The volume of publicly funded treatment delivered to lung cancer cases in Aotearoa New Zealand

With more than 1,800 deaths each year, lung cancer kills more New Zealanders than breast, prostate, melanoma and cervical cancers combined.¹ Around 2,400 New Zealanders are diagnosed with lung cancer each year,² and because of the appalling prognosis following diagnosis, around 65% of these cases will be dead within 1 year, and around 80% will be dead within 3 years.³ The lung cancer problem is particularly important for Māori, who are more than three times more likely to develop lung cancer, and 30% more likely to die once diagnosed compared with non-Māori.⁴ Progress made in tobacco control since the 1980s and 1990s means that rates of lung cancer are slowly reducing over time, primarily for men,⁵ and these reductions in incidence will lead to reductions in lung cancer deaths. However, because of factors including population growth, even assuming that we follow our current decreasing incidence trajectory, we will be diagnosing nearly 4,000 New Zealanders with lung cancer every year by 2044.⁵ Recent repeals to New Zealand’s world-leading tobacco legislation^6,7 will serve to further entrench this projected future burden.

Beyond the obvious burden to individuals and families caused by this disease, lung cancer is immensely complex and resource-intensive to treat.⁸ Early warning signs are vague and often mistaken for other respiratory complaints, which means that most cases are caught late in the natural course of the disease. Diagnosis is invasive and not straightforward. Curative and palliative treatments draw across the full range of modern cancer care,⁹ from complex surgery that can only be performed by specialists in a few treatment hubs, to radiation therapy only delivered in large cities, to a growing range of systemic therapies, mostly intravenously delivered in cancer centres. The complexity of the care required to treat lung cancer is complicated by a propensity for the disease to metastasise to other organs, as well as complexities arising from common comorbidities that need to be managed.¹⁰

The commonality, complexity, inherent treatment risk with consequent need for intensivist infrastructure and outright expense needed to treat lung cancer means that our public healthcare system will invariably carry the lion’s share of the care burden, with limited provision in private settings. However, the size of this burden remains unclear, and we currently do not know how much care we are providing to treat our most deadly cancer. Providing this care is the responsibility of central government within our universal healthcare setting, and so this information has immediate relevance for forward-planning, particularly given that despite reductions in incidence rates, the absolute number of lung cancers we will need to treat will still increase over the coming decades.⁵ Understanding the burden of lung cancer treatment in New Zealand is also relevant when we are discussing what national direction we want to take when it comes to tobacco control policy: this information is currently missing from this debate.

The purpose of this study was to describe current volumes of publicly funded surgery, radiation therapy and systemic therapy delivered to New Zealanders with lung cancer. Given that Māori are three times more likely to develop lung cancer than European New Zealanders,² and that we have previously shown disparities in access to surgery between Māori and Europeans with lung cancer,^11,12 we also sought to describe and compare treatment volumes by ethnicity.

Methods

Participants and data sources

We identified all lung cancers (International Classification of Diseases [ICD] codes C33–C34) registered on the New Zealand Cancer Registry (NZCR) from 1 January 2015 to 31 December 2021. Registrations on the NZCR were then linked to sources of publicly funded treatment data: the National Minimum Dataset (NMDS; for surgery and radiation therapy data), the National Non-Admitted Patient Collection (NNPAC; for radiation therapy data), and the Pharmaceutical Collection (PHARMS; for systemic therapy). All treatment data were extracted from 5 years prior to the onset of the study period to allow for the calculation of comorbidity (see Socio-demographic and clinical variables below).

Socio-demographic and clinical variables

Prioritised ethnicity was derived from the NZCR (or supplementary sources within the National Collections such as the NMDS, Primary Health Organisation [PHO] Collection, Mortality Collection and NNPAC) and categorised using prioritised ethnicity into Māori (ethnicity code 21), European (ethnicity codes 10–12), or “Other” (i.e., all non-Māori, non-European peoples). While we defined ethnicity data for everyone in the cohort, ethnic comparisons focussed on comparing Māori and European. Age at diagnosis was derived from the NZCR, and categorised as <50, 50–64, 65–74, and 75+ years. Gender was derived from the NZCR (or supplementary sources), and all people were recorded as either male or female (no other/unknown gender records within the dataset for this period).

