misogyny persist among certain people and within certain institutions. However, labels such as racism, systemic bias, conscious and unconscious bias and colonization not only may be applied unfairly, but possibly distract us from focusing on and addressing the true causes. On the other hand, New Zealand has work to do in order to identify how and why disparities in post-operative mortality and across the wider health sector occur in practice. 

 

Knight (2022) believes that systemic bias and racism do not exist in our health system, at least to any significant extent. He too suggests that disparities in health outcomes across demographic groups emerge largely from socioeconomic factors (especially in relation to housing), differences in genetics and lifestyle choices, including personal choices on exercise, consumption of alcohol and recreational drugs and smoking. He argues against a commonly-articulated view that primary contributing factors toward Māori ill-health include systemic racism, white privilege and unconscious bias. He says that Māori having a higher death rate from a range of causes is consistent with the presence of co-morbidities, genetics and poor primary health care.

 

Inequities in health need, access and outcomes do indeed persist for Māori at all levels of socio-economic deprivation, wherever they happen to live. 

Unfortunately, this assertion is correct. The same appears to be true for New Zealand’s Pacific people, whose indices of health and wellbeing are in many cases even poorer than those of Māori in both childhood and adult illness.

 

Racism distributes the determinants of health along ethnic lines and impacts health directly, so until racism is eliminated, ethnicity will be a valid marker of need.

The assertion of systemic bias or racism as a cause of disparity, and thereby providing justification for major policy and legislative change, is evident in domains other than health; for example, in education. However, recognizing the problems raised in Gurney et al., the extent of systemic bias within these domains is difficult to determine objectively and could only be evaluated through research rather than anecdote.   

 

Similarly, there is no basis for using the individual exception (for example, “I’m Māori and I don’t have high health needs”) as a justification for not targeting high-risk populations. This represents a fundamental misunderstanding of individual versus population risk and applies to any population characteristic, not just ethnicity.

Are individual exceptions used to justify not targeting at-risk populations?  In any case, within conversations on public health, particularly those concerning the Treaty, I wish to see similar attention to enhancing the health and wellbeing and economic status of Pacific people and other individuals, families and communities in need.

 

Like every country, we have a duty to allocate scarce health resources to those most at risk, and to use all available risk characteristics to identify those most in need as sensitively and specifically as possible.

I agree in full and suggest that self-reported ethnicity is one imperfect but potentially useful marker if used in conjunction with other markers. 

 

These measures are crucial to address discrimination that already exists in our health system. We must remember that the status quo is not a neutral starting point, but instead has a pre-existing ethnic bias towards our dominant ethnicity.

In what ways does discrimination play out in New Zealand’s health system? Quite possibly, our health system of decades ago was characterized by favorable bias towards Europeans, but is this assertion true today? Perhaps the answer is that at an aggregate level the system attempts fairness and colour-blind policies and provision, but imbalances have emerged of the kind identified by Gurney et al. Perhaps imbalances occur at some level of decision-making, selection, prioritization and allocation of medical care. Further work on the issues identified there is fully warranted. 

 

The government’s directive is not just an attack on Māori, but an attack on science and good medical practice.

In its Cabinet Paper, Government has committed to a continued role of Iwi-Māori Partnership Boards in determining local priorities for the health system. So, in what way is a decision to use indices other than ethnicity an attack on any ethnic group? If it constitutes an attack on one group, then does it not also constitute an attack on other groups as well? It is really an attack on science, when other effective markers are used to identify those in need and to monitor their progress? 

 

Anyone who supports this directive, either actively or complicitly through their silence, is supporting the undermining of our collective scientific knowledge and commitment to evidence-based medical practice.

I do not agree. The use of ethnicity as a marker is downplayed but not eliminated. Further, if systemic bias and racism are relatively minor today, then we should focus on the true causes of disparity - the Social Determinants of Health - but also attempt to raise awareness of health services and accessibility to those services. 

 

APPENDIX 2

Gurney et al. (2021) compared age-standardized rates between ethnic groups (Māori, Pacific, Asian, European, MELAA/Other) and calculated hazard ratios using Cox proportional hazards regression modelling, adjusted for age, sex, deprivation, rurality, comorbidity, ASA score, anaesthetic type, procedure risk and procedure specialty. They used the 2001 Total Māori Population as the standard population, an Indigenous standard that uses Māori as the target population (Ministry of Health, 2018).

 

In Table 1 of Gurney et al. we see that age standardization has reduced the mortality in acute procedures for Europeans much more than for Māori, Pacific and others, presumably because of a much older age distribution for the European group. The table below gives the relevant data, where the figures are death rates (mortality) per 100 patients, both crude (raw) data and the mid-points of the modelled rates, given in parentheses. I have calculated the given ratios of the crude rates and the adjusted rates in order to illustrate the difference between the European data and data for the other groups. Europeans have much higher crude mortality but the European data undergoes much greater deflation after age standardization. 

