Population and contact tracer uptake of New Zealand's QR-code-based digital contact tracing app for COVID-19.

This study aimed to understand the population and contact tracer uptake of the quick response (QR)-code-based function of the New Zealand COVID Tracer App (NZCTA) used for digital contact tracing (DCT). We used a retrospective cohort of all COVID-19 cases between August 2020 and February 2022. Cases of Asian and other ethnicities were 2.6 times (adjusted relative risk (aRR) 2.58, 99 per cent confidence interval (95% CI) 2.18, 3.05) and 1.8 times (aRR 1.81, 95% CI 1.58, 2.06) more likely than Maori cases to generate a token during the Delta period, and this persisted during the Omicron period. Contact tracing organization also influenced location token generation with cases handled by National Case Investigation Service (NCIS) staff being 2.03 (95% CI 1.79, 2.30) times more likely to generate a token than cases managed by clinical staff at local Public Health Units (PHUs). Public uptake and participation in the location-based system independent of contact tracer uptake were estimated at 45%. The positive predictive value (PPV) of the QR code system was estimated to be close to nil for detecting close contacts but close to 100% for detecting casual contacts. Our paper shows that the QR-code-based function of the NZCTA likely made a negligible impact on the COVID-19 response in New Zealand (NZ) in relation to isolating potential close contacts of cases but likely was effective at identifying and notifying casual contacts.

Examining the impact of COVID-19 on Maori:non-Maori health inequities in Aotearoa, New Zealand: an observational study protocol.

INTRODUCTION: The COVID-19 pandemic has had both direct and indirect impacts on the health of populations worldwide. While racial/ethnic health inequities in COVID-19 infection are now well known (and ongoing), knowledge about the impact of COVID-19 pandemic management on non-COVID-19-related outcomes for Indigenous peoples is less well understood. This article presents the study protocol for the Health Research Council of New Zealand funded project 'Ma te Mohio ka Marama: Impact of COVID-19 on Maori:non-Maori inequities'. The study aims to explore changes in access to healthcare, quality of healthcare and health outcomes for Maori, the Indigenous peoples of Aotearoa New Zealand (NZ) and non-Maori during the COVID-19 outbreak period across NZ. METHODS AND ANALYSIS: This observational study is framed within a Kaupapa Maori research positioning that includes Kaupapa Maori epidemiology. National datasets will be used to report on access to healthcare, quality of healthcare and health outcomes between Maori and non-Maori during the COVID-19 pandemic in NZ. Study periods are defined as (a) prepandemic period (2015-2019), (b) first pandemic year without COVID-19 vaccines (2020) and (c) pandemic period with COVID-19 vaccines (2021 onwards). Regional and national differences between Maori and non-Maori will be explored in two phases focused on identified health priority areas for NZ including (1) mortality, cancer, long-term conditions, first 1000 days, mental health and (2) rheumatic fever. ETHICS AND DISSEMINATION: This study has ethical approval from the Auckland Health Research Ethics Committee (AHREC AH26253). An advisory group will work with the project team to disseminate the findings of this project via project-specific meetings, peer-reviewed publications and a project-specific website. The overall intention of the project is to highlight areas requiring health policy and practice interventions to address Indigenous inequities in health resulting from COVID-19 pandemic management (both historical and in the future).

Trends in penicillin dispensing during an acute rheumatic fever prevention programme.