Area-level socio-economic deprivation was defined using the New Zealand Index of Deprivation (NZDep) area deprivation index,¹³ which determines area deprivation based on the domicile code available on the NZCR (or other supplementary sources within National Collections if unavailable on the NZCR). NZDep was categorised into five quintile categories (NZDep 1 [least deprived], 2, 3, 4 and 5 [most deprived]). Rurality was defined using the Geographical Classification for Health (GCH), based on domicile code drawn from the NZCR (or supplemented from other National Collections, as for the ethnicity and deprivation information above).¹⁴ The GCH categorises urban/rural status based on travel time to the edge of an urban area, population size and population density, with categories including Urban 1 (“most” urban), Urban 2, Rural 1, Rural 2 and Rural 3 (“most” rural).¹⁴

Comorbidity at cancer diagnosis was defined using the C3 cancer comorbidity index.¹⁵ C3 comorbidity score was categorised as 0 (score <=0), 1 (>0, <=1), 2 (>1, <=2) and 3 (>2). The C3 index was used as a categorical variable within descriptive analysis, and as a splined variable for regression modelling/marginal standardisation using restricted cubic splines with four knots at the median, 75th, 90th and 95th percentiles.¹⁶

Stage of disease was derived from the NZCR, using the SEER Summary Stage method (A to F),¹⁷ with stage collapsed into Local (SEER Summary Stage B), Regional (C and D), Advanced (E) and Unstaged (F).¹⁸ Tumour histological subtype (referred to as “tumour type” in this report) was defined using morphology data from the NZCR. To define tumour type, we used the approach taken by the International Agency for Research on Cancer (IARC) in their Cancer Incidence in Five Continents,¹⁹ and used the World Health Organization Blue Book and ICD-10-O code books²⁰ to set categories based on tumour histology. Small cell lung cancers were one category, and non-small cell cancers were disaggregated into adenocarcinomas, squamous cell carcinomas and other/unspecified tumours. 

Treatment variables

Surgery

Following a clinically guided process that excluded non-cancer-related surgeries, a final master list of 432 Australian Classification of Health Interventions surgical procedure blocks was created.²¹ Procedure blocks, hereafter termed “procedures”, often aggregate individual procedures up to a block of like-procedures; for example, the “lobectomy” procedure block comprises left lung and right lung lobectomies. This master list was then used to identify procedures within the NMDS dataset and define the count and average volume of cancer-related (“any cancer”) procedures delivered to cases. This “any cancer” expands beyond surgery specific to lung cancer to ensure that we captured a snapshot of all cancer-relevant treatment received by people in our cohort following diagnosis, even if a given surgery was more closely related to the treatment of a different primary cancer. We also used a lung cancer–specific list of procedures to define the number and average of cancer-specific procedures delivered to cases (“cancer-specific”). When counting surgery, we calculated “surgical-days”. When counting the total volume of surgery, a surgical-day was when one of our listed procedures was performed; if more than one of the procedures was performed on the same day, this was still only counted as 1 surgical-day. When counting the most common procedure types, we counted the number of unique days that a given procedure occurred on during follow-up. These measures reflect person-centred experiences of the healthcare journey.

Radiation therapy

We used NMDS (publicly funded data only) and NNPAC data to identify receipt of radiation therapy during the follow-up window. A list of the codes used to identify radiation therapy from these two data sources is shown in Appendix 1. If a radiation therapy attendance was recorded on the same day in both NMDS or NNPAC, we only counted one radiation therapy attendance (“attendance-day”). This is because such a scenario was considered to be a quirk of how attendances are recorded across these two datasets, given the low clinical likelihood of two separate radiation therapy treatments being delivered on the same day. Conceptually this means that this metric is a patient-centred variable that records the number of days on which a person attended a session for radiation therapy.

