The premature mortality to incidence ratio (PREMIER): a person-centred measure of cancer burden

David Banham, Jonathan Karnon, Alex Brown, David Roder, John Lynch
doi: https://doi.org/10.1101/2021.10.17.21265122

Abstract

Background Cancer control initiatives are informed by quantifying the capacity to reduce cancer burden through effective interventions. Burden measures using health administrative data are a sustainable way to support monitoring and evaluating of outcomes among patients and populations. The PREmature Mortality to IncidencE Ratio (PREMIER) is one such burden measure. We use data on Aboriginal and non-Aboriginal South Australians from 1990 to 2010 to show how PREMIER quantifies disparities in cancer burden: between populations; between sub-population cohorts where stage at diagnosis is available; and when follow-up is constrained to 24-months after diagnosis.

Method PREMIERcancer is the ratio of years of life expectancy lost due to cancer (YLLcancer) to life expectancy years at risk at time of cancer diagnosis (LYAR) for each person. The Global Burden of Disease standard life table provides referent life expectancies. PREMIERcancer was estimated for the population of cancer cases diagnosed in South Australia from 1990 to 2010. Cancer stage at diagnosis was also available for cancers diagnosed in Aboriginal people and a cohort of non-Aboriginal people matched by sex, year of birth, primary cancer site and year of diagnosis.

Results Cancers diagnoses (N=144,891) included 777 among Aboriginal people. Cancer burden described by PREMIERcancer was higher among Aboriginal than non-Aboriginal (0.55, 95%CIs 0.52-0.59 versus 0.39, 95%CIs 0.39-0.40). Diagnoses at younger ages among Aboriginal people, 7 year higher LYAR (31.0, 95%CIs 30.0-32.0 versus 24.1, 95%CIs 24.1-24.2) and higher premature cancer mortality (YLLcancer=16.3, 95%CIs 15.1-17.5 versus YLLcancer=8.2, 95%CIs 8.2-8.3) influenced this. Disparities in cancer burden between the matched Aboriginal and non-Aboriginal cohorts manifested 24-months after diagnosis with PREMIERcancer 0.44, 95%CIs 0.40-0.47 and 0.28, 95%CIs 0.25-0.31 respectively.

Conclusion PREMIER described disproportionately higher cancer burden among Aboriginal people in comparisons involving: all people diagnosed with cancer; the matched cohorts; and, within groups diagnosed with same staged disease. The extent of disparities were evident 24-months after diagnosis. This is evidence of Aboriginal peoples’ substantial capacity to benefit from cancer control initiatives, particularly those leading to earlier detection and treatment of cancers. PREMIER’s use of readily available, person-level administrative records can help evaluate health care initiatives addressing this need.

Background

Cancer is a leading cause of death and premature death globally [1, 2]. In Australia, cancer remains the largest contributor to years of life prematurely lost (YLL) despite the age standardised burden per head of population having declined by 11% from 2003 to 2011 [3]. Average burden may mask disparate trends in outcomes between and within populations [4, 5]. In the case of Aboriginal Australians (where “Aboriginal” is respectfully used to refer to people self-identifying as Aboriginal, Torres Strait Islander, or both [6]) comparable age-adjusted YLL were initially higher (52 versus 35 YLL per 1,000 population in 2003) and further increased to 55 versus 31 YLL per 1,000 population by 2013. This higher fatal burden is influenced by comparatively greater incidence of cancers with poor survival [5, 7, 8], diagnoses at more advanced stage [9-11], lower exposure to cancer treatment [9, 12], and excess case fatality concentrated in the first two-years after diagnosis [13]. Each of these influences suggest an unmet capacity to benefit from cancer control initiatives and actions including augmented cancer screening programs and addressing variations in treatment [14-16]. Such interventions need to be accompanied by relevant performance measures; measures which ensure system accountability [17], first by articulating disparity, then quantifying the capacity to benefit from prevention, early detection and intervention.

At a macro level, performance measures for population cancer outcomes [18] usually use relative survival [7, 19]. Relative survival is the ratio of observed survival among a group of people diagnosed with cancer and the expected survival of a similar, disease free group in the general population [20]. However, that method’s use can be severely limited for sub-populations of particular interest [7, 21, 22] or greatest need [22] where life tables detailing the background probabilities of death are not routinely available [23]. Such is the case with Aboriginal Australians, particularly at state and territory levels [7, 24]. An alternative is to use the Mortality to Incidence Ratio (MIR) which is the ratio of the observed cancer mortality and incidence rates in a given population in a specified time period [25, 26]. MIR is often used to illustrate disparate cancer outcomes between countries [27, 28] and the manner in which health system ranking [29] with components of cancer care such as cancer screening and treatment [28, 30-33], positively correlate with better, lower MIRs as illustrated in Figure 1 [27]. Australia’s health system is ranked thirty-second by the World Health Organization and has an average MIR of approximately 0.3, which is low by international standards and reflects well on Australia’s cancer control activities [34]. While less frequently used, MIR also describes cancer disparities within countries [35-37]. In this light, the favourable Australian average masks Aboriginal Australia’s poorer outcome of 0.5 [38].

