Original ArticleOriginal Article
Open Access

Impact of cancer diagnosis on life expectancy by area-level socioeconomic groups in New South Wales, Australia: a population-based study

Md Mijanur Rahman, Michael David, David Goldsbury, Karen Canfell, Kou Kou, Paramita Dasgupta, Peter Baade and Xue Qin Yu
Cancer Biology & Medicine August 2024, 21 (8) 692-702; DOI: https://doi.org/10.20892/j.issn.2095-3941.2024.0166

Abstract

Objective: Improvement in cancer survival over recent decades has not been accompanied by a narrowing of socioeconomic disparities. This study aimed to quantify the loss of life expectancy (LOLE) resulting from a cancer diagnosis and examine disparities in LOLE based on area-level socioeconomic status (SES).

Methods: Data were collected for all people between 50 and 89 years of age who were diagnosed with cancer, registered in the NSW Cancer Registry between 2001 and 2019, and underwent mortality follow-up evaluations until December 2020. Flexible parametric survival models were fitted to estimate the LOLE by gender and area-level SES for 12 common cancers.

Results: Of 422,680 people with cancer, 24% and 18% lived in the most and least disadvantaged areas, respectively. Patients from the most disadvantaged areas had a significantly greater average LOLE than patients from the least disadvantaged areas for cancers with high survival rates, including prostate [2.9 years (95% CI: 2.5–3.2 years) vs. 1.6 years (95% CI: 1.3–1.9 years)] and breast cancer [1.6 years (95% CI: 1.4–1.8 years) vs. 1.2 years (95% CI: 1.0–1.4 years)]. The highest average LOLE occurred in males residing in the most disadvantaged areas with pancreatic [16.5 years (95% CI: 16.1–16.8 years) vs. 16.2 years (95% CI: 15.7–16.7 years)] and liver cancer [15.5 years (95% CI: 15.0–16.0 years) vs. 14.7 years (95% CI: 14.0–15.5 years)]. Females residing in the least disadvantaged areas with thyroid cancer [0.9 years (95% CI: 0.4–1.4 years) vs. 0.6 years (95% CI: 0.2–1.0 years)] or melanoma [0.9 years (95% CI: 0.8–1.1 years) vs. 0.7 years (95% CI: 0.5–0.8 years)] had the lowest average LOLE.

Conclusions: Patients from the most disadvantaged areas had the highest LOLE with SES-based differences greatest for patients diagnosed with cancer at an early stage or cancers with higher survival rates, suggesting the need to prioritise early detection and reduce treatment-related barriers and survivorship challenges to improve life expectancy.

keywords

Introduction

People diagnosed with cancer are living longer than ever before and the number of cancer survivors is growing rapidly worldwide. Despite consistent improvement in survival for many cancer types over recent decades, disparities by area-level socioeconomic status (SES) persist in many countries, including Australia1–3. A substantial body of research has demonstrated socioeconomic and residence in remote area disparities in cancer survival4–7. While the reasons for these disparities are multi-faceted, some of the differences may stem from a more advanced stage of diagnosis and sub-optimal treatment options for patients in the most disadvantaged areas8. Quantifying the disparities in years of life lost (YLL) due to a cancer diagnosis by area-level SES is critical for addressing this public health issue, informing health policy decision-makers, and improving cancer outcomes.

Most population-based studies that have investigated survival disparities by area-level SES rely on the 5-year relative survival (RS)9,10. However, RS cannot be used to measure the lifetime impact of a cancer diagnosis. An alternative measure, the loss of life expectancy (LOLE), has been increasingly used in quantifying the expected remaining life years following a cancer diagnosis, thus providing a means to characterise the impact of disease through a comparison of the expected number of life years lost with the life expectancy of the general population11,12. The LOLE due to cancer can reflect the lifetime impact of a cancer diagnosis by removing the condition at a specific time point after diagnosis13. The LOLE also provides a better understanding of the actual impact of cancer at individual and population levels and is therefore of increasing interest to public health researchers, consumers, and policymakers.

Several studies have examined the influence of area-level SES on the LOLE due to a cancer diagnosis, including one study from England10 and another study from Queensland, Australia14. These studies reported noticeable differences in the impact of a cancer diagnosis on LOLE by area-level SES for common cancer types. However, these studies did not consider the stage of diagnosis during the statistical analysis. Although previous studies have shown that cancer patients living in more disadvantaged areas are more likely to be diagnosed with advanced cancer than those living in the least disadvantaged areas15, it is unclear whether the stage at diagnosis can explain all of the LOLE differences due to the SES.