Systemic therapy

The PHARMS collection was used to identify receipt of systemic therapy during the follow-up window (“dispensings”). Categories of systemic therapy were defined based on chemical names in consultation with our clinical team members, and set as chemotherapy, targeted therapy, immunotherapy and hormone therapy. Two medical oncologists independently reviewed all chemical names listed by the PHARMS collection under the “Oncology Agents and Immunosuppressants” therapeutic grouping, and either mapped a chemical into one of these categories or removed the chemical from consideration within our study. For chemotherapy and targeted therapy, which can be delivered intravenously or orally, we additionally stratified by method of delivery. The full list of included chemicals and their respective categories is included in Appendix 2.

Measuring treatment volume

For each person diagnosed with lung cancer, we examined treatment delivery across a time window beginning 90 days prior to cancer diagnosis and spanning to 2 years post-diagnosis to allow for sufficient time for treatment relevant to the cancer to be captured. While we recognise that an individual may continue to be treated for the same cancer beyond this 2-year window, we selected this window to a) maximise the likelihood that the treatment being delivered is related to the same primary cancer, b) provide a single, consistent time window for capturing treatment volume, and c) balance the need for recent data against data that were currently available at the time we conducted analysis. Because treatment data were only available until July 2023, when examining treatment volume we restricted the cohort to those diagnosed prior to July 2021. 

Measuring “any-receipt” of treatment

In addition to determining treatment volume, for descriptive context we also described “any-receipt” (that is, at least one instance) of treatment across a time window beginning 90 days prior to cancer diagnosis and spanning to 1 year post-diagnosis to allow for sufficient time for at least one treatment relevant to the cancer to be captured.

Statistical analysis

Descriptive analysis

We described cohort characteristics for both the total cohort for a given cancer, as well as separately for Māori and Europeans, including crude numbers and percentages and (for the ethnicity-stratified data) age-standardised percentages. We described any-receipt of treatment as crude numbers as well as crude and age-standardised proportions of the total and ethnicity-stratified cohort, with marginal-standardised proportions also calculated for the latter (see description below). We described the volume of surgery-days, radiation therapy attendance-days and systemic therapy dispensings delivered over the treatment window. For surgery, we also described the most common procedures by volume (e.g., lobectomy).

As well as describing volumes, we calculated how much of the total volume of a given outcome was delivered to a given ethnic group (e.g., the crude percentage of total surgeries delivered to Māori). We also determined the “average” (mean) rate of treatment per case, as well as the average treatment delivered per treated patient (e.g., mean number of radiation therapy attendance-days, per person who received any radiation therapy). Age standardisation of cohort characteristic and outcome data was calculated through direct age standardisation, with the Māori lung cancer cohort as the standard population. Age standardising these results allowed us to control for the younger age structure of the Māori lung cancer cohort compared with the European lung cancer cohort.

Marginal standardisation

To adjust for the potential confounding and/or mediating impact of other factors on access to a given treatment, for the any-receipt by ethnicity analysis we conducted marginal standardisation based on logistic regression modelling, while average treatment results were marginally standardised based on Poisson regression modelling¹⁹ using the %margins macro in SAS. This process allowed us to compare proportions/rates between ethnic groups that were adjusted for potential differences in covariate profile between these groups, using the Māori lung cancer socio-demographic and clinical covariate profile as the reference. The rcspline macro was used to include the C3 comorbidity score as a splined variable in the marginal standardisation, with four knots at the median, 75th, 90th and 95th percentiles. Within our results presentation, marginally adjusted rate/mean data are labelled as “fully adj.” (short for fully adjusted).