Figure 1
Figure 1 Mortality to Incidence Ratio (MIR) by the World Health Organization’s Health System ranking (Top 100)

MIR has limited application for routine performance reporting for several reasons. As with life tables [7, 21, 22], routine and/or localised estimates for calculating population incidence and mortality rates may not be readily accessible. This is the case for Aboriginal Australians with Census estimates before 2016 labelled as ‘experimental’ and yearly population updates by age and smaller geographical areas not routinely published [39]. Consequently, data availability also limits the use of MIR [40] in quantifying opportunities to tailor initiatives to the needs of relevant sub-populations [41]. In addition, population [42] and cancer registrations [5, 43] available for performance monitoring often have time lags of two years or more before their release. This is sub-optimal because disparities in cancer outcome are manifest within 24-months of diagnosis [13]. Earlier signals on outcomes are needed if we are to evaluate the effects of system change in a timely manner [44, 45].

We respond to the need to further develop performance measurement in cancer control by revising MIR with the aim of increasing comparison between and within population sub-groups and without relying on infrequently available population parameters. We do so by employing a burden of disease method and measuring the time gap [46] of optimal life expectancy [47] remaining at two critical points in a person’s experience of cancer: the age of a person’s cancer diagnosis and death from cancer. Optimal life expectancy here refers to an international standard derived from the best observed mortality rates globally [48]. By adopting this method means we re-evaluate the MIR’s underlying parameters at the person level, then aggregate results for (sub)population groups.

Consequently, we introduce the PREmature Mortality to IncidencE Ratio (PREMIER), a metric that reframes MIR within a burden of disease method. After outlining PREMIER’s components and construction, we provide four analyses demonstrating its application. Analysis One focuses on general disparities in cancer burden existing between populations and uses cancers diagnosed among Aboriginal and non-Aboriginal Australians. Given these populations experience differences in age and primary site of cancers diagnosed [5, 8], Analysis Two adjusts for those confounding variables and quantifies disparity between Aboriginal people with cancer and a cohort of cancer cases drawn from the non-Aboriginal population having the same sex, year of birth, year of cancer diagnosis and primary site. Analysis Three enumerates differences in PREMIER within the Aboriginal and matched non-Aboriginal cohorts on the basis of cancer stage at diagnosis. To assess the extent to which disparities in cancer burden are evident soon after diagnosis, our final Analysis Four evaluates cancer burden between and within the matched cohorts 24-months after diagnosis. We then consider the implications and responses to observed disparities.

Methods

Study design and participants

We first provide a population context of all cancer cases [excluding non-melanoma skin cancer] diagnosed among South Australians in the period 1990 to 2010 (N=144,891). A nested retrospective, matched cohort design [9, 49] is used to compare cancers cases diagnosed among Aboriginal people (N=777) with a one-to-one random selection of cancer cases among non-Aboriginals matched on the basis of sex, year of birth, primary cancer site and year of diagnosis [8]. Follow-up time is from diagnosis date to date of death, or censoring or records at 31 December 2011, whichever occurred first.

Data sources, related measurements and definition of PREMIER

Cancer data for the South Australian population were obtained from the South Australian Cancer registry (SACR) [50] in the course of developing an advanced cancer data system within the Cancer Data and Aboriginal Disparities (CanDAD) project [51]. SACR is a population registry collating dates of International Classification of Diseases for Oncology (ICD-O-3) [52] coded diagnoses and death (attributed as cancer or non-cancer death). Specialist clinical cancer registry staff further enhanced the nested cohort study records using diagnostic and pathology records available to SACR to include cancer stage at diagnosis using Surveillance, Epidemiology, and End Results Program methodologies [53]. Stage at diagnosis categories included: localised - confined to tissue of origin; regional - invaded adjacent tissue or regional nodes; distant/unknown - spread to distant lymph nodes or other organ sites; leukaemia; or insufficient staging data were available.

MIR parameters of mortality and incidence are reframed within a burden of disease framework in the following manner. Mortality among cancer cases is quantified using YLL [54, 55], the amount of life expectancy remaining at time at which death attributed to cancer occurred. Incidence is quantified using expected Life Years at Risk (LYAR) [56], that is, the amount of life expectancy remaining at time at which cancer diagnosis occurred. Both YLL and LYAR represent the years of optimal life expectancy remaining at the age a given event occurs. That optimal life expectancy, which is subsequently used as a standard against which other measures are made, was previously derived for the global burden of disease study using the lowest age-specific risk of death observed in populations greater than 5 million individuals across the world (further details are available in Appendix Table 18, p503 [54]). In the case of YLL, the relevant event is the age at death while LYAR refers to age at diagnosis.

