Impact of socioeconomic status on acute myeloid leukemia outcomes in New Zealand’s universal health care Open Access

Ethnic and socioeconomic disparities affect the survival of patients with acute myeloid leukemia (AML).^1-7 The ethnic differences in survival outcomes are typically attributed to a variety of complex and interconnected biologic and socioeconomic factors.^8-12 Importantly, socioeconomic disparities have modifiable causes that can be altered by measures taken at the community level. It is currently unknown what factors are the most relevant for ethnic differences compared with the impact of specific socioeconomic and health care factors.

Most of the racial and ethnic disparities analyzed have been in patients from the United States.^1,4,5,13-15 The United States has a high population of racial/ethnic minority groups and has an insurance-based health care system. Conversely, many European countries have universal social health care coverage but low populations of ethnic minority groups.^2,16-22 Access to care in the United States is markedly affected by race, and there has been a focus on medical insurance and access to treatment as to why race/ethnicity and socioeconomic disparities affect patient outcomes.^5,23-27 However, Black patients continue to have worse outcomes even within clinical trials,¹³ highlighting the complex interplay between ethnic and socioeconomic factors in AML outcomes.

Uniquely, New Zealand has universal health care coverage and a large population of ethnic minority Māori and Pacific Island (Polynesian) people. Polynesian peoples comprise 25.5% (16.5% Māori, 9.0% Pacific Peoples) of the total New Zealand population of ∼5 million.²⁸ Significant health inequities exist between New Zealand Polynesians and Europeans, and Polynesians have poorer survival in many cancers, including leukemia.^29,30 Because New Zealand has a unique combination of universal health care coverage and a large ethnic minority population, we took the opportunity to identify the effects of ethnicity and socioeconomics on the incidence and overall survival of adult patients with AML. Our analysis aimed to uncouple the independent effects of socioeconomic and ethnic factors on the outcomes of adult patients with AML.

The study population consisted of all adult patients (aged ≥20 years) with AML registered with the New Zealand Cancer Registry (NZCR) between 1 January 2000 and 31 December 2019.³¹ The NZCR is a population-based register of all new malignant tumors diagnosed in New Zealand. The registry follows the guidelines for recording and reporting cancer incidence recommended by the International Agency for Research on Cancer and the International Association of Cancer Registries. The primary function of the NZCR is to accumulate information on the incidence and mortality of cancer to provide a basis for cancer research. Our previous work demonstrated that the NZCR is a reliable hematologic cancer registry consistent with international registries and a valuable tool for monitoring the incidence and mortality of blood cancers in New Zealand.³¹ 

The study procedures were approved by the Auckland District Health Board research review committee (approval number A+7753). Deidentified data were obtained for each patient, including ICD-10 (International Statistical Classification of Diseases and Related Health Problems Tenth Revision) code, morphology code, date of diagnosis, age at diagnosis, date of death, sex, prioritized ethnicity code, District Health Board at which patients were diagnosed, and the domicile code. The NZCR records self-identified ethnicity and allows for multiple responses. Patients reporting >1 ethnicity are coded as a single ethnicity using the following prioritization system: Māori > Pacific > Asian > Other. All ethnicities were considered in this study, but the small numbers in some groups limited meaningful analyses; therefore, most comparisons were made between patients identifying as European, Māori, or Pacific. To account for the low numbers of Māori and Pacific patients in certain subgroups, and because findings for these ethnicities were similar, Māori and Pacific patients were combined into a single Polynesian group. Separated data for these ethnicities can be found in the supplemental Table 3 or requested from the corresponding author.

Socioeconomic deprivation was measured using the New Zealand Index of Deprivation (NZDep) 2018, which was assigned to patients based on their area of residence at diagnosis.³² The index allows for the comparison between deprivation and other variables over time.^32,33 NZDep2018 divides New Zealand's small geographic areas into deciles based on 8 dimensions of deprivation: communication, income, employment, qualifications, home ownership, social support, living space, and living conditions. The key variables within these dimensions, in order of decreasing weight, include: individuals without home internet access, recipients of means-tested benefits (aged 18-64 years), individuals living below an income threshold (after adjusting for household size and composition), unemployed individuals (aged 18-64 years), individuals without qualifications (aged 18-64 years), people not owning their home, individuals living in single-parent families (aged <65 years), individuals living below a bedroom occupancy threshold, and individuals residing in dwellings with persistent dampness and/or mold larger than A4 size. The NZDep index assigns a deprivation score to statistical areas typically comprising 100 to 200 people. The ordinal scale ranges from decile 1 (representing the 10% of areas with the least deprivation) to decile 10 (representing the 10% of areas with the highest deprivation).³² Patients were grouped into 2 deprivation categories: less deprivation (deciles 1-7) and higher deprivation (deciles 8-10), as reported by Belyaev et al.³³ 