AIM: Acute rheumatic fever (ARF), a serious inflammatory condition, often leads to rheumatic heart disease (RHD). Between 2011 and 2016, Aotearoa New Zealand implemented a rheumatic fever prevention programme (RFPP) to reduce high rates of ARF through improved community access to timely diagnosis and early treatment of group A streptococcal (GAS) pharyngitis, which has been shown to prevent subsequent ARF. This study aimed to quantify the change in penicillin antibiotic dispensing rates among children aged 18 years or younger during the RFPP. METHOD: This retrospective analysis utilised administrative data from the National Pharmaceutical Collection. Using a controlled, interrupted time series analysis, the effect of the RFPP on antibiotic dispensing rates was explored. Poisson regression models were used to assess the change in dispensing rates during the RFPP among control regions (those not in the RFPP) and regions participating in the RFPP. The primary measure was rate ratio (RR) for the difference between the observed versus counterfactual rates of penicillin dispensing. RESULT: A total of 12,154,872 dispensing records between 2005 and 2018 were included. Amoxicillin was the most frequently dispensed penicillin (57.7%), followed by amoxicillin-clavulanate (23.4%). Amoxicillin dispensing increased by 4.3% in regions operating the RFPP compared to the increase in control regions (p<0.001). The overall rate of penicillin dispensing decreased, driven by a rapid decline in amoxicillin-clavulanate dispensing. CONCLUSION: During the RFPP an increase in amoxicillin dispensing was seen in regions participating in the programme and regions outside of the programme, indicating the programmatic approach led to improved adherence to recommended first-line antibiotics.

Predicting high health-cost users among people with cardiovascular disease using machine learning and nationwide linked social administrative datasets.

OBJECTIVES: To optimise planning of public health services, the impact of high-cost users needs to be considered. However, most of the existing statistical models for costs do not include many clinical and social variables from administrative data that are associated with elevated health care resource use, and are increasingly available. This study aimed to use machine learning approaches and big data to predict high-cost users among people with cardiovascular disease (CVD). METHODS: We used nationally representative linked datasets in New Zealand to predict CVD prevalent cases with the most expensive cost belonging to the top quintiles by cost. We compared the performance of four popular machine learning models (L1-regularised logistic regression, classification trees, k-nearest neighbourhood (KNN) and random forest) with the traditional regression models. RESULTS: The machine learning models had far better accuracy in predicting high health-cost users compared with the logistic models. The harmony score F1 (combining sensitivity and positive predictive value) of the machine learning models ranged from 30.6% to 41.2% (compared with 8.6-9.1% for the logistic models). Previous health costs, income, age, chronic health conditions, deprivation, and receiving a social security benefit were among the most important predictors of the CVD high-cost users. CONCLUSIONS: This study provides additional evidence that machine learning can be used as a tool together with big data in health economics for identification of new risk factors and prediction of high-cost users with CVD. As such, machine learning may potentially assist with health services planning and preventive measures to improve population health while potentially saving healthcare costs.

We still don't count: the under-counting and under-representation of Maori in health and disability sector data.

AIM: To examine ethnicity data quality; in particular, the representation and potential under-counting of Maori in health and disability sector data, as well as implications for inequities. METHODS: Maori and non-Maori ethnicity data are analysed at: 1) a population aggregate level across multiple 2018 datasets (Estimated Resident Population, Census Usually Resident Population, Health Service User (HSU) population and Primary Health Organisation (PHO) enrolments); and 2) an individual level for those linked in PHO and 2018 Census datasets. Ethnicity is drawn from the National Health Index (NHI) in health datasets and variations by age and gender are explored. RESULTS: Aggregate analyses show that Maori are considerably under-represented in HSU and PHO data. In linked analysis Maori were under-counted on the NHI by 16%. Under-representation in data and under-counting occur across both genders but are more pronounced for Maori men with variations by age. CONCLUSION: High quality ethnicity data are fundamental for understanding and monitoring Maori health and health inequities as well as in the provision of targeted services and interventions that are responsive to Maori aspirations and needs. The continued under-counting of Maori in health and disability sector data is a breach of Te Tiriti o Waitangi and must be addressed with urgency.

Where Are We Dying? Ethnic Differences in Place of Death Among New Zealanders Dying of Cancer.

