Regional and ethnic projections of gastric cancer incidence in Aotearoa New Zealand to 2045: identifying opportunities for targeted action

Gastric cancer (stomach cancer) is an important contributor to cancer morbidity and mortality in Aotearoa New Zealand.^1–3 It is a top 10 cause of cancer death among Māori, with 5-year survival rates of around 27%.⁴ Although national incidence has declined over recent decades, substantial differences persist between population groups. Māori and Pacific peoples experience higher incidence than the European/Other group, with some reports indicating three to six times higher rates.^2,5 These differences reflect variation in exposure to Helicobacter pylori (H. pylori), dietary and environmental factors, and broader structural conditions that influence access to timely diagnosis and treatment.^6,7 Although gastric cancer represents a smaller proportion of total cancer burden, its high case fatality, marked ethnic inequities and the presence of a well-established and modifiable causal pathway through H. pylori make it a priority for targeted prevention and service planning.^2,5,6,8,9

H. pylori infection is typically acquired in childhood, and household crowding, which is more prevalent in Pacific communities, is an important driver of transmission, contributing to the high infection prevalence observed in this population.¹⁰ In addition, inherited cancer susceptibility plays a role in a subset of cases. Pathogenic germline mutations in the CDH1 gene, which cause hereditary diffuse gastric cancer, are over-represented among Māori whānau and contribute to early onset diffuse gastric cancer and elevated lifetime risk.^11,12 Recognition of both acquired and inherited risk factors is therefore important for understanding the observed burden of gastric cancer and its unequal distribution in the New Zealand population.

The distribution of tumour sub-types differs across population groups. Diffuse-type gastric cancer, often associated with H. pylori, is more common among Māori and Pacific peoples than among New Zealand Europeans.^6,7 Māori and Pacific populations are also diagnosed at a younger age and experience poorer survival outcomes that are shaped by structural inequities including socio-economic deprivation and inequitable access to care.^2,5,6,9,13 Although gastric cancer carries high case fatality and generally poor prognosis,⁶ it is often overlooked in national cancer control discussions due to the relatively low annual case numbers compared to other cancers. These low annual case numbers, geographic dispersal and regional variation in surgical volumes also create unique challenges for care delivery and outcomes.^14,15

Gastric cancer often requires intensive treatment, including surgery, chemotherapy and prolonged follow-up.¹⁶ Even a modest increase in absolute case numbers has implications for diagnostic pathways, endoscopy services, surgical oncology and supportive care. Regional variation in population growth and demographic ageing will also influence future demand.

Age-period-cohort (APC) modelling provides a method for projecting future cancer incidence by drawing on historical trends and future population projections.¹⁷ There are no published projections of gastric cancer incidence in New Zealand that present results disaggregated by ethnicity and region, and existing work is limited to national-level estimates without ethnic or regional breakdown.¹⁸ This paper contributes new insights by presenting national and regional projections of gastric cancer incidence in New Zealand through to 2045. Incidence is stratified by ethnicity, sex and Health New Zealand – Te Whatu Ora region to provide a more detailed picture of expected trends to support planning and wider cancer control activities.

Methods

Data sources

National cancer registry data for New Zealand were used to model trends in gastric cancer incidence from 2001 to 2022 and to generate projections through to 2045. The dataset comprised all primary registrations of gastric cancer (International Classification of Diseases, 10th revision [ICD-10] code C16), stratified by calendar year, 5-year age group, sex, prioritised ethnicity (Māori, Pacific peoples, Asian and European/Other) and Health New Zealand – Te Whatu Ora region. Population estimates and projections were sourced from Statistics New Zealand and aligned to the same demographic and geographic groupings (Statistics New Zealand, customised report, licensed for re-use under the Creative Commons Attribution 4.0 International licence).

Modelling approach

Incidence was projected using a generalised linear modelling framework with a Poisson distribution and log link. The outcome variable was the count of incident gastric cancer cases within strata defined by calendar year, 5-year age group, sex, prioritised ethnicity and Health New Zealand – Te Whatu Ora region. The natural logarithm of the corresponding population denominator was included as an offset, such that the model estimates stratum-specific incidence rates on the log scale.