Crude and age-standardized capture rates were calculated using New Zealand’s estimated resident population at 30 June 2018 as the denominator population.³⁴ Age-standardized rates were calculated using the 2000 to 2025 World Health Organization world population standard and the 2001 census total Māori population standard.³⁵ Comparisons were made between Māori and non-Māori rates; more detailed population data separating other ethnicities were unavailable for the entire study period. A χ² test was used to compare proportions between European and Polynesian patients with AML and the general population between each decile group and age decade using data from census 2018 and New Zealand’s estimated resident population at 30 June 2018.^34,36 

Data were analyzed using SPSS Statistics software version 29.0. Descriptive statistics were compared with the χ² and Fisher exact tests for categorical variables, and the Mann-Whitney U test for continuous variables. Kaplan-Meier curves were generated to measure overall survival from the date of diagnosis to the extraction date (29 October 2021); groups were compared using the log-rank test. Cox proportional hazards regression analysis was used to generate multivariate models of survival hazards. Patient age at diagnosis, sex, disease type, ethnicity, and deprivation group were included as possible predictors of overall survival. The level of statistical significance was defined as P < .05.

Between 1 January 2000 and 31 December 2019, 3500 adult patients with AML were reported to the NZCR (Table 1; supplemental Tables 1 and 2). Most patients were European (76.5%), with smaller populations of Māori (11.3%) and Pacific Peoples (6.2%), roughly proportional to the make-up of the New Zealand population.²⁸ Overall, male patients were more common than female patients; however, female patients were slightly more common in the Polynesian group (P = .049; Table 1). This higher percentage of female patients was primarily driven by Pacific Peoples (52.8%), with Māori (49.1%) and European (46.0%) female patients showing lower frequencies (supplemental Table 3).

Most (64.4%) European patients were aged >65 years at diagnosis, with a median age of 71.0 years (Table 1). In contrast, Polynesian patients were diagnosed at a younger median age of 57.0 years (P < .001) and most (67.7%) were aged between 20 and 65 years. The lower median age of Polynesian patients compared with European can be explained, in part, by the age distribution of these populations in New Zealand as a whole (Table 2). New Zealand Europeans aged ≥65 years, make up 18.7% of the European population, whereas Māori and Pacific peoples aged ≥65 years, only make up 6.2% and 5.2% of their respective populations.²⁸ 

The proportions of patients with AML living in areas of higher deprivation (deciles 8-10) were higher for Polynesian than European patients (61.9% vs 27.0%, P < .001; Table 1; Figure 1A). This pattern resembles the general New Zealand population, in which Polynesians are also more likely than Europeans to live in areas of higher deprivation (59.2% vs 24.3%, P < .001; Figure 1B).

Other ethnicities accounted for <5% of patients each (supplemental Table 2). Asians were the most common (4.9%), including 1.8% Chinese and 1.4% Indians. Asian data were analyzed, but results should be interpreted cautiously because of small numbers. Asian patients with AML were younger at diagnosis than Europeans (median age of 59.0 years for Chinese, 64.0 for Indians, and 71.0 for Europeans (P < .001; supplemental Table 4). This difference in age could also be explained by the overall characteristics of these populations. New Zealand Asians have the age distribution profile of a relatively recent migrant population, with 41.3% aged between 20 and 39 years and only 6.5% are aged ≥65 years.²⁸ The percentage of Asian patients living in the most deprived areas (26.6%) was comparable with Europeans (27.0%), although this was lower for Chinese (21.0%) and higher for Indians (37.5%; supplemental Table 4). Acute promyelocytic leukemia (APL) was more common in Asian patients (12.2% overall, 12.5% in Chinese, and 12.2% in Indians) compared with 5.0% in Europeans and 7.3% in Polynesians (P < .001; Table 1; supplemental Table 4). Patients with other ethnicities were too few to meaningfully analyze (supplemental Table 2).