PURPOSE: Around a third of people with cancer will die outside of their preferred place of death, with substantial variation occurring between and within countries in terms of place of death. Here, we examine place of death within the New Zealand cancer context, with specific focus on differences between Indigenous Maori and other ethnic groups. METHODS: Using national-level data, we identified all those who died in New Zealand between 2007 and 2018 of cancer (N = 107,373), stratified by ethnicity and cancer type, and linked these patients to national health and mortality records. We then described the crude and age-standardized proportions of cancer deaths by location separately by ethnic group, and conducted logistic regression to compare odds of death within a given location between ethnic groups. RESULTS: After adjusting for age, sex, and deprivation, we found that Maori people with cancer are more likely to die in a private residence than Europeans (46% v 26%; odds ratio [OR] 2.45; 95% CI, 2.36 to 2.55), and also somewhat more likely to die in hospital (27% v 23%; OR 1.26; 95% CI, 1.21 to 1.32). Commensurately, Maori are less likely to die in either hospice inpatient unit (14% v 27%; OR 0.48; 95% CI, 0.45 to 0.51) or residential care (12% v 30%; OR 0.56; 95% CI, 0.52 to 0.59). Pacific patients generally follow the same pattern as Maori patients. These findings were largely repeated across cancer types, with some variation in the magnitude not overall pattern. CONCLUSION: It remains unclear whether these differences reflect differences in preferences for place of death between ethnic groups, or whether they reflect differences in access to appropriate supportive care. Further research is required to examine these differences in greater detail.

Long term exposure to air pollution, mortality and morbidity in New Zealand: Cohort study.

OBJECTIVES: To investigate associations between long-term exposure to PM2.5, NO2, mortality and morbidity in New Zealand, a country with low levels of exposure. DESIGN: Retrospective cohort study. SETTING: The New Zealand resident population. METHOD: The main analyses included all adults aged 30 years and over with complete data on covariates: N = 2,223,507. People who died, or were admitted to hospital, (2013-2016) were linked anonymously to the 2013 census, and to estimates of ambient PM2.5, and NO2 concentration. We fitted Poisson regression models of mortality and morbidity in adults (≥30) for all natural causes of death, and by sub- group of major cause. Person-time of exposure, censored at the time of death, was included as an offset. We adjusted for confounding by age, sex, ethnicity, income, education, smoking status and ambient temperature. Further analyses stratified by ethnic group, and investigated respiratory hospital admissions in children. RESULTS: There were statistically significant positive associations between pollutants and natural causes of death: RR (per 10 µg/m3) for PM2.5 1.11 (1.07 to 1.15) and for NO2 1.10 (1.07 to 1.12). For morbidity, the strongest associations were for PM2.5 and ischaemic heart disease in adults, RR: 1.29 (1.23 to 1.35) and for NO2 and asthma in children, RR: 1.18 (1.09 to 1.28). In models restricted to specific ethnic groups, we found no consistent differences in any of the associations. CONCLUSIONS: The results for NO2 are higher than those published previously. Other studies have reported that the dose-response for PM2.5 may be higher at low concentrations, but less is known about NO2. It is possible NO2 is acting as a proxy for other traffic-related pollutants that are causally related to health impacts. This study underlines the importance of controlling pollution caused by motor vehicles.

National Study of Pain Medicine Access Among Maori and Non-Maori Patients With Lung Cancer in New Zealand.

PURPOSE: Pain is among the most common and consequential symptoms of cancer, particularly in the context of lung cancer. Maori have extremely high rates of lung cancer, and there is evidence that Maori patients with lung cancer are less likely to receive curative treatment and more likely to receive palliative treatment and to wait longer for their treatment than non-Maori New Zealanders. The extent to which Maori patients with lung cancer are also less likely to have access to pain medicines as part of their supportive care remains unclear. METHODS: Using national-level Cancer Registry and linked health records, we describe access to subsidized pain medicines among patients with lung cancer diagnosed over the decade spanning 2007-2016 and compare access between Maori and non-Maori patients. Descriptive and logistic regression methods were used to compare access between ethnic groups. RESULTS: We observed that the majority of patients with lung cancer are accessing some form of pain medicine and there do not appear to be strong differences between Maori and non-Maori in terms of overall access or the type of pain medicine dispensed. However, Maori patients appeared more likely than non-Maori to first access pain medicines within 2 weeks before their death and commensurately less likely to access them more than 24 weeks before death. CONCLUSION: Given the plausibility that there are differences in first access to pain medicines (particularly opioid medicines) among Maori approaching end of life, further investigation of the factors contributing to this disparity is required.