No published study has determined the impact of a cancer diagnosis on the LOLE by area-level SES and stage at diagnosis. Therefore, this study aimed to quantify the LOLE resulting from a cancer diagnosis by area-level SES and gender in New South Wales (NSW), Australia, while adjusting for age and cancer stage at the time of diagnosis. The objectives of this study were as follows: estimate how long cancer patients are expected to live based on their residential area-level SES and compare the life expectancy to the average life expectancy of the NSW general population; estimate the differences in life expectancy by area-level SES for various common cancer types based on factors, including gender, age, and cancer stage at the time of diagnosis; and calculate the additional years of life that could be gained if the survival disparities related to area-level SES were eliminated in a given year in NSW.

Patients and methods

Data and study population

Data were collected from the Cancer Institute NSW Enduring Cancer Data Linkage (CanDLe), a multiple-linked administrative data initiative of all cancer patients in NSW, which represents greater than one-third of cancer diagnoses in Australia16,17. The current study cohort included 422,680 patients between 50 and 89 years of age who were registered in the NSW Cancer Registry (NSWCR) and diagnosed with cancer from 2001–2019. The mortality follow-up status was obtained until 31 December 2020 from the NSW Registry of Birth, Death, and Marriage (RBDM) with a 1-year follow-up evaluation minimum. Cases diagnosed at death, in-situ cases, and cases with missing area-level SES information were excluded. Only the first primary cancer diagnosis was included because a prior cancer diagnosis may affect patient prognosis. Patients diagnosed with cancer before 2001 were not included because reliable estimates of the general population mortality rate by age, gender, and area-level SES were not available for earlier years. Like previous studies13,14,18, the current study cohort was restricted to patients 50–89 years of age to avoid longer extrapolations when calculating the LOLE for a small number of younger patients and unreliable estimates of RS for older patients. Cancers of the lymphohematopoietic system were also excluded because staging information is not clinically relevant. Finally, analyses were conducted separately for 12 common cancer types [International Classification of Disease (ICD)-10 codes in Table S1] considering Statistical Area 2 (SA2) as the geographic unit of analysis19. These 12 common cancers have a significant impact on the population with respect to incidence and mortality outcomes. Additionally, information on the stage of cancer at the time of diagnosis was available for these cancer types. Ethics approval was obtained for the sub-study protocol as part of the Cancer Institute NSW CanDLe by the NSW Population and Health Services Research Ethics Committee [NSWPHSREC (Approval No. 2022UMB0903)].

Area-level SES

The Index of Relative Socioeconomic Advantage and Disadvantage (IRSAD), a census-based aggregated general socioeconomic index, was used to measure the area-level SES20. IRSAD summarises a range of information about the economic and social conditions of people and households within SA2, with a population range of 3,000–25,00021. A lower score indicates a relatively greater level of disadvantage (i.e., many households with low income and many people with no qualifications and in low-skilled occupations), while a higher score indicates a relative lack of disadvantage and an overall greater level of advantage (i.e., many households with high income and many people with high qualifications and in skilled occupations). The IRSAD score was collapsed into 3 broad groups: ‘most disadvantaged’ (quintile 1); ‘middle SES’ (quintiles 2–4); and ‘least disadvantaged’ (quintile 5).

Stage of cancer at the time of diagnosis

The NSWCR recorded the spread (stage) of cancer at the time of cancer diagnosis as localized, regional (adjacent organs or regional lymph nodes), distant metastases, and unknown or missing stage. Missing information on the spread of cancer was grouped with the unknown group in this study.

LOLE

The LOLE is defined as the difference between the life expectancy of patients with cancer and the life expectancy of comparable individuals from the general population matched by gender, age, and calendar year. A comparison of the life expectancy of cancer patients with the general population with similar sociodemographic characteristics provides a straightforward measure of the burden of cancer on the population of interest and reflects the accumulated excess mortality over the lifespan of cancer patients9,18,22,23. However, the LOLE measure is highly dependent on age at the time of cancer diagnosis. An alternative measure, the proportion of remaining YLL, which is defined as the LOLE divided by the life expectancy, provides a comparable metric across groups with different age distributions. The YLL is the average absolute number of years lost due to cancer.

Furthermore, the LOLE estimates can also be used to quantify the average life years that could be saved if disparities in cancer survival are addressed18.