Age group was included as a categorical variable. Calendar year and birth cohort were included as linear terms to provide a simplified APC structure. Birth cohort was defined as calendar year minus the start age of the relevant 5-year age group. Sex, prioritised ethnicity and region were included as categorical main-effect covariates. No interaction terms were included between age, period, cohort, sex, ethnicity or region. As such, the groups share a common underlying linear temporal pattern, while differences between groups represent multiplicative shifts in baseline incidence on the log scale. Projected differences in age-standardised rates (ASRs) and case numbers over time therefore arise primarily from differences in baseline risk and from changes in population size and age structure, rather than from group-specific temporal trend effects.

The modelling framework draws on APC principles used in cancer incidence projections and is conceptually aligned with the approach applied in the Nordpred model.¹⁷ In contrast to the Nordpred model, which estimates a separate drift parameter representing the overall log-linear trend and applies attenuation to this component in future periods, the present model incorporated calendar year directly as a linear term within a Poisson regression framework. Under this specification, the linear period term captures the overall temporal trend without estimating a distinct drift parameter. No explicit trend dampening was applied. This approach was selected to provide a simplified and internally consistent representation of temporal change across demographic and regional strata where some counts can be small.

To place greater emphasis on recent incidence patterns while retaining information from earlier years, time-based recency weights were applied. Calendar years were assigned increasing weights across the observed period, such that more recent years contributed more strongly to parameter estimation. Weights were normalised to have mean 1 across the observed period and were incorporated directly into model fitting. All analyses were undertaken in R (version 4.4.3, R Core Team, 2024, R Foundation for Statistical Computing).

Bootstrap procedure and projection uncertainty

Uncertainty in projections was quantified using non-parametric bootstrapping with 1,000 iterations. Within each iteration, strata defined by age group, sex, ethnicity and region were resampled with replacement separately within each calendar year. This stratified resampling preserves the overall period structure while allowing sampling variability across demographic and regional strata within year.

Within each bootstrap replicate, historical population denominators were perturbed multiplicatively to represent uncertainty in population estimates. Specifically, observed population counts were multiplied by a normally distributed random variable with mean 1 and standard deviation 0.01. Resulting values were constrained to be at least 1 to avoid non-positive denominators. The Poisson model was refitted to each bootstrap sample using the recency weights described above.

Future population denominators were similarly perturbed multiplicatively using normally distributed random noise with mean 1 and year-specific standard deviations to reflect increasing uncertainty over the projection range. Standard deviations were set to 0.005 for 2023 to 2025, 0.010 for 2026 to 2030, 0.015 for 2031 to 2035, 0.020 for 2036 to 2040 and 0.025 for 2041 to 2045.

Projections were generated through to 2045. For each bootstrap replicate, predicted log incidence was obtained from the fitted model, including the population offset, and transformed to predicted case counts on the original scale. Projected case numbers were summarised across bootstrap replicates using the median as the point estimate. Ninety-five percent uncertainty intervals (UIs) were defined by the 2.5th and 97.5th percentiles of the bootstrap distribution.

Age-standardised rates (ASRs)

ASRs were calculated for both observed and projected data for the periods 2020–2022, 2035 and 2045, stratified by Health New Zealand – Te Whatu Ora region, sex and prioritised ethnicity, using the 2001 World Health Organization world standard population. For observed data, 95% confidence intervals (CIs) were calculated using standard methods for directly standardised rates, while for projected data, 95% UIs were derived from the distribution of ASRs across bootstrap replicates.

Study approval

This study was considered low risk and was out of scope for Health and Disability Ethics committee review as it used de-identified administrative health data. Locality authorisation including Māori research review was granted by Health New Zealand – Te Whatu Ora Waitematā District, Research & Knowledge Centre (approval code: WAI20420).

Results

National and regional projections

Nationally, gastric cancer cases are projected to increase from around 492 per year in 2020–2022 to 725 by 2045 (95% UI 612–860), a 48% rise. The ASR is projected to decrease from 5.9 per 100,000 (95% CI 5.6–6.2) to 5.3 (95% UI 4.3–6.5), a 10% reduction. Female cases increase from 185 to 263 and male cases from 307 to 461, with ASRs declining in both groups. Cases among those aged ≥75 years accounted for 30% of all cases (159 per year) in 2020–2022, increasing to around 43% (270 cases per year; 95% UI 235–311) by 2035 and nearly 50% (354 cases per year; 95% UI 294–427) by 2045.