We make three assumptions in adopting those standard life expectancy estimates. First, we assume it is fair that all people aspire to optimal life expectancy because health differentials between sub-populations are influenced through societal and environmental risk factor exposures [47, 48] rather than fixed biological determinants aside from age. Second, we assume a uniform estimate of life expectancy across time, place and circumstance facilitates fair comparisons, regardless of changing geographic or sub-population specific mortality rates. We also assume a consistent method to deriving measures facilitates comparison between those measures and such comparisons are valuable.

PREMIER represents the amount of life expectancy lost as a fraction of life expectancy remaining at the time a sentinel health event is diagnosed. In the case of premature loss of life from cancer death after cancer diagnosis (PREMIERcancer), this is the ratio of years of life lost attributed to cancer (YLLcancer) to expected life years at risk at the time of cancer diagnosis (LYAR) represented as: Embedded Image

As a fraction of YLL and LYAR, PREMIER ranges from 0, where death after cancer diagnosis does not occur within the observation period, to 1, where death occurs at the same age as diagnosis. As an example, a person diagnosed with cancer at age 55 is taken as having 32.9 years of life expectancy remaining, thus LYAR is 32.9. Where death from cancer follows at age 65 the remaining life expectancy represents 23.8 years of life lost to cancer, YLLcancer. PREMIERcancer is 23.8 / 32.9, or 0.72, indicating that 72% of life expectancy at time of diagnosis was subsequently lost.

Individual PREMIER, and its LYAR and YLL components, can be grouped across population groups, or cohorts of people diagnosed with cancer. PREMIER can refer to a variety of observation periods. For instance, populations or cohorts may be observed for: varying periods from time of diagnosis to right-censoring of observations at a given date; a fixed period after cancer diagnosis; or, a combination of the two.

Statistical analysis

Under the heading of Risk, we summarise the mean age at cancer diagnosis and the accompanying LYAR. Subsequent Loss to premature mortality describes the number and mean age of deaths observed and attributed to cancer by SACR. Where deaths were not attributed to cancer, YLLcancer is zero. The Loss to Risk ratio, comprises the averaged PREMIERcancer for individuals within each group.

Table 1 includes three groups of cancer cases: the population of cancer cases diagnosed from 1990 to 2010 among non-Aboriginal South Australians; cancer cases diagnosed among Aboriginal South Australians in the same period; and, a matched cohort of cancer cases among non-Aboriginal people. Table 2 focuses on the Aboriginal and non-Aboriginal cohorts disaggregated by stage at diagnosis. Table 3 repeats this focus while limiting observation time to a maximum of 24-months after diagnosis.

View this table:
Table 1: Cancer diagnoses, premature mortality and PREMIERcancer, South Australia 1990-2010*
View this table:
Table 2: Cancer diagnoses, premature mortality and PREMIERcancer by stage at diagnosis, South Australia 1990-2010*
View this table:
Table 3: Cancer diagnoses, premature mortality and PREMIERcancer at 24-months by stage at diagnosis, South Australia 1990-2010*

Our multivariable analysis used the matched cohorts to evaluate the relationship between: PREMIERcancer at 24-months after diagnosis (PREMIERcancer 24-months) as the outcome with Aboriginality as the exposure and, cancer stage at diagnosis as a covariate. Interactions between Aboriginality and stage at diagnosis were also examined. We used fractional response regression [57], a quasi-likelihood estimation method available within Stata 15.1 as fracreg [58], and assumed a probit model for the conditional mean. This approach accommodates PREMIER’s attributes as: a fraction of two continuous quantities with life expectancy lost as numerator, life expectancy at time of diagnosis as denominator; having a denominator which is also the maximum value for the numerator; and, thus having values in the range of 0 to 1 inclusive. We clustered the data by the cohorts’ matched pairs and report 95% confidence intervals (95% CIs) based on robust standard errors. We report the modelled parameter coefficients which provide the sign of each covariate’s effect on PREMIERcancer 24-months. However, because the coefficients are difficult to interpret we also assessed the simultaneous average marginal effects of Aboriginality and stage at diagnosis on the proportion of life at risk lost in the 24-month period from diagnosis. That is, we report the change in PREMIERcancer 24-months where the cancer case involved an Aboriginal person rather than non-Aboriginal; and localised or distant stages rather than regional stage disease at diagnosis.