Statistical analysis

The study cohort was initially evaluated based on the distribution of cancer patients by cancer type, gender, and area-level SES. The mean age at the time of cancer diagnosis was also calculated for each cancer type by gender and area-level SES. To calculate the life expectancy of the cancer cohort, we utilized flexible parametric survival models that regress the cumulative excess hazard on the log scale and enable extrapolation of RS data by age at the time of cancer diagnosis, calendar year, gender, area-level SES, and the spread of disease at the time of cancer diagnosis using spline models. Details involving flexible parametric survival models have been published elsewhere13,24,25. The commonly used Cox regression model is not suitable for extrapolating the observed survival curve beyond the available follow-up evaluations. Unlike the Cox regression model, the flexible parametric model does not require a proportional hazard assumption. Therefore, the ratio of hazard rates across different groups remains constant over time. This model uses splines (i.e., constrained piecewise polynomials joined at “knots”) to estimate the underlying hazard function. Several earlier studies have used this method to quantify the LOLE in cancer patients9,10,12,14,18,26–29.

Flexible parametric survival models were fitted separately for each cancer type, adjusting for age and year at the time of cancer diagnosis, gender (when relevant), area-level SES, and the spread of disease at the time of cancer diagnosis. To capture the non-linear relationships and potential complexities associated with age and year at the time of cancer diagnosis, we modelled these variables using restricted cubic splines with four and two degrees of freedom, respectively. Interactions between year and age were included in all models.

Complete and smoothed population mortality tables were generated using death and population counts by 5-year age group, gender, year, and SA2 obtained from the RBDMs30. Death rates and probabilities of death by 1-year age groups were estimated using a flexible Poisson multivariable model, which has been shown to provide more robust and unbiased estimates of smoothed age-specific mortality rates for small populations than the standard life table approach31. Smoothing was carried out by combining groups of neighbouring SA2s to enhance the stability of the population mortality tables.

The life expectancy of cancer cohorts was estimated based on extrapolation of all-cause survival by multiplying the predicted RS from the flexible parametric models with the expected survival estimates. The survival curves for the cancer cohort and general population were calculated based on age, gender, area-level SES, and calendar year. The difference between the two curves was the LOLE due to cancer diagnosis. The predicted LOLE and the proportion of remaining YLL obtained from the models for each cancer patient were summed within each SES, gender, and weighted by age (or age and gender for results by persons), with the weights reflecting the distribution of age (or age-gender) for those diagnosed in 2019, the most recent year for our cancer cohort.

Finally, we calculated the predicted total number of YLL due to cancers diagnosed in 2019 by area-level SES with addition of the model-based predicted LOLE across all observations. The potential total gain in life years for the 2019 cancer cohort was quantified by removing SES disparity in the RS. The total YLL was estimated from the model in the hypothetical scenario. All patients diagnosed in 2019 experienced the same RS as the patients living in the least disadvantaged area, while maintaining the expected survival of the SES group. The reduced total life years lost experienced by the whole cohort, if everyone had the same RS as the least disadvantaged SES group, was the potential gain in life years. All analyses were performed using Stata/SE version 18 (StataCorp, College Station, TX, USA). Flexible parametric models were fitted using the stpm2 package.

Results

Nearly one-quarter (24%) of the cohort (n = 422,680) was from the most disadvantaged areas and 18% were from the least disadvantaged areas (Table 1). The mean age at the time of cancer diagnosis was moderately higher for patients from the most disadvantaged areas than the least disadvantaged SES (69.5 years vs. 67.5 years for any cancer), with the largest differences for female melanoma patients (69.5 years vs. 66.1 years).

View this table:
Table 1

Distribution of the patient population by area-level socioeconomic status (SES), gender and cancer type in New South Wales, Australia, between 2001 and 2019

Furthermore, compared to the least disadvantaged areas, patients from the most disadvantaged areas were less likely to be diagnosed with localized cancer (43.7% vs. 52.9%) but more likely to be diagnosed with distant metastases (17.5% vs. 11.3%; Table S2).

Figure 1 presents the differences in life expectancy by area-level SES in the general population and cancer cohort and the LOLE due to cancer according to age at the time of cancer diagnosis. The SES disparity in life expectancy was wider in cancer patients than in the general population independent of age. The LOLE due to cancer was noticeably higher among the most disadvantaged SES areas than the least disadvantaged areas with the gap narrowing with increasing age at the time of cancer diagnosis.

Figure 1
Figure 1

Life expectancy by age at cancer diagnosis and the loss of life expectancy (LOLE) due to cancer diagnosis for all cancers combined between 2001 and 2020 in New South Wales, Australia.