The crude capture rate for AML between 1 January 2000 and 31 December 2019 was 5.62 cases per 100 000 people per year. This was higher for non-Māori (5.70) than Māori patients (5.09; P < .05; Table 3). However, Māori patients had higher age-standardized capture rates than non-Māori patients when adjusted using either the Māori population standard (3.13 vs 2.49; P < .05; Table 3) or the World Health Organization standard (3.64 vs 3.21; P < .05; data not shown). The incidence of AML in Pacific Peoples or other ethnic minorities was unable to be calculated because the Māori population was the only ethnicity separately recorded for the entire study period. The capture rates of AML were higher for patients who lived in the areas with higher deprivation (deciles 8-10) than those living with less deprivation (deciles 1-7; 2.70 and 2.23, respectively; P < .05; Table 3).

Without age stratification, Polynesian and Asian patients had significantly better overall survival than Europeans (P < .001; supplemental Figure 1A). However, when split by age groups (between 20-65 and >65 years), there were no statistically significant differences in survival outcomes (supplemental Figure 1B-C). European, Māori, and Pacific patients aged 20 to 65 years had equivalent survival, but for those aged >65 years, Pacific patients survived better (supplemental Figure 1D-F). Asian patients, in particular Chinese, appeared to survive better than European (supplemental Figure 1G-I), which may be, at least partially, because of their younger age at diagnosis and higher rates of APL. Further analyses of Asian patients were not informative because of their small numbers.

When stratified by age at diagnosis, the overall survival of European and Polynesian patients was worse for each decade of life (P < .001; Figure 2A-B). Compared with Europeans, there was a sharp survival decline for Māori patients in their 40s (Figure 2C) and for Pacific patients in their 30s and 40s (Figure 2D). When patients were compared by decade-of-age at diagnosis, Polynesian patients had significantly worse survival at ages 30 to 39 and 40 to 49 years than European patients (P < .05; supplemental Figure 2). For all other decade-of-age at diagnosis groups, there was no significant difference in survival outcomes between European and Polynesian patients (supplemental Figure 2).

To determine the impact of deprivation on AML outcomes, we compared the overall survival time of European and Polynesian patients living with different levels of deprivation (Figure 3; supplemental Figure 3). We found that when younger (aged 20-65 years) European patients were split into deprivation quintiles, the less deprived a patient was, the higher the survival (P = .049; supplemental Figure 3A). Older (>65 years) European patients with less deprivation also had significantly higher survival (P = .007; supplemental Figure 3B). The graduated impact of deprivation was not the same for Polynesian patients, but younger Polynesian patients (aged 20-65 years) living with the least deprivation (quintile 1) had much higher survival rates than those living with the most deprivation (quintile 5; supplemental Figure 3C). A graduated trend was also observed for older Polynesian patients (aged >65 years), except for those living in quintile 1 (n = 8 only; supplemental Figure 3D).

Because most Polynesian patients live in areas with high deprivation (deciles 8-10; 61.9%; Table 1), we compared patients living with higher deprivation (deciles 8-10) and less deprivation (deciles 1-7; Figure 3A). We found that both European and Polynesian younger patients (aged 20-65 years) living with higher deprivation (deciles 8-10) had significantly worse survival than those living with less deprivation (deciles 1-7; Figure 3Ai-ii). Older European and Polynesian patients (aged >65 years) living with higher deprivation (deciles 8-10) also had worse survival than those living with less deprivation (deciles 1-7), but this only reached significance for European patients (P = .015, n = 1717; Figure 3Aiii), likely, in part, because of low numbers of Polynesian patients (P = .193, n = 196; Figure 3Aiv).

To determine whether the impact of deprivation persists in the context of more targeted treatments, we examined the survival of patients APL and non-APL separately (Figure 3B-C). We found that higher deprivation deciles (NZDep 8-10) were associated with worse survival for both APL and non-APL, and this effect was evident in both European and Polynesian patients (Figure 3B-C). Small numbers of Polynesian patients with APL likely limited statistical significance in that group (P = .059; Figure 3Bii).