Results

Cancer burden between population groups

Table 1 shows SACR recorded 144,891 invasive cancer diagnoses among South Australians from 1990 to 2010. Cancer diagnoses among Aboriginal people accounted for a small number of those cases (N=777) and these are described in detail elsewhere [8]. Notably though, the latter cases were diagnosed at considerably younger age (57.7 years) compared to those among non-Aboriginal people (65.5 years). Consequently, life expectancy at risk at time of cancer diagnosis was almost 7 years higher among Aboriginal people with LYAR=31.0 (95% CIs 30.0-32.0) compared to the non-Aboriginal average of LYAR=24.1 (95%CIs 24.1-24.2). Proportionately more case fatalities, and at younger average age, were also observed among Aboriginal people with cancer. Taken together, average loss to premature mortality from cancer among Aboriginal cases was twice that of the broader group of non-Aboriginal cases (YLLcancer=16.3, 95%CIs 15.1-17.5 versus YLLcancer=8.2, 95%CIs 8.2-8.3). In turn, PREMIERcancer was markedly higher among Aboriginal compared to non-Aboriginal cases at 0.55 (95%CIs 0.52-0.59) versus 0.39 (95%CIs 0.39-0.40) respectively.

Cancer burden between and within matched cohorts

Table 1 also compares cases among Aboriginal people compared to a randomly selected cohort of diagnoses among non-Aboriginal cases (N=777) matched by sex, year of birth, year of diagnosis and primary cancer site. LYAR among the Aboriginal and non-Aboriginal cohort are therefore equivalent because of age matching. Fewer case fatalities at comparatively older ages among the non-Aboriginal cohort led to an average YLLcancer at 11.2 (95% CIs 10.1-12.3) and PREMIERcancer at 0.40 (95% CIs 0.37-0.44) which were markedly lower than their matched Aboriginal contemporaries with PREMIERcancer=0.55 (95%CIs 0.52-0.59). Indeed, PREMIERcancer for all non-Aboriginal and the subset of cases within the non-Aboriginal cohort were very similar (0.39, 95%CIs 0.39-0.40 and 0.40, 95% CIs 0.37-0.44 respectively).

Table 2 disaggregates Aboriginal and matched non-Aboriginal cohort results by stage at diagnosis. Cancers among Aboriginal people were more likely to involve distantly spread disease (n=333 or 42.8% of cases) than among non-Aboriginal people (n=255 or 32.8% of cases). Within each stage at diagnosis cancer case fatality was relatively more common among Aboriginal than non-Aboriginal people. Also, the average age at cancer death was lower among Aboriginal people than non-Aboriginal people diagnosed with regionally staged disease (58.9 versus 63.1 years) and distant staged disease (60.8 versus 63.2 years). Both factors contributed to markedly greater average YLLcancer in the Aboriginal cohort than the non-Aboriginal cohort with differences ranging from 2.0 (95%CIs 1.7-2.3) in localised stage to 6.2 (6.1-6.2) in regionally spread disease. For both cohorts, PREMIERcancer increased as cancer spread at diagnosis increased. However, PREMIERcancer also showed the relative amount of life at risk and subsequently lost was higher within the Aboriginal cohort at each stage of disease at diagnosis.

Cancer burden two years after diagnosis

Table 3 shows cohort outcomes up to two years after cancer diagnosis. Case fatality increased as stage at diagnosis increased from local to regional to distant stages with consistently higher loss observed among Aboriginal compared to non-Aboriginal people. Again, age at cancer death was younger among Aboriginal people than non-Aboriginal people for each stage at diagnosis. Average YLLcancer was also higher among Aboriginal cases at each stage of disease at diagnosis. Consequently, PREMIERcancer differed between cohorts 24-months after diagnosis with higher losses among Aboriginal (PREMIERcancer 24-months=0.44, 95%CIs 0.40-0.47) than non-Aboriginal (PREMIERcancer 24-months=0.28, 95%CIs 0.25-0.31). This difference of 0.16 in the limited 24-month follow-up period (using PREMIERcancer 24-months) was very similar to the difference of 0.15 observed across the full observation period (using PREMIERcancer).

PREMIERcancer 24-months also differed within cohorts and increased as stage at diagnosis increased. For example, point estimates for PREMIERcancer 24-months within the Aboriginal cohort increased from 0.17 in cases of localised disease to 0.68 where disease spread was distant or unknown, an overall change of 0.51. Overall change within the non-Aboriginal cohort was slightly less at 0.47 and ranged from 0.10 in localised disease to 0.57 in distant spread disease.