The average LOLE and the proportion of remaining YLL by area-level SES and cancer type stratified by gender are presented in Table 2. Patients from the most disadvantaged SES had a greater average LOLE than patients from the least disadvantaged SES. There were significant differences in cancer types with higher survival rates, including prostate [2.9 years (95% CI: 2.5–3.2 years) vs. 1.6 years (95% CI: 1.3–1.9 years)] and female breast cancers [1.6 years (95% CI: 1.4–1.8 years) vs. 1.2 years (95% CI: 1.0–1.4 years)]. The highest average LOLE was observed for males from the most disadvantaged areas with pancreatic [16.5 years (95% CI: 16.1–16.8 years)], liver [15.5 years (95% CI: 15.0–16.0 years)], and lung cancer [15.3 years (95% CI: 15.1–15.5 years)]. The corresponding estimates for female patients from the most disadvantaged areas were moderately lower than the least disadvantaged areas [pancreatic cancer: 14.1 years (95% CI: 13.8–14.4 years); liver cancer: 13.0 years (95% CI: 12.5–13.5 years); and lung cancer: 12.4 years (95% CI: 12.2–12.6 years)]. The area-level SES differences in the LOLE for these cancer types with high fatality rates were not significant (P > 0.05). The lowest average LOLE was observed for females from the least disadvantaged areas with thyroid cancer [0.6 years (95% CI: 0.2–1.0 years)], melanoma [0.7 years (95% CI: 0.5–0.8 years)], and breast cancer [1.2 years (95% CI: 1.0–1.4 years)].

View this table:
Table 2

Average loss of life expectancy and proportion of life years lost by area-level SES, cancer type, and gender in New South Wales, 50–89 years of age between 2001 and 2020

In agreement with the LOLE, the maximum proportion of remaining YLL was observed for patients with pancreatic cancer (most vs. least disadvantaged SES: 90% vs. 83% for males and 87% vs. 80% for females), followed by lung and liver cancers. Patients from the most disadvantaged areas had a greater average LOLE for most cancer types (except for cancers with high fatality rates) regardless of age at the time of cancer diagnosis (Figure 2). However, the gap in the LOLE by area-level SES decreased with increasing age at the time of diagnosis.

Figure 2
Figure 2

Loss in life expectancy by area-level socioeconomic status for common cancer types between 2001 and 2020 in New South Wales, Australia.

Patients diagnosed with distant metastases at younger ages had the highest LOLE, followed by unknown, regional, and localized cancer (Figure 3). At 50 years of age, patients from the most disadvantaged SES who were diagnosed with distant metastases lost approximately 30 years of life, whereas patients from the least disadvantaged SES lost approximately 24 years. The corresponding estimates for those diagnosed with localized cancer were nearly 5 years among the most disadvantaged and 2 years among the least disadvantaged SES groups. The results regarding the difference in the LOLE by area-level SES due to the spread of cancer at the time diagnosis for five major cancers are presented in Figures S1–S5.

Figure 3
Figure 3

Loss in life expectancy due to any malignancy by area-level socioeconomic status and cancer stage at diagnosis between 2001 and 2020 in New South Wales, Australia.

Figure 4 presents the total YLL by area-level SES and the gain in life years by cancer types if the SES differential was removed in NSW in 2019 (i.e., if patients from the most disadvantaged areas had the same RS as patients from the least disadvantaged areas). The total YLL due to cancer diagnosis among patients from the most disadvantaged areas was 37,068 in 2019. Due to greater LOLE coupled with an increased number of patients from the most disadvantaged SES, lung cancer had the highest YLL in 2019 (most disadvantaged: 12,941 vs. least disadvantaged: 3,886), followed by bowel and prostate cancer. The total gain in life years in 2019 if the area-level SES disparity in RS was removed reached 5,864 with the highest total life years gain for patients with prostate (1,769) followed by lung (1,339) and bowel cancer (839).

Figure 4
Figure 4

Total years of life lost (YLL) by most and least disadvantaged socio-economic status (SES) group and the gain in life years if inequality in SES groups was removed in New South Wales, 2019.

Discussion

In this population-based study the impact of cancer diagnosis was estimated by area-level SES groups in NSW, Australia between 2001 and 2020 by quantifying the LOLE and the proportion of remaining life years lost for 12 common cancer types. The results indicate a notable reduction in life expectancy due to cancer diagnosis for males and females with substantial variations by cancer type, age, the spread of cancer at the time of diagnosis, and area-level SES. Patients with pancreatic, liver, and lung cancers had the highest average LOLE and the proportion of remaining YLL, whereas patients with melanoma, and thyroid and breast cancers had the lowest average LOLE and the proportion of remaining YLL. These findings are consistent with the average LOLE estimates of previous Australian14,23,26,28 and international studies10,18,27,29. However, most previous studies only focused on major cancer types and examined the temporal trend in LOLE. The current study, for the first time, estimated the average LOLE for a broad range of common cancer types by area-level SES, adjusting for the spread of cancer at the time of diagnosis.