Early diagnosis of acute rheumatic fever and rheumatic heart disease as part of a secondary prevention strategy: Narrative review.

Acute rheumatic fever (ARF) and its sequela rheumatic heart disease (RHD) remain significant causes of morbidity and mortality. In New Zealand, ARF almost exclusively affects Indigenous Maori and Pacific children. This narrative review aims to present secondary interventions to improve early and accurate diagnosis of ARF and RHD, in order to minimise disease progression in New Zealand. Medline, EMBASE and Scopus databases were searched as well as other electronic publications. Included were 56 publications from 1980 onwards. Diagnosing ARF and RHD as early as possible is central to reducing disease progression. Recent identification of specific ARF biomarkers offer the opportunity to aid initial diagnosis and portable echocardiography has the potential to detect undiagnosed RHD in high-risk areas. However, further research into the benefits and risks to children with subclinical RHD is necessary, as well as an economic evaluation.

Age disparities in lung cancer survival in New Zealand: The role of patient and clinical factors.

OBJECTIVE: Age is an important prognostic factor for lung cancer. However, no studies have investigated the age difference in lung cancer survival per se. We, therefore, described the role of patient-related and clinical factors on the age pattern in lung cancer excess mortality hazard by stage at diagnosis in New Zealand. MATERIALS AND METHODS: We extracted 22 487 new lung cancer cases aged 50-99 (median age = 71, 47.1 % females) diagnosed between 1 January 2006 and 31 July 2017 from the New Zealand population-based cancer registry and followed up to December 2019. We modelled the effect of age at diagnosis, sex, ethnicity, deprivation, comorbidity, and emergency presentation on the excess mortality hazard by stage at diagnosis, and we derived corresponding lung cancer net survival. RESULTS: The age difference in net survival was particularly marked for localised and regional lung cancers, with a sharp decline in survival from the age of 70. No identified factors influenced age disparities in patients with localised cancer. However, for other stages, females had a greater difference in survival between middle-age and older-age than males. Comorbidity and emergency presentation played a minor role. Ethnicity and deprivation did not influence age disparities in lung cancer survival. CONCLUSION: Sex and stage at diagnosis were the most important factors of age disparities in lung cancer survival in New Zealand.

The impact of timely cancer diagnosis on age disparities in colon cancer survival.

OBJECTIVE: We described the role of patient-related and clinical factors on age disparities in colon cancer survival among patients aged 50-99 using New Zealand population-based cancer registry data linked to hospitalisation data. METHOD: We included 21,270 new colon cancer cases diagnosed between 1 January 2006 and 31 July 2017, followed up to end 2019. We modelled the effect of age at diagnosis, sex, ethnicity, deprivation, comorbidity, and emergency presentation on colon cancer survival by stage at diagnosis using flexible excess hazard regression models. RESULTS: The excess mortality in older patients was minimal for localised cancers, maximal during the first six months for regional cancers, the first eighteen months for distant cancers, and over the three years for missing stages. The age pattern of the excess mortality hazard varied according to sex for distant cancers, emergency presentation for regional and distant cancers, and comorbidity for cancer with missing stages. Ethnicity and deprivation did not influence age disparities in colon cancer survival. CONCLUSION: Factors reflecting timeliness of cancer diagnosis most affected age-related disparities in colon cancer survival, probably by impacting treatment strategy. Because of the high risk of poor outcomes related to treatment in older patients, efforts made to improve earlier diagnosis in older patients are likely to help reduce age disparities in colon cancer survival in New Zealand.