Multivariable analysis

Table 4 shows the association between life at risk and life subsequently lost up to 24-months after cancer diagnosis in the cohorts and the concurrent effects of Aboriginality and stage at diagnosis. Both Aboriginality and advancing disease stage at diagnosis were associated with higher PREMIERcancer. The model’s marginal effects indicate Aboriginal cases experienced an average of 0.10 or 10% (95%CIs 0.06-0.14) higher PREMIERcancer than non-Aboriginal cohort cases diagnosed with the same stage of disease. Simultaneously, and when compared to regionally spread disease at diagnosis, localised disease was associated with 0.21 or 21% (95%CIs 0.14-0.27) lower PREMIERcancer and distant/unknown spread with 0.27 or 27% (95%CIs 0.20-0.34) higher PREMIERcancer. No further interaction of the effects of Aboriginality by stage at diagnosis was evident.

View this table:
Table 4: Fractional outcome regression and average marginal effects on PREMIERcancer at 24-months, South Australia 1990-2010*

Discussion

PREMIER combines life expectancy at the time of cancer diagnosis and the resultant loss of life due to cancer death in order to quantify cancer burden. This is calculated for each person diagnosed with subsequent aggregation to groups. Our first analysis demonstrated PREMIER’s application in describing disparities in cancer burden for the entire population of invasive cancers diagnosed among South Australians. PREMIER described substantially higher cancer burden among the population of Aboriginal people with cancer compared to other South Australians (PREMIERcancer of 0.55 versus 0.39). These differences were bought about by Aboriginal South Australians with cancer having lower average age and more life expectancy (7 years) at risk of loss while also experiencing higher average premature mortality loss due to higher case fatality (59.3% versus 43.7%) and younger age at death (62 versus 72 years). Our second analysis focussed on Aboriginal and non-Aboriginal cohorts with equivalent sex, age, year of diagnosis and primary cancer site. While life expectancy at diagnosis was equivalent, PREMIER enumerated 15% more cancer burden among Aboriginal South Australians with cancer (PREMIERcancer of 0.55 versus 0.40). This was influenced by more frequent cancer deaths (59.3% versus 43.8%) and these deaths being at a younger age (61.5 versus 63.7 years). With the availability of stage at diagnosis for the cohorts, we then considered the variation of cancer burden within the cohorts. In both cohorts PREMIER increased as stage increased from local to regional to distant spread. In addition, PREMIER remained higher among Aboriginal people at each stage (PREMIERcancer=0.30 versus 0.22 for localised disease; 0.56 versus 0.41 for regional spread; and, 0.77 versus 0.68 for distant spread). These disparities by stage and Aboriginality were not only apparent for the broader observation period. They were fully manifested 24-months after diagnosis and our fourth analysis showed 16% higher cancer burden among Aboriginal than non-Aboriginal contemporaries (PREMIERcancer 24-months of 0.44 versus 0.28 respectively). Disparity of this size then continued across longer term observations.

Our analyses align with other reports of MIR, the ratio of observed cancer mortality and incidence rates in a given population in a specified time period, which describe intra-country disparities in cancer outcomes. For example, MIR differences between Black (MIR=0.48) and White (MIR=0.40) in South Carolina are clear [35, 37], yet recent differences between Aboriginal (MIR=0.51) and Australia generally (MIR=0.30) are even more pronounced [38]. These disparate results are echoed by PREMIER within the population of South Australians diagnosed with cancer where substantially more cancer burden among Aboriginal than non-Aboriginal (PREMIERcancer=0.55 versus 0.39 respectively) was quantified.

There are notable points of difference between MIR and PREMIER though. MIR makes use of mortality and incidence rates calculated on people diagnosed or dying in any given period. Those dying may have been diagnosed in different time periods meaning different groups of people are being compared [19]. One of the consequences of this back-scattering of incident cases is to make it difficult to observe rapid changes in prognosis [19]. PREMIER however, draws directly on each individual case for both numerator (LYAR) and denominator (YLL). Because incidence and mortality are observed within the same person the need to adjust for back-scattering is avoided. This is an advantage because it enables PREMIER to provide an earlier signal on cancer outcomes. Earlier measures can inform timely evaluations of system change, particularly system change aimed at improving outcomes within 24-months of diagnosis, a time when disparities are entrenched but also able to be detected using PREMIER.

PREMIER’s perspective on cancer burden is relevant to evidence-based policy development in cancer control [59] in other ways. For example, PREMIER’s estimation provides absolute measures of life at risk and life lost from cancer in a manner that is useful to planning activities. This is achieved by anchoring age at diagnosis and age at cancer death against a defined, optimal outcome. By describing disparities in age at diagnosis LYAR determined the amount of life expectancy amenable to change by preventing cancer, or at least deferring cancer incidence to later ages, through reduced exposure to cancer risks. As a relative measure, PREMIER revealed disparities across stage at diagnosis where more advanced disease led to higher cancer mortality and higher PREMIER. This information can help prioritise activities leading to earlier case detection and increased participation in cancer screening activities to detect cancers at an earlier stage. PREMIER also demonstrated an ability to enumerate disparities in cancer burden associated with stage and ethnicity 24-months after cancer diagnosis, a time during which people are more likely to be receiving care through health services [45]. This becomes particularly useful in supporting activities that promote access [60], uptake and quality [15, 61] of effective and available cancer treatments. In short, PREMIER enumerates people’s capacity to benefit from cancer control initiatives involving prevention, early detection and treatment and thus contributes to prioritising health system activities.