One of the key findings of the current study is the significant differences in the average LOLE and the proportion of remaining YLL by area-level SES for major cancer types, including prostate, female breast, and bowel cancer, and melanoma but not cancer types with high fatality rates, such as lung, pancreatic, and liver cancer. Patients from the most disadvantaged areas had a greater reduction in life expectancy than patients from the least disadvantaged areas. This finding differs from previous studies conducted in Queensland, Australia14 and England10. The Queensland study reported that the differences in the LOLE by area-level SES were not significant for major cancer types, including colorectal, female breast, lung, and prostate cancer, and melanoma separately but were significant for all cancers combined. However, there were significant differences when the area-level SES was combined with accessibility to services. Patients from the most disadvantaged and low accessibility areas had a greater average LOLE than patients from the least disadvantaged and high-accessibility areas. Differences in the LOLE by SES group were not significant and were inconsistent across different cancer types in the English study. We extended these previous studies by incorporating information on the spread of cancer at the time of diagnosis, which is an important prognostic factor for survival. In so doing, we accounted for differences in the distribution of cancer stage by area-level SES, hence the resulting estimates mainly reflect differences in factors related to treatment or other survivorship challenges.

Patients diagnosed with distant metastases had a greater LOLE than patients diagnosed at localized stages, which is consistent with another Australian study26. The gap in the area-level SES disparity was wider for patients who were diagnosed with cancer at a younger age and a lower spread of cancer at the time of diagnosis. Additionally, the area-level SES disparities in the LOLE were relatively high for cancers with higher survival rates. Several reasons may explain these disparities. One important reason may be the disparities in cancer treatment and survivorship challenges, including financial toxicity. Patients from the most disadvantaged areas are more likely to receive sub-optimal cancer treatment and poorer survivorship care than patients from the least disadvantaged SES. Sub-optimal cancer treatment and poor survivorship care may be associated with increased mortality that can contribute to greater LOLE for patients from the most disadvantaged areas32. For cancer types with higher fatality rates, the disparities in treatment and survivorship challenges may contribute to relatively smaller differences in the mortality rates but differences in early detection can contribute to some differences in the LOLE. Furthermore, the lower area-level SES disparities in the average LOLE among patients who were diagnosed with cancer at older ages are consistent with previous studies reporting that SES disparity in health decreases with age among older adults33.

Approximately 16% of YLL (5,864) could have been saved in 2019 if the patients from the most disadvantaged areas had the same RS as patients from the least disadvantaged areas. This estimate is like a previous Australian study, which reported that 19% of remaining YLL could be reduced if all patients had the same RS as patients from the least disadvantaged and high accessibility groups14. Our findings indicate a pattern across cancer types with a greater proportion of life years that could be saved for cancer types with higher survival rates. As a single cancer type, patients with prostate cancer had the highest total life years gain in 2019 (1,769), given a high number of cases with significant area-level SES differences. In contrast, the greater LOLE for cancer type with high fatality rates, such as pancreatic, lung, and liver cancer, were also accompanied by a higher absolute gap in YLL between patients from the most and least disadvantaged areas, which was reflected by the total gain in life years for lung cancer patients in 2019 (1,339). This finding suggests that providing optimal treatment and support to mitigate the survivorship challenges for patients from the most disadvantaged areas may reduce the disparities in YLL by area-level SES.

The key strength of our study was the utilization of a large population-based registry dataset comprising greater than one-third of cancer cases in Australia. Unlike other registry data, the NSW CR routinely collects information on the spread of cancer at the time of diagnosis and is linked with multiple administrative datasets, including the NSW RBDM data, through the NSW Cancer Institute CanDLe data initiative. This procedure allowed us, for the first time, to examine area-level SES disparity in the LOLE and the proportion of remaining YLL by the spread of cancer at the time of diagnosis, with additional mortality follow-up for at least 1 year. We conducted separate analyses for 12 common cancer types, which allowed us to identify the patterns in area-level SES disparities across cancer types. By including the spread of cancer in the model, we accounted for the complex association of area-level SES and the spread of cancer into a subsequent impact on the LOLE and the proportion of remaining YLL. The findings on the area-level SES differences in the LOLE stratified by the spread of cancer at the time of diagnosis provide more relevant information to patients with cancer and clinicians regarding disease prognosis, treatment decisions, and future survivorship challenges.

Our study had some limitations. First, the estimated life expectancy was based on extrapolated information from the flexible parametric model, in which more recently diagnosed patients were pronounced and had a relatively shorter follow-up period. The model assumed a linear trend in the log cumulative excess hazard in the future, which may change in practice. Hence, our results may not fully reflect the actual observed survival in the future, as well as the impact of more recent changes in treatment and survivorship care. However, using a spline-based model for the baseline cumulative hazard reduces the potential impact of this model assumption. Previous studies reported satisfactory extrapolation of life expectancies from the survival function, and different modelling assumptions and approaches have small differences in the resulting life expectancies13,34. We also restricted our analyses to patients > 50 years of age to avoid longer extrapolation for younger patients.