Software

Data planning, management, analysis and presentation was completed using SAS v9.4 (SAS Institute Inc., North Carolina, United States of America), Microsoft Excel (Microsoft Corporation, Washington, United States of America) and R version 4 (R Foundation for Statistical Computing, Vienna, Austria).

Results

A total of 16,850 registrations of lung cancer occurred across the study period. Their socio-demographic and clinical characteristics are shown in Figure 1. Around 22% of the total cohort were Māori (n=3,642), while 68% were European (11,410) and 10% were non-Māori/non-European (Other, 1,798).

View Figure 1–3, Table 1–2.

Any-receipt of cancer treatment for lung cancer by treatment type across the cohort, in total and by ethnicity, is shown in Appendix 3. Around a third of lung cancer registrations received lung cancer–specific surgery (age standardised 31%), more than 40% received radiation therapy (age std. 43%) and more than a third received systemic therapy (age std. 37%). After adjusting for covariates, Māori and Europeans were similarly likely to receive these treatments (Appendix 3).

The total and average volumes of publicly funded treatment delivered to people diagnosed with lung cancer across our study period are shown in Table 1 and Figure 2 below. For each year of people diagnosed, the following publicly funded treatment was delivered across a 2-year treatment window:

• Surgery: A total of 1,468 surgical-days of cancer-related (any cancer) surgery were delivered per year of diagnosed cancers (Māori 259 days, crude proportion 18%; Europeans 1,024, 70%), giving an age-standardised average of 0.6 surgical-days per person diagnosed (fully adjusted: Māori 0.5, European 0.6). A total of 978 surgical-days of lung cancer–specific surgery were delivered per year of people diagnosed (Māori 176, European 662), at an age-adjusted average of 0.4 surgical-days per people diagnosed (fully adjusted: Māori 0.3, European 0.4).
• Radiation therapy: A total of 11,027 attendance-days of radiation therapy were delivered per year of people diagnosed (Māori 2,814, 26%; European 7,141, 65%), at an age-adjusted average of 5 attendance-days per person diagnosed (fully adjusted: Māori 6, European 5).
• Systemic therapy: A total of 9,108 dispensings of systemic therapy were delivered per year of people diagnosed (Māori 2,278, 25%; European 5,527, 61%), at an age-adjusted average of five dispensings per person diagnosed (fully adjusted: Māori 4, European 4). The vast majority of systemic therapy was chemotherapy, with a total of 7,461 (Māori 28%, European 61%) chemotherapy dispensings delivered per year of people diagnosed, mostly intravenous (IV) (6,670, Māori 28%, European 62%) compared to oral (791, Māori 20%, European 60%). A total of 1,253 (Māori 9%, European 52%) targeted therapy dispensings were delivered per year of people diagnosed, mostly oral (1,183, Māori 9%, European 50%) compared to IV (69, Māori 8%, European 75%). The non-Māori/non-European (Other) group received 39% of the delivered targeted therapy (Figure 2). Both hormone and immunotherapy delivery were uncommon and not individually interpreted but are presented in Figure 2 and Appendix 4.

For each person who received a given treatment, the following average treatment was delivered across a 2-year treatment window (Table 2):

• Surgery: An age-standardised mean of 1.7 surgical-days of cancer-related (any cancer) surgery were delivered for each person who received any cancer-related surgery (fully adjusted: 1.7 Māori, 1.7 European), while an average of 1.5 surgical-days of cancer-specific surgery were delivered for each person who received any cancer-specific surgery (Māori 1.5, European 1.5).
• Radiation therapy: An age-standardised mean of 11.5 attendance-days of radiation therapy were delivered for each case who received any radiation therapy (fully adjusted: 11.5 Māori, 11.5 European).
• Systemic therapy: An age-standardised mean of 11.8 systemic therapy dispensings were delivered for each case that received any systemic therapy (fully adjusted: 11.7 Māori, 11.8 European). An average of 11.5 (Māori 11.5, European 11.6) dispensings of chemotherapy were delivered to each case who received chemotherapy, with an average of 11.1 IV dispensings (Māori 11.1, European 11.0) and 5.3 oral dispensings (Māori 4.5, European 5.8). An average of 10.1 (Māori 8.1, European 9.6) dispensings of targeted therapy were delivered to each case who received targeted therapy, with an average of 10.0 oral dispensings (Māori 8.3, European 9.5). Average rates of hormone and immunotherapy dispensings were not interpreted but are presented in Appendix 5.