Similarly, while we report aggregated outcomes, it is important to remember PREMIER is calculated for each individually diagnosed case which become available for grouping and analysed in many configurations. We grouped observations by Aboriginality, however groups could be based on: shared area level geography; socio-economic position; or, by attending a certain service or receiving the care of particular provider. This adaptability is not only relevant to policy and planning but has further application in relating system performance to outcomes for individuals and the population groups to whom they belong [41]. PREMIER offers a robust and contemporary measure of performance with which to assess the effectiveness of early detection and treatment efforts. This is because PREMIER is free of the immediate need for background population information and time lags in reporting are reduced with counting and observations beginning as soon as diagnosis is made. This suggests the use of clinical records for reporting at patient (micro) and service (meso) levels in the first instance. As the underlying cancer and mortality records are integrated into population registries as we have used, macro level reporting for populations and the whole of system can follow. Information at these varying levels lend themselves to continued quality improvement processes and ongoing applied research. The use of existing, routine administrative data also helps address the evaluation needs of health services and government [62] while promoting public accountability [63]. Indeed, incorporating YLL within PREMIER facilitates comparison with other health system indicators and targets around reducing avoidable and premature mortality, particularly among vulnerable populations [63].

PREMIER has other strengths. Our analyses demonstrate the feasibility of assessing PREMIER using existing, routine, administrative and/or clinical records which also suggests it is readily sustainable. Other parameters from hospital systems could inform stratification within patient groups, for example, by stage at diagnosis. As cancer mortality outcomes improve and it becomes increasingly important to assess patient morbidity, the burden of disease method also provides for health adjusting the age relevant life expectancy and incorporating this into PREMIER estimates [56, 64]. In the meantime, PREMIER responds to the call for ever-increasing comparability and granularity in reporting [64] in two ways. We showed PREMIER’s comparability across populations and within small cohort groups. Further comparison with the wider Australian community, or even globally and for other time periods is quite possible because by measuring against the same, global standard. PREMIER has additional scope to generalise across conditions such as stroke or heart attack where there are definitive times of diagnosis enabling assessment of LYAR and subsequent YLL components. This would inform further comparison between and within people groups on the basis of health condition.

Limitations

PREMIER has several limitations. Interpreting relative outcome measures expressed as ratios which depend on different numerators and denominators is challenging. It is also a commonly occurring issue when considering issues of health disparity [65]. Our suggested response is to accompany PREMIER with reports of LYAR and YLL as absolute measures based on life expectancy. This raises the major limitation of PREMIER in that both LYAR and YLL are predicated on a global standard life table while local life expectancy for population groups of interest will likely be different. That is, PREMIER makes use of two biased measures and overestimates outcome disparities [66, 67] suggesting a prudent approach to its use as recommended with other survival methods [68]. The counter argument is to avoid bias by using population specific life tables [69-71]. However, life tables reflecting jurisdiction or group averages do not necessarily remedy the issue because such averages may mask considerable variation within the relevant jurisdictions or population group. For example, average life expectancy within one US county having the benefit of one of the highest observed life expectancies at birth was recently shown to subsume variations of up to 18 years among males and 15 years for females [72]. Nevertheless, when relevant life tables become available, the bias within our analysis can be approximated as done in other instances assessing the need for intra-country socio-economic position life tables [68]. Until such time though, our analysis makes use of the fall-back recommendation of using cancer specific mortality. This is justified because where health inequities exist, it is unacceptable to wait until complete information is to hand before acting. Therefore, we adopt an imperfect but well based and transparent method to quantifying health inequity by measuring against a gold standard, optimal outcome. In our case, this outcome is a standard attained by some but markedly less so by others within the same country and served by the same universal, healthcare system.

We further acknowledge our analysis of PREMIER did not account for the influence of comorbid conditions [73, 74]. In their own right, these are a major point of difference in the health status of Aboriginal and other Australians. However, PREMIER estimates for all-causes of death among people with cancer are easily calculated. Where higher risk of death from non-cancer causes are experienced [23] PREMIER estimates would increase and potentially exacerbate the disparities we documented. Other cancer survival studies do in fact report changes in the risk of death from cancer or non-cancer causes in the five years after cancer diagnosis [23] and this issue will benefit from further investigation.