Second, the LOLE was estimated within a RS framework with the assumption that cancer deaths are a negligible proportion of all deaths in the general population. A higher proportion of deaths from a specific cancer, may inflate the expected mortality in the population so that LOLE estimates are likely to be underestimated. However, this was not a major problem for most cancer types except for prostate cancer in the older age group35. Additionally, another key assumption of the independence of cancer and non-cancer causes of death may not be valid for some cancer types due to shared risk factors, such as cigarette smoking, obesity, and co-morbidities36,37. Co-morbidity and lifestyle factor information was not available in the NSW CR, and when combined with the lack of a general population mortality table that included these factors, we were not able to explore these factors in the current study. This limitation may be a topic for future research.

Conclusions

Our study demonstrated that patients from the most disadvantaged areas had a significantly greater reduction of life expectancy than patients from the least disadvantaged areas across major cancer types, except for cancers with high fatality rates. The area-level SES disparity was higher in cancer types with higher survival rates and patients diagnosed with cancer at earlier ages and a lower spread of cancer. Further research is warranted to identify whether delays in diagnosis, treatment-related barriers, or survivorship challenges among the most disadvantaged SES groups are drivers of these differences. Notable reductions in life expectancy were observed for cancers with high fatality rates or patients diagnosed with distant metastases. Our findings can help patients and clinicians better understand the burden of cancer and how socio-demographic differences affect cancer patient survival across lifetimes. Finally, our findings provide an essential public health perspective on prioritising early detection and reducing treatment-related barriers and survivorship challenges to improve life expectancy, especially among patients from the most disadvantaged areas.

Supporting Information

Conflict of interest statement

No potential conflicts of interest are disclosed.

Author contributions

Conceived and designed the analysis: Md Mijanur Rahman, Xue Qin Yu.

Collected the data: Third party (Cancer Institute NSW).

Contributed data or analysis tools: Md Mijanur Rahman, Xue Qin Yu, Peter Baade, Kou Kou, David Goldsbury, Karen Canfell.

Performed the analysis: Md Mijanur Rahman.

Wrote the paper: Md Mijanur Rahman wrote the first draft, and all authors critically reviewed the manuscript.

Data availability statement

This data analysis was conducted under conditions approved by the relevant Ethics Committee(s). As a condition of approval, data are not sharable. Access to data by other individuals or agencies would require appropriate ethical approvals to be in place.

Acknowledgements

This research was completed using data from the CanDLe Initiative. The CanDLe Initiative is led by the Cancer Institute NSW and supported by the NSW Ministry of Health. Record linkage was provided by the Centre for Health Record Linkage.

Footnotes

  • ↵*Joint senior author.

  • Received May 7, 2024.
  • Accepted June 10, 2024.