We then analyzed the impact of deprivation on the overall survival across the decade-of-age at diagnosis. Higher deprivation deciles (NZDep 8-10) were associated with poorer outcomes for patients aged 30-39 (P = .002), 60-69 (P = .015), and 70-79 (P = .029) years, with no statistically significant effect detected for other age groups (data not shown).

Because we had found that age, ethnicity, and deprivation all affected the survival outcomes of patients with AML, we used Cox regression analysis to separate these factors (Table 4). The risk factors analyzed were age at diagnosis, sex, APL or non-APL disease, ethnicity, and deprivation. Multivariate Cox regression analysis revealed that when deprivation was not included as a variable, ethnicity was a significant risk factor (P = .003). However, when deprivation was included, the impact of ethnicity was no longer statistically significant (P = .063). The independent risk factors for survival included socioeconomic deprivation (P = .001), non-APL disease (P < .001), and older age at diagnosis (P < .001; Table 4).

Improving outcomes of patients with AML is a strong focus of clinical and research hematologists. As part of this goal, it is necessary to better understand the interplay between ethnic and socioeconomic factors to inform an actionable plan, reduce disparities and improve patient outcomes across all societal groups.

In our study on adult patients with AML in the context of universal health care coverage in New Zealand, we found that socioeconomically deprived patients with AML had worse overall survival, and this affected both European and Polynesian (Māori and Pacific) patients. A high proportion of Polynesian patients with AML reside in the most deprived areas and therefore are most affected by this risk factor. Multivariate analysis revealed that when deprivation was not included as a variable, ethnicity was a significant risk factor. However, when deprivation was included, ethnicity was no longer a statistically significant risk. These findings point to the need of a modified approach when managing patients that are socioeconomically disadvantaged.

The age-adjusted incidence of AML was significantly higher for patients living with higher deprivation than those with less deprivation and higher for Māori than non-Māori patients, raising the possibility that socioeconomic deprivation plays a role in AML development. Similar was found for patients in the United Kingdom,²² but in the United States, the age-adjusted incidence was highest for White patients, and lower for Hispanic, Black, Asian, and Pacific Island patients.¹⁵ We could not find published data on the effect of deprivation on AML incidence in the United States.

Risk factors for developing AML include smoking, obesity, exposure to chemicals or radiation, and cancer history and treatment.^37,38 Data for these risk factors were not available for our study cohort but are known to be ethnically and socioeconomically patterned.³⁷ Smoking and obesity are also predictors of poorer survival.^37-42 Smoking rates are ∼2 to 3 times higher in Polynesians than other New Zealanders, as well as ∼3 times higher in the most deprived neighborhoods.⁴³ Similarly, obesity rates are higher for Polynesians and those living in the most deprived neighborhoods.^43-46 

The most striking finding of our study is the strong impact that deprivation has on overall survival of patients with AML. Surprisingly, this also included APL patients. Our previous research found that ethnicity did not affect survival outcomes between Polynesian and European patients with APL.⁴⁷ This follows a worldwide trend of APL being well-managed with a specific treatment regimen incorporating all-trans retinoic acid and arsenic trioxide.^48-52 However, we found that deprivation did affect APL survival outcomes for both European and Polynesian patients. Thus, it will be important to identify which specific deprivation-related factors have contributed to this worse survival, so they can be better addressed.

New Zealand Māori have worse cancer survival in many cancer types than non-Māori in each deprivation quintile.²⁹ Gurney et al suggest this may be because of the delay in diagnosis, whereas treatment may not meaningfully differ between Māori and non-Māori patients.²⁹ Differences in the time of detection between ethnic groups are unlikely to be crucial in AML, because most patients present acutely. Treatment for AML in New Zealand has been relatively uniform during the study period given our universal health care coverage. The mainstay of AML therapy consisted of 7+3 chemotherapy backbone for fit patients without comorbidities. In the last decade, addition of immunotherapy or targeted treatment were available through UK National Cancer Research Institute AML trials, which were conducted in all New Zealand centers. For patients unfit for intensive induction chemotherapy, treatment consisted of either azacitidine for those with low blast counts or low-dose cytarabine for other patients. Novel agents were not funded in New Zealand during the study period. Anecdotally, only a very small proportion of patients with AML decline treatment or clinical trial participation. Nevertheless, comorbidities such as cardiac, pulmonary, renal, or hepatic diseases, which are intertwined with deprivation⁵³ and more prevalent in Polynesian patients,^54-58 may have influenced the intensity and tolerance of chemotherapy. The NZCR does not provide information about comorbidities or the treatment received, so we could not investigate these issues further.