Conclusion

We demonstrated PREMIER’s application in quantifying cancer burden disparities using Aboriginal and non-Aboriginal comparisons in South Australia. Cancer burden was markedly higher among Aboriginal people than non-Aboriginal in all comparisons based on: all people diagnosed with cancer; groups matched by sex, age, primary site and year of diagnosis; and, within groups experiencing similarly staged disease at diagnosis. Importantly, the extent of disparities were evident 24-months after diagnosis and persisted at similar levels thereafter. This points to a substantial capacity to benefit from improved cancer control initiatives among Aboriginal people, particularly those health system activities aimed at earlier detection and treatment of cancers. Our analyses also suggest PREMIER’s use of readily available, person-level information can provide important information helping evaluate person-centred cancer care as one dimension of high-quality health care delivery addressing this need.

Data Availability

The datasets generated and/or analysed during the current study are not publicly available due to privacy reasons, including the provisions of the Australian Privacy Principles. The study's data comprised of de-identified unit record administrative records and were used under privileged arrangements set out in a study specific confidentiality deed. The data cannot be accessed by another party without relevant data custodian and human research ethics approvals.

Declarations

Ethics approval and consent to participate

South Australia’s Aboriginal Health Research Ethics Committee (AHREC 04-12-461) and SA Health’s Human Research Ethics Committee (SA Health HREC HREC/12/SAH/35) approved the use of population cancer registry records. CanDAD’s Aboriginal Community Reference Group governance ensured alignment of the study protocol with South Australian Aboriginal Health Research Accord principles [75].

Consent for publication

Not applicable.

Availability of data and material

The datasets generated and/or analysed during the current study are not publicly available due to privacy reasons, including the provisions of the Australian Privacy Principles. The study’s data comprised of de-identified unit record administrative records and were used under privileged arrangements set out in a study specific confidentiality deed. The data cannot be accessed by another party without relevant data custodian and human research ethics approvals.

Competing interests

The authors declare they have no competing interests.

Funding

The CanDAD project was funded by National Health and Medical Research Council (APP1072243), Cancer Council SA’s Beat Cancer Project with project partners’ in-kind support. DB is supported in part by an Australian Government Research Training Program Scholarship. AB is supported by the Sylvia and Charles Viertel Senior Medical Research Fellowship. The funders were not involved in the analysis, data interpretation and writing of this manuscript.

Authors’ contributions

DB conceived the project, performed the analyses and drafted the manuscript; JL, JK, AB and DR made important contributions to operationalising this study, interpreting the statistical analysis, and revised the manuscript. All authors read and approved the final version of the manuscript.

Acknowledgements

Dr Graeme Tucker for statistical advice and the Cancer Data and Aboriginal Disparities (CanDAD) Aboriginal Community Reference Group (ACoRG), particularly Aunty Roslyn Weetra.

Footnotes

  • Email: jonathan.karnon{at}flinders.edu.au

  • Email: alex.brown{at}sahmri.com

  • Email: david.roder{at}unisa.edu.au

  • Email: john.lynch{at}adelaide.edu.au

Abbreviations

95% CIs
95% confidence intervals
CanDAD
Cancer Data and Aboriginal Disparities
LYAR
Life Years at Risk
PREMIER
PREmature Mortality to IncidencE Ratio calculated as YLL/LYAR for each case
PREMIERcancer
PREmature Mortality to IncidencE Ratio calculated as YLLcancer/LYAR for each case
PREMIERcancer 24-months
PREmature Mortality to IncidencE Ratio up to 24-months after diagnosis calculated as YLLcancer 24-months/LYAR for each case
PROMs
Patient Reported Outcome Measures
YLL
Years of Life Lost
YLLcancer
Years of Life Lost associated with cancer death
YLLcancer 24-months
Years of Life Lost associated with cancer death up to 24-months after diagnosis
SACR
South Australian Cancer Registry