The most common cancer-specific surgical procedures delivered to Māori and Europeans across a 2-year treatment window are shown in Figure 3, and for the total cohort in Appendix 6. The two most common procedures delivered were application, insertion or removal procedures on chest wall, mediastinum or diaphragm (total 524 surgical-days per year of people diagnosed, Māori 95, European 351), followed by lobectomy of lung (total 227, Māori 36, European 156). The list of the most common procedures was broadly comparable between Māori and Europeans.

Discussion

Key findings

High volumes of surgery, radiation therapy and systemic therapy were delivered to people diagnosed with lung cancer across our study period, with radiation therapy and systemic therapy the most common treatment modalities. For each year of people diagnosed, our public healthcare system delivered 978 surgical-days of lung cancer–specific surgery, 11,027 attendance days of radiation therapy and 9,108 dispensings of systemic therapy, primarily IV chemotherapy, within a 2-year window following diagnosis.

Because of incidence disparities, Māori are receiving nearly 20% of all publicly funded lung cancer surgery and nearly a quarter of all lung radiation therapy and systemic therapy. There were no discernible differences between Māori and Europeans in terms of the average volume of treatment delivered per person diagnosed. The relatively high volume of oral targeted therapy among the Other ethnic group is likely due to a higher incidence of targetable mutations, including epidermal growth factor receptor in Asian and Pacific populations.²² These mutations have also been observed to be more common among Māori than Europeans,²² so the observation of similar average levels of targeted therapy between these groups could indicate differential access to mutation testing and subsequent prescribing. When focussing on those who were treated, we found no strong differences between Māori and Europeans in the average volume of surgery, radiation therapy or systemic therapy. A possible exception was the slightly smaller level of dispensing of targeted therapy to Māori compared with Europeans (Māori approximately eight dispensings per treated person, European approximately 10 dispensings). Deeper enquiry with more granular data is required to understand possible differences in access to and subsequent volume of received targeted therapy among Māori.

The most common lung cancer–specific procedures included chest and pleural excision procedures, lung resections and common lymph node metastases. The most common procedures delivered to Māori were broadly comparable with those delivered to Europeans. The broader cancer-related (any cancer) procedures included resections of organs that are frequent sites of primary lung cancer metastases, including the liver and brain. This list additionally included some procedures unlikely to be lung cancer metastases, including skin and colorectal procedures (Appendix 7)—highlighting the fact that lung cancer does not occur in isolation from the management of other common cancers.

Ramifications and recommendations

Lung cancer is complex, resource-intensive and expensive to diagnose and treat. In this study we have quantified the volume of surgery, radiation therapy and systemic therapy delivered by New Zealand’s public health system in the treatment of our most lethal cancer to provide a snapshot of the current state of delivery. This was achieved using national routinely collected data, meaning that the kind of monitoring presented here could (and arguably should) be business-as-usual in a universal healthcare system. We recommend that this type of reporting becomes an ongoing concern for central government.

If implemented, a national lung cancer screening programme would ideally shift diagnoses to an earlier stage, which, over time, will increase the need for curative care relative to palliative care. This is a positive outcome, since more lives will be saved than without screening, and as such we recommend that central government implements lung cancer screening nationally in an equitable and evidence-based manner.^23,24 Since the volume of treatment will increase with a well-run screening programme, particularly surgical treatment of localised disease,¹² we also recommend that planning toward a national screening programme must anticipate increased downstream clinical need within our system. Screening should not be implemented prior to the completion of such planning.