References

  1. 1.↵
    2. Mariotto AB,
    3. Zou Z,
    4. Johnson CJ,
    5. Scoppa S,
    6. Weir HK,
    7. Huang B.
    Geographical, racial and socio-economic variation in life expectancy in the US and their impact on cancer relative survival. PLoS One. 2018; 13: e0201034.
  2. 2.
    2. Mokdad AH,
    3. Dwyer-Lindgren L,
    4. Fitzmaurice C,
    5. Stubbs RW,
    6. Bertozzi-Villa A,
    7. Morozoff C, et al.
    Trends and patterns of disparities in cancer mortality among US counties, 1980-2014. J Am Med Assoc. 2017; 317: 388–406.
    OpenUrlCrossRef
  3. 3.↵
    2. Maomao C,
    3. He L,
    4. Dianqin S,
    5. Siyi H,
    6. Xinxin Y,
    7. Fan Y, et al.
    Current cancer burden in China: epidemiology, etiology, and prevention. Cancer Biol Med. 2022; 19: 1121–38.
    OpenUrlAbstract/FREE Full Text
  4. 4.↵
    2. Stanbury JF,
    3. Baade PD,
    4. Yu Y,
    5. Yu XQ.
    Cancer survival in New South Wales, Australia: socioeconomic disparities remain despite overall improvements. BMC Cancer. 2016; 16: 48.
    OpenUrlPubMed
  5. 5.
    2. Afshar N,
    3. English DR,
    4. Blakely T,
    5. Thursfield V,
    6. Farrugia H,
    7. Giles GG, et al.
    Differences in cancer survival by area-level socio-economic disadvantage: a population-based study using cancer registry data. PLoS One. 2020; 15: e0228551.
  6. 6.
    2. Afshar N,
    3. English DR,
    4. Milne RL.
    Factors explaining socio-economic inequalities in cancer survival: a systematic review. Cancer Control. 2021; 28: 10732748211011956.
  7. 7.↵
    2. Dalton SO,
    3. Olsen MH,
    4. Johansen C,
    5. Olsen JH,
    6. Andersen KK.
    Socioeconomic inequality in cancer survival – changes over time. A population-based study, Denmark, 1987-2013. Acta Oncol. 2019; 58: 737–44.
    OpenUrlPubMed
  8. 8.↵
    2. Yu XQ.
    Socioeconomic disparities in breast cancer survival: relation to stage at diagnosis, treatment and race. BMC Cancer. 2009; 9: 364.
    OpenUrlCrossRefPubMed
  9. 9.↵
    2. Andersson TM,
    3. Dickman PW,
    4. Eloranta S,
    5. Sjovall A,
    6. Lambe M,
    7. Lambert PC.
    The loss in expectation of life after colon cancer: a population-based study. BMC Cancer. 2015; 15: 412.
    OpenUrl
  10. 10.↵
    2. Syriopoulou E,
    3. Bower H,
    4. Andersson TM,
    5. Lambert PC,
    6. Rutherford MJ.
    Estimating the impact of a cancer diagnosis on life expectancy by socio-economic group for a range of cancer types in England. Br J Cancer. 2017; 117: 1419–26.
    OpenUrl
  11. 11.↵
    2. Wyper GMA,
    3. Devleesschauwer B,
    4. Mathers CD,
    5. McDonald SA,
    6. Speybroeck N.
    Years of life lost methods must remain fully equitable and accountable. Eur J Epidemiol. 2022; 37: 215–6.
    OpenUrlCrossRefPubMed
  12. 12.↵
    2. Dasgupta P,
    3. Andersson TM,
    4. Garvey G,
    5. Baade PD.
    Quantifying differences in remaining life expectancy after cancer diagnosis, Aboriginal and Torres Strait Islanders and other Australians, 2005-2016. Cancer Epidemiol Biomarkers Prev. 2022; 31: 1168–75.
    OpenUrl
  13. 13.↵
    2. Andersson TM,
    3. Dickman PW,
    4. Eloranta S,
    5. Lambe M,
    6. Lambert PC.
    Estimating the loss in expectation of life due to cancer using flexible parametric survival models. Stat Med. 2013; 32: 5286–300.
    OpenUrlCrossRefPubMed
  14. 14.↵
    2. Kou K,
    3. Dasgupta P,
    4. Aitken JF,
    5. Baade PD.
    Impact of area-level socioeconomic status and accessibility to treatment on life expectancy after a cancer diagnosis in Queensland, Australia. Cancer Epidemiol. 2020; 69: 101803.
  15. 15.↵
    2. Hardy D,
    3. Du DY.
    Socioeconomic and racial disparities in cancer stage at diagnosis, tumor size, and clinical outcomes in a large cohort of women with breast cancer, 2007-2016. J Racial Ethn Health Disparities. 2021; 8: 990–1001.
    OpenUrl
  16. 16.↵
  17. 17.↵
    Australian Institute of Health and Welfare. Cancer in Australia. Canberra: Australian Institute of Health and Welfare; 2021.
  18. 18.↵
    2. Rutherford MJ,
    3. Andersson TM,
    4. Björkholm M,
    5. Lambert PC.
    Loss in life expectancy and gain in life years as measures of cancer impact. Cancer Epidemiol. 2019; 60: 168–73.
    OpenUrl
  19. 19.↵
    Australian Bureau of Statistics. Australian Statistical Geography Standard (ASGS): volume 1 – main structure and Greater Capital City statistical areas. Canberra: Australian Bureau of Statistics; 2016.
  20. 20.↵
  21. 21.↵