Personality disorder prevalence and correlates in a whole of nation dataset.

PURPOSE: The epidemiology of personality disorder is poorly understood. This study aims to describe the population in contact with mental health services with a diagnosis of personality disorder and compare service use between this group and those with a diagnosis of depression. METHODS: Investigation of a routinely collected clinical data set across New Zealand was conducted. We used data from 2008 to 2017 and 1-year data from 2016, the most complete dataset. This allowed for variation over the years and confirmation within a 1-year prevalence. These data were analysed focusing on patients with a primary diagnosis of any personality disorder and the subset with borderline personality disorder (BPD). BPD was chosen as the most common clinical personality disorder diagnosis and that most researched. RESULTS: 8884 (2.8%) of the population in contact with mental health services carried a primary diagnosis of personality disorder. Personality diagnosis other than antisocial personality disorder (ASPD) in men and borderline personality disorder (BPD) in either gender was negligible. Rates of personality pathology increased with social deprivation and were commonest in young adults. Multi-morbidity was the norm, with internalising disorder common. Maori diagnosed with PD were over-represented both in the patient group and by population. CONCLUSION: Borderline personality disorder and antisocial personality disorder are routinely diagnosed in routine clinical practice in New Zealand; however, other categorical diagnoses are not made. Patients with PD are significant users of resources in comparison to depressed patients. Resource utilisation was significantly greater in those with PD, in particular use of inpatient services compared to those with depression.

Changing social inequalities in smoking, obesity and cause-specific mortality: Cross-national comparisons using compass typology.

BACKGROUND: In many countries smoking rates have declined and obesity rates have increased, and social inequalities in each have varied over time. At the same time, mortality has declined in most high-income countries, but gaps by educational qualification persist-at least partially due to differential smoking and obesity distributions. This study uses a compass typology to simultaneously examine the magnitude and trends in educational inequalities across multiple countries in: a) smoking and obesity; b) smoking-related mortality and c) cause-specific mortality. METHODS: Smoking prevalence, obesity prevalence and cause-specific mortality rates (35-79 year olds by sex) in nine European countries and New Zealand were sourced from between 1980 and 2010. We calculated relative and absolute inequalities in prevalence and mortality (relative and slope indices of inequality, respectively RII, SII) by highest educational qualification. Countries were then plotted on a compass typology which simultaneously examines trends in the population average rates or odds on the x-axis, RII on the Y-axis, and contour lines depicting SII. FINDINGS: Smoking and obesity. Smoking prevalence in men decreased over time but relative inequalities increased. For women there were fewer declines in smoking prevalence and relative inequalities tended to increase. Obesity prevalence in men and women increased over time with a mixed picture of increasing absolute and sometimes relative inequalities. Absolute inequalities in obesity increased for men and women in Czech Republic, France, New Zealand, Norway, for women in Austria and Lithuania, and for men in Finland. Cause-specific mortality. Average rates of smoking-related mortality were generally stable or increasing for women, accompanied by increasing relative inequalities. For men, average rates were stable or decreasing, but relative inequalities increased over time. Cardiovascular disease, cancer, and external injury rates generally decreased over time, and relative inequalities increased. In Eastern European countries mortality started declining later compared to other countries, however it remained at higher levels; and absolute inequalities in mortality increased whereas they were more stable elsewhere. CONCLUSIONS: Tobacco control remains vital for addressing social inequalities in health by education, and focus on the least educated is required to address increasing relative inequalities. Increasing obesity in all countries and increasing absolute obesity inequalities in several countries is concerning for future potential health impacts. Obesity prevention may be increasingly important for addressing health inequalities in some settings. The compass typology was useful to compare trends in inequalities because it simultaneously tracks changes in rates/odds, and absolute and relative inequality measures.

An estimate of limited duration cancer prevalence in New Zealand using 'big' data.