 

	Māori	Pacific	Asian	MELAA/Other	European
Crude and Adjusted	1.4  ( 1.1)	1.3  ( 1.0)	1.3  (0.8)	1.3  (0.7)	2.3  ( 0.7)
Ratio	1.27	1.3	1.63	1.86	3.29

 

Table 1 of Gurney et al. also shows higher deflation of the crude mortality for Europeans across all specialities - Cardiovascular, Digestive system, Respiratory system, Neurosurgery, Musculoskeletal, Urinary system and Other. 

 

The table below gives the equivalent data for Elective/Waiting list procedures.

 

	Māori	Pacific	Asian	MELAA/Other	European
Crude and Adjusted	0.2  ( 0.2)	0.2  ( 0.1)	0.1  (0.1)	0.1  (0.1)	0.2  ( 0.1)
Ratio	1.0	2.0	1.0	1.0	2.0

 

As expected, we see somewhat higher deflation of crude mortality for Europeans than the other groups, except Pacific. This deflation for Europeans is entirely reasonable and within normal bounds. 

 

Age standardization is a commonly-used method for comparing populations in medical studies and the adjusted rates take account of age, sex, socioeconomic level and other factors. The approach of Gurney et al. is fully in accordance with internationally-accepted methods and the higher downward adjustments are justified if the European group had an older age distribution than the others. 

 

However, that the rates for Europeans were reduced by a much larger extent than for the other groups for acute surgery, means that great care must be exercised in reporting and commenting on modelled disparities, especially if research studies such as Gurney et al. are to be used in health policy.  

 

In any case, the large downward adjustment of the modelled European mortality for acute surgery leads to large hazard ratios for the other groups. The combination of standardization and subsequent calculation of hazard is a perfectly acceptable and commonly-used approach but, in addition to high hazard ratios, involves compounding errors, so that the predicted 14% additional mortality for Māori in acute surgery over that of Europeans must be accepted in principle but treated with caution. The calculated disparities must be taken seriously but to some extent the higher mortalities for Māori and other groups than for Europeans are an artefact of the statistical procedures. 

 

Gurney et al. do note that they may not have adjusted completely for the impact of severity of illness at time of surgery. Severity of illness must be included within any future analysis. Further, the greater average capacity of Europeans and Asians to meet the cost of private care, differences in access to post-operative care and indeed self-care following operations, may also be significant. 

 

Appendix 3

Age-standardization is applied to account for differences in age distributions across groups. However, age-standardized rates have known limitations. Standardizing to the age distribution of one population (e.g. Māori or Pacific) to those of other populations can obscure insight into the population of interest. Thus, age-specific statistics centered on Māori can indeed provide good measures of health risks and identify high-risk subgroups within the population of interest.

However, age-standardized rates do not provide accurate measures of actual rates where populations have very different age structures. 

 

Accordingly, we must consider the limitations of age-standardization because average measures may not convey the true magnitude of risks or disparities and often are insensitive to change in demographics over time (Thurber et al., 2022). Further, age-standardized rates can vary considerably with the choice of reference population and so the choice of reference population can affect any interpretation of comparisons. However, in spite of known limitations, often age-standardized rates are reported without necessary context.  

 

Before undertaking age-standardization, researchers should investigate the data in order to identify any limitations that may affect the analyses. If the age-standardized rates lie largely outside the age-specific rates, then additional reporting of age-specific measures should be provided (Thurber et al., 2022).

 

Irrespective of the reference population that is used, standardized statistics provide summary measures only and may in fact mask relationships within the data. Aggregating all age groups within a single summary statistic implicitly combines subgroups with variable mortality rates, leading to imprecise estimates.

 

Focusing analysis on a given population produces standardized mortality rates that are similar to the crude (actual) mortality rates of that population, but the calculated rates for other populations may be greatly different from their crude rates (Thurber et al., 2022).

 

Use of an Indigenous Standard tends to increase the calculated risk ratios, as it accords proportional weight to young and middle-aged groups who are under-represented in the total population and where age-specific Indigenous/non-Indigenous risk ratios are highest. Accordingly, it gives reduced weights to the older age groups where crude death rates are high. In New Zealand, use of the Māori Standard led to larger Māori vs. non-Māori risk ratios (Thurber et al., 2022).

 

Finally, when Indigenous Standards are introduced, the most current population should be used.

Dr David Lillis trained in physics and mathematics at Victoria University and Curtin University in Perth, working as a teacher, researcher, statistician and lecturer for most of his career. He has published many articles and scientific papers, as well as a book on graphing and statistics.  

 

References

Bickel et al. (1975) (Science 187: 398-404) 

 

DPMC (2024). CO (24) 5: Needs-based Service Provision

https://www.dpmc.govt.nz/publications/co-24-5-needs-based-service-provision#purpose

 

Gurney, Jason; McLeod, Melissa; Stanley, James; Sarfati, Diana; Campbell, Doug; Davies, Cheryl; Dennett, Elizabeth; Himona, Peter; Jackson, Sarah; Ongley, Dick; Robson, Bridget; Rumball-Smith, Juliet; Signal, Virgina; Stairmand, Jeannine; Thomas, Courtney and Koea, Jonathan (2021).