In the Northern Region, cases are projected to rise from 195 to 306 by 2045 (95% UI 257–364), a 57% increase. The ASR decreases from 6.9 (95% CI 6.3–7.5) to 6.0 (95% UI 4.9–7.4). Female cases rise from 77 to 112 and male cases from 118 to 194, with ASRs declining in both groups.

In Te Manawa Taki, cases increase from 103 to 150 by 2045 (95% UI 126–180), a 46% rise. The ASR falls from 6.2 (95% CI 5.5–6.9) to 5.3 (95% UI 4.3–6.6). Female cases rise from 40 to 55 and male cases from 64 to 95, with ASR reductions for both sexes.

In the Central Region, cases are projected to grow from 83 to 118 by 2045 (95% UI 99–142), a 42% increase. The ASR declines slightly from 4.9 (95% CI 4.3–5.6) to 4.8 (95% UI 3.9–6.0). Female cases rise from 31 to 43 and male cases from 53 to 75, with stable or declining ASRs.

In Te Waipounamu, cases rise from 110 to 150 by 2045 (95% UI 125–178), a 36% increase. The ASR decreases from 5.1 (95% CI 4.6–5.8) to 4.4 (95% UI 3.6–5.5). Female cases rise from 38 to 53 and male cases from 72 to 96, with reductions in ASRs for both groups.

View Table 1–2.

Ethnic-specific projections

Among Māori, gastric cancer cases are projected to increase from 84 in 2020–2022 to 165 by 2045 (95% UI 141–198), a 96% rise. The ASR is projected to decline from 10.7 (95% CI 9.4–12.1) to 9.8 (95% UI 8.0–12.0). In both sexes, case numbers increase but ASRs decrease, and Māori continue to experience among the highest rates across all ethnic groups.

For Pacific peoples, cases are projected to rise from 45 to 87 by 2045 (95% UI 71–104), a 93% increase. The ASR declines from 14.4 (95% CI 12.1–17.1) to 12.2 (95% UI 9.6–14.9). Female and male case numbers both increase, although projected rates remain higher for males.

Among those identified in the Asian ethnic group, case numbers rise from 54 to 161 by 2045 (95% UI 133–196), a 198% increase. The ASR declines from 6.7 (95% CI 5.7–7.8) to 5.1 (95% UI 4.1–6.4). Increases in cases for both sexes are driven largely by demographic growth rather than rising underlying risk.

For European/Other populations, case numbers remain stable, from 309 in 2020–2022 to 311 by 2045 (95% UI 261–370). The ASR declines from 4.6 (95% CI 4.3–5.0) to 3.6 (95% UI 2.9–4.4). Patterns for females and males are consistent with future demographic growth.

Discussion

This study provides national and regional projections of gastric cancer incidence in New Zealand through to 2045, disaggregated by sex, prioritised ethnicity and Health New Zealand – Te Whatu Ora region. Although ASRs are projected to decline, absolute case numbers are expected to rise across most population groups over the next two decades, driven primarily by demographic change, particularly population ageing. The projections therefore largely reflect underlying demographic processes; however, these processes interact with existing inequities in incidence and population structure to shape the future distribution of disease burden.

The projected trends show a persistent and unequal burden across ethnic groups. Māori and Pacific peoples have both the highest current incidence and a large proportional increase in projected case numbers by 2045, reflecting higher baseline risk alongside more rapid population growth and ageing. Although ASRs decline for all groups, these reductions are insufficient to offset demographic pressures for Māori, Pacific and Asian populations. In contrast, the European/Other population shows relatively stable case numbers and lower rates, although the absolute burden remains higher due to population size. Regional variation reflects population growth and demographic composition, with the Northern Region experiencing the largest increase in case numbers.