AIMS: Increases in cancer survival may increase cancer prevalence and demand for healthcare. We aimed to estimate cancer prevalence in the New Zealand population. METHODS: We used national linked health, social and census datasets from the Stats NZ Integrated Data Infrastructure to identify the number of New Zealand residents who had at least one cancer diagnosis in New Zealand. We included all primary cancers recorded on the New Zealand Cancer Registry from January 1995 to June 2013, and used the 2013 census for demographic and socioeconomic data. RESULTS: On 30 June 2013, 140,600 of 4,438,900 (3.2%) New Zealand residents had been diagnosed with cancer in the last 18.5 years. In ≥15 year olds, the age-standardised prevalence of cancer diagnosed 0 to ≤1 year, and >1 to ≤5 years, prior to 30 June 2013 was 0.4% and 1.1% in men and 0.3% and 0.9% in women, respectively. Over the 18.5-year period prevalence was greatest in the oldest ages, European/Other, highest qualified, highest income, least deprived, ex-smokers, and Canterbury, Bay of Plenty and Nelson/Marlborough District Health Boards (age-standardised). CONCLUSIONS: Groups with the highest survival and the greatest access to healthcare had the highest cancer prevalences.

Is mode of transport to work associated with mortality in the working-age population? Repeated census-cohort studies in New Zealand, 1996, 2001 and 2006.

BACKGROUND: Increasing active transport is proposed as a means to address both health and environmental issues. However, the associations between specific modes, such as cycling, walking and public transport, and health outcomes remain unclear. We examined the association between mode of travel to work and mortality. METHODS: Cohort studies of the entire New Zealand working population were created using 1996, 2001 and 2006 censuses linked to mortality data. Mode of travel to work was that reported on census day, and causes of death examined were ischaemic heart disease and injury. Main analyses were Poisson regression models adjusted for socio-demographics. Sensitivity analyses included: additional adjustment for smoking in the 1996 and 2006 cohorts, and bias analysis about non-differential misclassification of cycling vs car use. RESULTS: Walking (5%) and cycling (3%) to work were uncommon. Compared with people reporting using motor vehicles to travel to work, those cycling had a reduced all-cause mortality (ACM) in the socio-demographic adjusted models RR 0.87 (0.77-0.98). Those walking (0.97, 0.90-1.04) and taking public transport (0.96, 0.88-1.05) had no substantive difference in ACM. No mode of transport was associated with detectable statistically significant reductions in cause-specific mortality. Sensitivity analyses found weaker associations when adjusting for smoking and stronger associations correcting for likely non-differential misclassification of cycling. CONCLUSIONS: This large cohort study supports an association between cycling to work and reduced ACM, but found no association for walking or public-transport use and imprecise cause-specific mortality patterns.

Health system costs for individual and comorbid noncommunicable diseases: An analysis of publicly funded health events from New Zealand.