The overall survival from AML can be affected by patient access to hematopoietic stem cell transplantation (HSCT). A local study from 1999 showed that Polynesian patients had a reduced chance of finding a 6/6-matched unrelated donor on international registries.⁵⁹ This has likely improved later via establishing the New Zealand Bone Marrow Donor Registry, although there are no published data to support it. Nevertheless, socioeconomic deprivation did not affect access to, or survival after, HSCT for 851 adult patients treated in Auckland between 2010 and 2020.³³ The authors suggested that the equal survival between socioeconomic groups was partly because of a comprehensive multidisciplinary support team available to transplant recipients, including pharmacists, social workers, nurse practitioners, and Polynesian support workers.³³ Future research could investigate the survival benefits to patients with AML with a similarly comprehensive support team.

It would also be useful to prospectively gather individual patient level data on factors affected by deprivation, including comorbidities (especially obesity, smoking, cardiac, pulmonary, renal, and hepatic diseases), treatment type, tolerability and adherence, transport accessibility, health literacy and education, occupation and income level, and household social support. An immediate actionable change may involve designing a standardized questionnaire to incorporate such data in clinical trials, aiming to make it a component of standard of care for all patients. It will be important to understand how to frame and deliver such a questionnaire, so that data are accurately recorded in different settings and collected in a culturally appropriate and sensitive manner. Currently, social determinants of health and even race and/or ethnicity are not routinely reported from clinical trials. Only 67% of trials leading to drug approval by the US Food and Drug Administration for hematological therapies between 2011 and 2021 reported on race.⁶⁰ The Children’s Oncology Group diversity and health disparities committee has prioritized a standardized collection of demographics and social determinants of health data from pediatric patients.⁶¹ The process has been shown to be feasible, with one study showing an 87.5% parent opt-in rate.⁶² Similar initiatives are needed for adult patients with AML.

The limitations of our study include using the NZDep index to measure socioeconomic deprivation. Although the local geographic areas in which people live reflect the probable sociodemographic environment, the index does not identify deprivation at the individual level. Other limitations include the retrospective design, lack of information on genetic risk and biologic characteristics of the leukemia, patient comorbidities, treatment received including HSCT, and the causes of death; therefore, multivariate analysis could not identify all confounders. The strength of our work is that we analyzed a large, relatively uniform ethnic population under the care of a universal public health care system delivering uniform standard-of-care treatment to patients with AML.

In conclusion, we identified the strong impact of socioeconomic deprivation on the overall survival of adult patients with AML within New Zealand’s universal health care system. Our results highlight the need for an in-depth investigation into the mechanisms behind socioeconomic disparities in AML incidence and outcomes. Future research should identify specific risk factors contributing to survival differences due to socioeconomic deprivation. Clinical systems must adapt to mitigate deprivation-related factors when possible, to help improve patient outcomes.

Chris Lewis, an information analyst for the New Zealand Ministry of Health, extracted patient data from the New Zealand Cancer Registry.

This work was supported by funding from the Bone Marrow Cancer Research Trust (Christchurch; UoA# 3720536), Cancer Research Trust (CRTNZ 1922 PG), and donations from the Norman Family (UoA# 3715253) to M.L.K.-Z. A.V.H. and S.P. received Summer Research Scholarships from the University of Auckland.

Contribution: T.I. analyzed data and wrote the manuscript; T.N.G. conducted and supervised statistical analyses and contributed to data interpretation and writing of the manuscript; A.V.H. and S.P. performed statistical analyses; N.C. helped interpret data and write the manuscript; M.L.K.-Z. designed the study and contributed to data analysis, interpretation, and writing of the manuscript; and all authors have read and agreed to the final version of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Maggie L. Kalev-Zylinska, Department of Molecular Medicine and Pathology, Faculty of Medical and Health Sciences, University of Auckland, 85 Park Rd, Grafton, Auckland 1023, New Zealand; email: [email protected].