H. pylori testing, treatment and opportunities for prevention

H. pylori infection is a major causal factor for non-cardia gastric cancer and is a key contributor to the higher incidence observed among Māori and Pacific peoples in New Zealand.^1,19,20 International evidence indicates that long-term declines in H. pylori prevalence have been central to reducing incidence in many settings.^21,22 However, in New Zealand, in addition to barriers to accessing primary care, access to symptomatic H. pylori testing and treatment is inequitable. Local research has demonstrated higher infection prevalence among Māori, Pacific peoples and some Asian populations, and has also reported lower treatment and post-treatment clearance among Māori and Pacific peoples.^10,23,24

Global modelling suggests that a substantial proportion of future gastric cancer cases may be attributable to persistent H. pylori infection and therefore potentially preventable through effective detection and eradication strategies.²¹ These findings are consistent with the patterns observed in New Zealand and reinforce the relevance of population-based or targeted H. pylori test-and-treat approaches,¹ as articulated in the new International Agency for Research on Cancer (IARC) H. pylori test-and-treat recommendations.²⁵ A national H. pylori seroprevalence study has recently been completed²⁶ and will provide an updated evidence base to inform the development of targeted prevention strategies in the New Zealand context.^6,27

While H. pylori is a large driver of population-level risk, inherited susceptibility contributes to a subset of cases. Pathogenic CDH1 variants, which are over-represented among some Māori whānau, are associated with early-onset diffuse gastric cancer.^11,12 Evidence suggests that H. pylori infection may further increase risk among genetically susceptible individuals.²⁸ Recognition of hereditary diffuse gastric cancer is relevant for targeted genetic testing and management and reinforces the importance of H. pylori detection and eradication in high-risk individuals. However, this does not alter the central role of H. pylori–focussed prevention strategies at the population level.

Test-and-treat strategies represent an opportunity to reduce future incidence and are likely to be cost effective, particularly in higher-risk populations.²⁹ Development of a test-and-treat programme, as recommended in the New Zealand Cancer Action Plan 2019–2029,³⁰ should consider appropriate test modalities, culturally safe delivery and clear pathways to diagnosis and treatment.³¹ An equity-focussed approach is required to address system-level barriers including cost, access and the delivery of culturally appropriate care. Antimicrobial stewardship, including antibiotic resistance, also requires consideration, particularly in higher-prevalence populations where prior treatment exposure may affect success.³²

Given that most Māori and Pacific peoples access care through general practice, rather than through Māori- or Pacific-specific providers, equitable delivery will require that all primary care services are supported to provide effective and culturally appropriate H. pylori testing and treatment. Existing programmes such as the National Bowel Screening Programme may offer future opportunities for integrated stool-based testing, as trialled in Taiwan,³³ although this would require careful consideration given current inequities in screening uptake.

Implications for service planning

The projected increase in gastric cancers will place additional demand on diagnostic and treatment services. In addition, the increasing proportion and number of cases among those aged ≥75 years has important implications for clinical management and service delivery, given the greater complexity of care and higher burden of comorbidity in older populations. Despite declining incidence rates, rising absolute case numbers are expected to increase demand for gastroscopy, pathology, imaging, surgical care and peri-operative support. Ensuring adequate regional diagnostic capacity is important given existing constraints in endoscopy services. Increased use of diagnostic endoscopy may also lead to greater detection of early-stage disease, creating opportunities for minimally invasive treatment. For some adenocarcinomas, endoscopic resection is considered curative and may reduce the need for formal gastrectomy.³⁴ An increase in early-stage diagnoses may therefore shift aspects of treatment demand from major surgery towards advanced therapeutic endoscopy. This will have corresponding workforce and training implications.

New Zealand is a low-volume setting for gastric cancer surgery, with variation in resection volumes across centres.¹⁵ Low case numbers create challenges for maintaining specialist expertise and consistent access to advanced operative techniques. As case numbers increase, strengthening referral pathways, improving co-ordination between primary and secondary care, and ensuring timely access to diagnosis and treatment will be increasingly important. Improving access to evaluation for dyspeptic symptoms and reducing referral delays may support earlier diagnosis and improved outcomes. Implementation of updated guidance on post-treatment H. pylori clearance testing will also support appropriate antimicrobial stewardship and treatment selection.