BACKGROUND: There is little systematic assessment of how total health expenditure is distributed across diseases and comorbidities. The objective of this study was to use statistical methods to disaggregate all publicly funded health expenditure by disease and comorbidities in order to answer three research questions: (1) What is health expenditure by disease phase for noncommunicable diseases (NCDs) in New Zealand? (2) Is the cost of having two NCDs more or less than that expected given the independent costs of each NCD? (3) How is total health spending disaggregated by NCDs across age and by sex? METHODS AND FINDINGS: We used linked data for all adult New Zealanders for publicly funded events, including hospitalisation, outpatient, pharmaceutical, laboratory testing, and primary care from 1 July 2007 to 30 June 2014. These data include 18.9 million person-years and $26.4 billion in spending (US$ 2016). We used case definition algorithms to identify if a person had any of six NCDs (cancer, cardiovascular disease [CVD], diabetes, musculoskeletal, neurological, and a chronic lung/liver/kidney [LLK] disease). Indicator variables were used to identify the presence of any of the 15 possible comorbidity pairings of these six NCDs. Regression was used to estimate excess annual health expenditure per person. Cause deletion methods were used to estimate total population expenditure by disease. A majority (59%) of health expenditure was attributable to NCDs. Expenditure due to diseases was generally highest in the year of diagnosis and year of death. A person having two diseases simultaneously generally had greater health expenditure than the expected sum of having the diseases separately, for all 15 comorbidity pairs except the CVD-cancer pair. For example, a 60-64-year-old female with none of the six NCDs had $633 per annum expenditure. If she had both CVD and chronic LLK, additional expenditure for CVD separately was $6,443/$839/$9,225 for the first year of diagnosis/prevalent years/last year of life if dying of CVD; additional expenditure for chronic LLK separately was $6,443/$1,291/$9,051; and the additional comorbidity expenditure of having both CVD and LLK was $2,456 (95% confidence interval [CI] $2,238-$2,674). The pattern was similar for males (e.g., additional comorbidity expenditure for a 60-64-year-old male with CVD and chronic LLK was $2,498 [95% CI $2,264-$2,632]). In addition to this, the excess comorbidity costs for a person with two diseases was greater at younger ages, e.g., excess expenditure for 45-49-year-old males with CVD and chronic LLK was 10 times higher than for 75-79-year-old males and six times higher for females. At the population level, 23.8% of total health expenditure was attributable to higher costs of having one of the 15 comorbidity pairs over and above the six NCDs separately; of the remaining expenditure, CVD accounted for 18.7%, followed by musculoskeletal (16.2%), neurological (14.4%), cancer (14.1%), chronic LLK disease (7.4%), and diabetes (5.5%). Major limitations included incomplete linkage to all costed events (although these were largely non-NCD events) and missing private expenditure. CONCLUSIONS: The costs of having two NCDs simultaneously is typically superadditive, and more so for younger adults. Neurological and musculoskeletal diseases contributed the largest health system costs, in accord with burden of disease studies finding that they contribute large morbidity. Just as burden of disease methodology has advanced the understanding of disease burden, there is a need to create disease-based costing studies that facilitate the disaggregation of health budgets at a national level.

Socioeconomic and Tobacco Mediation of Ethnic Inequalities in Mortality over Time: Repeated Census-mortality Cohort Studies, 1981 to 2011.

BACKGROUND: Racial/ethnic inequalities in mortality may be reducible by addressing socioeconomic factors and smoking. To our knowledge, this is the first study to estimate trends over multiple decades in (1) mediation of racial/ethnic inequalities in mortality (between Maori and Europeans in New Zealand) by socioeconomic factors, (2) additional mediation through smoking, and (3) inequalities had there never been smoking. METHODS: We estimated natural (1 and 2 above) and controlled mediation effects (3 above) in census-mortality cohorts for 1981-1984 (1.1 million people), 1996-1999 (1.5 million), and 2006-2011 (1.5 million) for 25- to 74-year-olds in New Zealand, using a weighting of regression predicted outcomes. RESULTS: Socioeconomic factors explained 46% of male inequalities in all three cohorts and made an increasing contribution over time among females from 30.4% (95% confidence interval = 18.1%, 42.7%) in 1981-1984 to 41.9% (36.0%, 48.0%). Including smoking with socioeconomic factors only modestly altered the percentage mediated for males, but more substantially increased it for females, for example, 7.7% (5.5%, 10.0%) in 2006-2011. A counterfactual scenario of having eradicated tobacco in the past (but unchanged socioeconomic distribution) lowered mortality for all sex-by-ethnic groups and resulted in a 12.2% (2.9%, 20.8%) and 21.2% (11.6%, 31.0%) reduction in the absolute mortality gap between Maori and Europeans in 2006-2011, for males and females, respectively. CONCLUSIONS: Our study predicts that, in this high-income country, reducing socioeconomic disparities between ethnic groups would greatly reduce ethnic inequalities in mortality over the long run. Eradicating tobacco would notably reduce ethnic inequalities in absolute but not relative mortality.