    Australian Bureau of Statistics. Statistical area level 2, Australian Statistical Geographic Standard (ASGS) Edition 3. 2024. Available from: https://www.abs.gov.au/statistics/standards/australian-statistical-geography-standard-asgs-edition-3/jul2021-jun2026/main-structure-and-greater-capital-city-statistical-areas/statistical-area-level-2.
  22. 22.↵
    2. Kocarnik JM,
    3. Compton K,
    4. Dean FE,
    5. Fu W,
    6. Gaw BL,
    7. Harvey JD, et al.
    Cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted life years for 29 cancer groups from 2010 to 2019: a systematic analysis for the Global Burden of Disease Study 2019. JAMA Oncol. 2022; 8: 420–44.
    OpenUrl
  23. 23.↵
    2. Baade PD,
    3. Youlden DR,
    4. Andersson TM,
    5. Youl PH,
    6. Kimlin MG,
    7. Aitken JF, et al.
    Estimating the change in life expectancy after a diagnosis of cancer among the Australian population. BMJ Open. 2015; 5: e006740.
  24. 24.↵
    2. Nelson CP,
    3. Lambert PC,
    4. Squire IB,
    5. Jones DR.
    Flexible parametric models for relative survival, with application in coronary heart disease. Stat Med. 2007; 26: 5486–98.
    OpenUrlCrossRefPubMed
  25. 25.↵
    2. Lambert PC,
    3. Royston P.
    Further development of flexible parametric models for survival analysis. Stata J. 2009; 9: 265–90.
    OpenUrlCrossRef
  26. 26.↵
    2. Baade PD,
    3. Youlden DR,
    4. Andersson TM,
    5. Youl PH,
    6. Walpole ET,
    7. Kimlin MG, et al.
    Temporal changes in loss of life expectancy due to cancer in Australia: a flexible parametric approach. Cancer Causes Control. 2016; 27: 955–64.
    OpenUrl
  27. 27.↵
    2. Botta L,
    3. Dal Maso L,
    4. Guzzinati S,
    5. Panato C,
    6. Gatta G,
    7. Trama A, et al.
    Changes in life expectancy for cancer patients over time since diagnosis. J Adv Res. 2019; 20: 153–9.
    OpenUrl
  28. 28.↵
    2. Kou K,
    3. Dasgupta P,
    4. Cramb SM,
    5. Yu XQ,
    6. Andersson TM,
    7. Baade PD.
    Temporal trends in loss of life expectancy after a cancer diagnosis among the Australian population. Cancer Epidemiol. 2020; 65: 101686.
  29. 29.↵
    2. Devasia TP,
    3. Howlader N,
    4. Dewar RA,
    5. Stevens JL,
    6. Mittu K,
    7. Mariotto AB.
    Increase in the Life Expectancy of Patients with Cancer in the United States. Cancer Epidemiol Biomarkers Prev. 2024; 33: 196–205.
    OpenUrl
  30. 30.↵
    Registry of Births Deaths and Marriages. The coroners and the National Coronial Information System, Unit record mortality data for Australia by area of usual residence Brisbane. 2024. Available from: https://www.qld.gov.au/law/births-deaths-marriages-and-divorces/data/national-data.
  31. 31.↵
    2. Rachet B,
    3. Maringe C,
    4. Woods LM,
    5. Ellis L,
    6. Spika D,
    7. Allemani C.
    Multivariable flexible modelling for estimating complete, smoothed life tables for sub-national populations. BMC Public Health. 2015; 15: 1240.
    OpenUrlCrossRefPubMed
  32. 32.↵
    2. Papageorge MV,
    3. Woods AP,
    4. de Geus SW,
    5. Ng SC,
    6. McAneny D,
    7. Tseng JF, et al.
    The persistence of poverty and its impact on cancer diagnosis, treatment and survival. Ann Surg. 2023; 277: 995–1001.
    OpenUrl
  33. 33.↵
    2. Celeste RK,
    3. Fritzell J.
    Do socioeconomic inequalities in pain, psychological distress and oral health increase or decrease over the life course? Evidence from Sweden over 43 years of follow-up. J Epidemiol Community Health. 2018; 72: 160–7.
    OpenUrlAbstract/FREE Full Text
  34. 34.↵
    2. Andersson TM,
    3. Rutherford MJ,
    4. Lambert PC.
    Illustration of different modelling assumptions for estimation of loss in expectation of life due to cancer. BMC Med Res Methodol. 2019; 19: 145.
    OpenUrl
  35. 35.↵
    2. Hinchliffe SR,
    3. Dickman PW,
    4. Lambert PC.
    Adjusting for the proportion of cancer deaths in the general population when using relative survival: a sensitivity analysis. Cancer Epidemiol. 2012; 36: 148–52.
    OpenUrlPubMed
  36. 36.↵
    2. Sarfati D,
    3. Blakely T,
    4. Pearce N.
    Measuring cancer survival in populations: relative survival vs cancer-specific survival. Int J Epidemiol. 2010; 39: 598–610.
    OpenUrlCrossRefPubMed
  37. 37.↵
    2. Bright CJ,
    3. Brentnall AR,
    4. Wooldrage K,
    5. Myles J,
    6. Sasieni P,
    7. Duffy SW.
    Errors in determination of net survival: cause-specific and relative survival settings. Br J Cancer. 2020; 122: 1094–101.
    OpenUrl
Back to top

In this issue

Print
Download PDF
Email Article
Citation Tools

Jump to section

Keywords