Recent commentary has called for New Zealand to adopt the internationally recognised model of comprehensive cancer centres, integrating multidisciplinary clinical care, research, education and regional networks.³⁵ This approach has been proposed in response to rising demand, capacity constraints and workforce pressures, and may offer advantages in low-volume settings by promoting consistent clinical pathways and consolidating specialist expertise. Such models may reduce unwarranted variation in care and strengthen regional service integration. However, international evidence, largely from the United States of America, indicates that centralised specialist cancer care can exacerbate inequities when access is uneven, with lower utilisation among rural populations and people living in areas of high socio-economic deprivation.^36–39 These patterns have been linked to factors such as travel distance, referral pathways and access to specialist navigation, and may contribute to later stage at diagnosis.⁴⁰ In the context of New Zealand’s publicly funded and geographically dispersed health system, comprehensive cancer centres would require substantial investment and careful design. If pursued, equity would need to be embedded in design and implementation to ensure that centralisation improves, rather than widens, access to high-quality cancer care for populations already facing barriers in access to care.

Equity and system alignment

The projected increases among Māori and Pacific peoples highlight the importance of equity in designing appropriate solutions. Previous work has identified persistent barriers to timely diagnosis and treatment for Māori, including delays in referral, fragmented pathways, limited cultural safety and experiences of racism within the health system.^41,42 As with other cancers, strengthened navigation roles, improved communication and whānau-centred care models may support earlier engagement and reduce avoidable delays.

Alongside H. pylori-focussed prevention, population-level action to reduce exposure to other established gastric cancer risk factors including tobacco use,⁴³ harmful alcohol consumption⁴⁴ and excess body weight⁴⁵ remains relevant, particularly where these exposures are socially patterned and contribute to inequitable risk. Addressing these determinants requires sustained public health and regulatory approaches rather than reliance on individual behaviour change alone.

These findings should be interpreted within the broader context of health system reform in New Zealand. As regional and district service networks are strengthened, planning for future gastric cancer care must acknowledge the differential burden and ensure appropriately tailored responses that adequately meet the needs of populations experiencing the highest incidence and mortality. Embedding equity considerations within prevention, diagnostic pathways and specialist services is essential to avoid perpetuating or widening existing disparities.

Strengths

This study provides the first projections of gastric cancer incidence in New Zealand incorporating APC effects, with results presented by sex, prioritised ethnicity and region. The use of a bootstrap approach allows clear representation of statistical uncertainty. Producing disaggregated projections offers value for both national and regional planning and supports equity-focussed prioritisation.

Limitations

Interpretation of the projections should consider several limitations. The modelling assumes continuation of recent incidence trends and does not incorporate potential future changes in H. pylori prevalence, antibiotic resistance, dietary exposures or improvements in early diagnosis pathways, including screening. Although historical changes in risk factors are reflected within the APC components, future shifts in exposure or policy are not captured. The model includes main effects for sex, prioritised ethnicity and region, such that all groups share a common underlying temporal pattern. As a result, projected differences between groups arise from baseline incidence and population structure, rather than group-specific temporal trends.

The analysis was conducted at the level of total gastric cancer (ICD-10 C16) and did not distinguish between anatomical subsites, such as proximal and distal cancers, which have differing aetiologies and risk factor profiles. Subsite-specific modelling was not undertaken due to data structure and small numbers within stratified groups but represents an area for future work.

Use of prioritised ethnicity, while consistent with national reporting standards, may under-estimate burden for some groups, particularly Pacific peoples, and does not capture multi-ethnic identification. Broad level 1 ethnic groupings may also mask important heterogeneity, and known misclassification and undercounting, including under-representation of Māori in health datasets, may contribute to under-estimation.^46,47

Population denominators were derived from official Statistics New Zealand projections, with uncertainty increasing over time. While the bootstrap approach incorporated perturbation of population estimates, structural changes such as migration, differential ageing and future policy or behavioural shifts are not fully captured. The reported uncertainty intervals reflect statistical variation within the modelling framework but do not encompass all sources of uncertainty, including model specification or alternative future scenarios. These estimates should therefore be interpreted as indicative trends to support planning, rather than precise forecasts.

Conclusion

These projections indicate that gastric cancer will remain an important and unequally distributed cancer in Aotearoa New Zealand through to 2045, with declining age-standardised incidence but increasing absolute case numbers driven by demographic change. The highest ongoing burden remains among Māori and Pacific peoples, while substantial future growth in case numbers is also projected for Asian populations and in regions experiencing rapid population growth. This reinforces the need for equity-focussed prevention and service planning. Strengthened H. pylori control, including targeted and accessible test-and-treat approaches, offers a key opportunity to reduce future incidence. These national and regional projections can support alignment of prevention, diagnostic pathways and specialist service capacity with projected need.