The vast majority of lung cancers are preventable through robust tobacco control (as well as clean air) policies, legislation and regulation. We recommend reinstating the world-leading tobacco legislation reversed by the current Government,^6,7 which, over time, will not only save lives, but also reduce the numbers of people with lung cancer who need access to operating rooms, radiation therapy bunkers and chemotherapy lounges. Because the absolute number of cancers is increasing over time,⁵ lung cancer will remain a burden on our health system for the foreseeable future; however, for the benefit of future generations, we can choose now to enact policies that will eradicate tobacco-caused lung cancer as swiftly as possible.

Finally, our observations highlight the variation and frequency of treatment experienced by treated cases of lung cancer: an average of 1–2 surgeries, approximately 12 radiation therapy attendance-days likely split across multiple courses and approximately 12 systemic therapy dispensings, which would primarily be delivered in outpatient IV suites. Given that most people diagnosed with lung cancer will also have comorbidity, this treatment would ideally be occurring amid the management of other conditions. All told, lung cancer ushers a significant decline in health and wellbeing from the time of symptom onset to the end of life, a decline punctuated by a frequent necessity to engage with health services and receive major treatment. This burden, carried by individual lung cancer cases as well as their wider whānau, cannot be under-estimated, and must be centred within care plans and national cancer control planning alike.

Strengths and limitations

This was a national study of all lung cancer registrations linked to all publicly funded cancer treatment, which enabled a snapshot of lung cancer treatment delivery. Rather than focussing only on lung cancer–specific therapy, we have reported the totality of cancer-related care delivered to cases of lung cancer. This approach enables us to describe the lived reality of patients diagnosed with lung cancer, for whom not all cancer-related care will be directed to their primary cancer. Taking this approach reduces the lung cancer specificity of our results, but we note that a) we have provided lung cancer–specific results for surgery, since this was possible with the available data, and b) the vast bulk of treatment for patients with lung cancer will likely be directly related to their primary lung cancer. Hormone therapy is not among the systemic therapies used to treat lung cancer; as such, the hormone therapy and IV-targeted therapy observed in this study was likely delivered in the treatment of concurrent cancer (e.g., breast or prostate). Immunotherapy delivery was extremely rare because such medicines were not publicly funded across the study period (2015–2021), though New Zealand now has three funded immunotherapies for lung cancer: durvalumab (funded 2022), pembrolizumab (funded 2023) and atezolizumab (funded 2023). As data for these years become available, the volume of immunotherapies delivered in the treatment of lung cancer will invariably be higher than that reported here. We note that while all systemic therapy captured within the current study was publicly funded, it may have been prescribed in a privately funded setting. Finally, we note that undercounting of Māori within health research is an ongoing concern.²⁵ In this study we mitigated this issue by supplementing NZCR ethnicity data with ethnicity data from other National Collections where NZCR data were missing or non-specific.

Conclusion

In this national study of treatment received by people newly diagnosed with lung cancer between 2015 and 2021, we described the total count and average per-person volume of surgery, radiation therapy and systemic therapy delivered by our public healthcare system. We found that for each year of newly diagnosed people, our public system delivered 978 surgical-days of lung cancer–specific surgery, 11,027 attendance-days of radiation therapy and 9,108 dispensings of systemic therapy over a 2-year window following diagnosis. We found a high per-person volume of radiation therapy and systemic therapy, primarily IV chemotherapy. Encouragingly, there were no strong differences between Māori and Europeans in terms of the average volume of surgery, radiation therapy or systemic therapy per person diagnosed, or per person treated, and the most common surgical procedures delivered to Māori were broadly comparable with those delivered to Europeans. With the absolute number of cases set to increase over time, we must return to our world-leading tobacco cessation pathway in order to curb the significant pressure placed on our system by this common, complex and lethal cancer.

View Appendices.