Disparities in post-operative mortality between Māori and non-Indigenous ethnic groups in New Zealand NZMJ 17 September 2021, Vol 134, No 1542. ISSN 1175-8716

https://nzmj.org.nz/media/pages/journal/vol-134-no-1542/disparities-in-post-operative-mortality-between-maori-and-non-indigenous-ethnic-groups-in-new-zealand/0a35c7817e-1696476828/disparities-in-post-operative-mortality-between-maori-and-non-indigenous-ethnic-groups-in-new-zealand-open-access.pdf 

 

Health Geography and Deprivation (2018). New Zealand Data Zones 2018

https://hgd.auckland.ac.nz/datazones/

 

Institute for Health Metrics and Evaluation (2024). Mental Health

https://www.healthdata.org/research-analysis/health-risks-issues/mental-health  

 

Knight, Lawrie (2022). Fact checking Māori health claims that led to the Pae Ora (Healthy Futures) Bill

https://breakingviewsnz.blogspot.com/2022/03/dr-lawrie-knight-fact-checking-maori.html

 

Lillis, David (2023). Our Prioritised Health System and Pacific People

https://breakingviewsnz.blogspot.com/2023/01/david-lillis-our-prioritised-health.html

 

Loring, Belinda; Reid, Papaarangi; Curtis, Elana; McLeod, Melissa; Harris, Ricci and Jones, Rhys. (2024). ‘An attack on science and good medical practice’

New Zealand Medical Journal, 137(1603), 9-13.

https://e-tangata.co.nz/comment-and-analysis/an-attack-on-science-and-good-medical-practice/?utm_source=twitter&utm_medium=social&utm_campaign=07Oct2024

 

Marie, D., Fergusson, D. M. and Boden, J. M. (2008). Educational Achievement in Maori: The Roles of Cultural Identity and Social Disadvantage. Australian Journal of Education. 52: 2, 183-196. Article first published online: August 1, 2008; Issue published August 1, 2008.

https://journals.sagepub.com/doi/epdf/10.1177/000494410805200206

 

Ministry of Health (2018). Position paper on Māori health analytics – age standardisation.

Wellington, New Zealand: Ministry of Health.

 

Ministry of Health (2024). Disestablishment of the Māori Health Authority – Next Steps on Māori health

https://www.health.govt.nz/system/files/2024-08/disestablishment_of_the_mha_next_steps_hauora_maori_bundle_120824_black_box.pdf

 

Moir, Jo (2024). Health NZ drops tool that factored in ethnicity for waitlists, despite review findings

https://www.rnz.co.nz/news/political/523825/health-nz-drops-tool-that-factored-in-ethnicity-for-waitlists-despite-review-findings

 

Parliamentary Counsel Office (2022).Pae Ora (Healthy Futures) Act 2022

https://www.legislation.govt.nz/act/public/2022/0030/latest/versions.aspx

 

Patterson, R., Durie, M., Disley, B. and Tiatia-Seath S. (2018). He Ara Oranga: Report of the government inquiry into mental health and addiction.

https://mentalhealth.inquiry.govt.nz/inquiry-report/he-ara-oranga/

 

portal.zero.govt.nz (2023). Mandated standards candidates selected for Māori business, ethnicity, Māori descent, and iwi affiliation

https://portal.zero.govt.nz/8660980c5dd14f440d65d61e6131f730/blog/mandated-standards-candidates-selected-for-maori-business-ethnicity-maori-descent-and-iwi-affiliation

 

Signal, V., Sarfati, D., Cunningham, R., Gurney, J., Koea, J. and Ellison-Loschmann, L. (2015). Indigenous inequities in the presentation and management of stomach cancer in New Zealand: a country with universal health care coverage. Gastric Cancer. 2015;18(3):571-9.

https://pubmed.ncbi.nlm.nih.gov/25098927/

 

Sutcliffe, K., Ball, J., Clark, T. C., Archer, D., Peiris-John, R., Crengle, S. and Fleming, T. (2023). Rapid and unequal decline in adolescent mental health and well-being 2012–2019: Findings from New Zealand cross-sectional surveys. Australian & New Zealand Journal of Psychiatry. Feb;57(2):264-82.

 

Te Whatu Ora (2024a). Evaluation of two tools used for waitlist prioritisation for planned care in Health New Zealand | Te Whatu Ora

https://www.tewhatuora.govt.nz/publications/evaluation-of-two-tools-used-for-waitlist-prioritisation-for-planned-care-in-health-new-zealand-te-whatu-ora

 

Te Whatu Ora (2024b). National Minimum Dataset Data Mart Dictionary.  https://www.tewhatuora.govt.nz/for-health-professionals/data-and-statistics/nz-health-statistics/data-references/data-dictionaries

 

Thurber, K. A., Thandrayen, J., Maddox, R., Barrett, E. M., Walker, J., Priest, N., Korda, R. J., Banks, E., Williams. D. R. and Lovett, R. (2022). Reflection on modern methods: statistical, policy and ethical implications of using age-standardized health indicators to quantify inequities. International Journal of Epidemiology, 2022, 324–333; doi: 10.1093/ije/dyab132Education Corner

 

World Health Organization (2024). Social determinants of health

https://www.who.int/health-topics/social-determinants-of-health#tab=tab_1