References

  1. 1.↵
    Abajobir AA, Abate KH, Abbafati C, Abbas KM, Abd-Allah F, Abdulkader RS, Abdulle AM, Abebo TA, Abera SF, Aboyans V, et al: Global, regional, and national disability-adjusted life-years (DALYs) for 333 diseases and injuries and healthy life expectancy (HALE) for 195 countries and territories, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016. The Lancet 2017, 390: 1260–1344.
    OpenUrlCrossRef
  2. 2.↵
    Global Burden of Disease Cancer C: Global, regional, and national cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted life-years for 32 cancer groups, 1990 to 2015: A systematic analysis for the global burden of disease study. JAMA Oncology 2017, 3: 524–548.
    OpenUrl
  3. 3.↵
    Australian Institute of Health and Welfare: Australian Burden of Disease Study: Impact and causes of illness and death in Australia 2011.. Canberra: AIHW; 2016.
  4. 4.↵
    Cancer in Aboriginal & Torres Strait Islander people of Australia [https://www.aihw.gov.au/reports/cancer/cancer-in-indigenous-australians/report-editions]
  5. 5.↵
    Australian Institute of Health and Welfare: Cancer in Aboriginal and Torres Strait Islander peoples of Australia: An overview. Canberra: AIHW; 2013.
  6. 6.↵
    Health Performance Council Secretariat: Monitoring the implementation of Transforming Health Indicator Report. Adelaide: Government of South Australia; 2017.
  7. 7.↵
    Condon JR, Zhang X, Baade P, Griffiths K, Cunningham J, Roder DM, Coory M, Jelfs PL, Threlfall T: Cancer survival for Aboriginal and Torres Strait Islander Australians: a national study of survival rates and excess mortality. Population Health Metrics 2014, 12.
  8. 8.↵
    Banham D, Roder D, Keefe D, Farshid G, Eckert M, Cargo M, Brown A, for the CanDAD Aboriginal Community Reference Group and other CanDAD investigators: Disparities in cancer stage at diagnosis and survival of Aboriginal and non-Aboriginal South Australians. Cancer Epidemiology 2017, 48: 131–139.
    OpenUrl
  9. 9.↵
    Valery P, Coory M, Stirling J, Green A: Cancer diagnosis, treatment, and survival in Indigenous and non-Indigenous Australians: a matched cohort study. Lancet 2006, 367: 1842–1848.
    OpenUrlCrossRefPubMedWeb of Science
  10. 10.
    Condon J, Cunningham J, Barnes A, Armstrong B, Selva-Nayagam S: Cancer diagnosis and treatment in the Northern Territory - assessing health service performance for Indigenous Australians. Intern Med J 2006, 36: 498–505.
    OpenUrlPubMed
  11. 11.↵
    Roder D: Comparative cancer incidence, mortality and survival in Indigenous and non-Indigenous residents of South Australia and the Northern Territory. Cancer Forum 2005, 29: 7–9.
    OpenUrl
  12. 12.↵
    Condon J, Armstong B, Barnes A, Cunningham J: Cancer in Indigenous Australians: a review. Cancer Causes & Control 2003, 14: 109–121.
    OpenUrlCrossRefPubMedWeb of Science
  13. 13.↵
    Cramb S, Garvey G, Valery P, Williamson J, Baade P: The first year counts: cancer survival among Indigenous and non-Indigenous Queenslanders, 1997-2006. Med J Aust 2012, 196: 270–274.
    OpenUrlCrossRefPubMed
  14. 14.↵
    Adelson P, Fusco K, Karapetis C, Wattchow D, Joshi R, Price T, Sharplin G, Roder D: Use of guideline-recommended adjuvant therapies and survival outcomes for people with colorectal cancer at tertiary referral hospitals in South Australia. Journal of Evaluation in Clinical Practice 2017:n/a-n/a.
  15. 15.↵
    Liu JB, Ko CY: Disparities in Rectal Cancer: Moving from Descriptions to Solutions. Annals of Surgical Oncology 2017, 24: 299–301.
    OpenUrl
  16. 16.↵
    Woods LM, Rachet B, Coleman MP: Origins of socio-economic inequalities in cancer survival: a review. Annals of Oncology 2006, 17: 5–19.
    OpenUrlCrossRefPubMedWeb of Science
  17. 17.↵
    The Lancet: What’s next for Indigenous health in Australia? The Lancet 2018, 391: 632.
    OpenUrl
  18. 18.↵
    Clauser SB: Use of Cancer Performance Measures in Population Health: A Macro-level Perspective. JNCI Monographs 2004, 2004: 142–154.
    OpenUrlCrossRefPubMed
  19. 19.↵
    Ellis L, Woods LM, Esteve J, Eloranta S, Coleman MP, Rachet B: Cancer incidence, survival and mortality: explaining the concepts. Int J Cancer 2014, 135: 1774–1782.
    OpenUrlCrossRefPubMed
  20. 20.↵
    Ederer F, Axtell LM, Cutler SJ: The relative survival rate: A statistical methodology. National Cancer Institute Monograph 1961, 6: 101–121.
    OpenUrlPubMed
  21. 21.↵
    Mariotto AB, Noone A-M, Howlader N, Cho H, Keel GE, Garshell J, Woloshin S, Schwartz LM: Cancer Survival: An Overview of Measures, Uses, and Interpretation. JNCI Monographs 2014, 2014: 145–186.
    OpenUrlCrossRefPubMed
  22. 22.↵
    Morris M, Woods LM, Rachet B: A novel ecological methodology for constructing ethnicmajority life tables in the absence of individual ethnicity information. Journal of Epidemiology and Community Health 2015, 69: 361–367.
    OpenUrlAbstract/FREE Full Text
  23. 23.↵
    He VY, Condon JR, Baade PD, Zhang X, Zhao Y: Different survival analysis methods for measuring long-term outcomes of Indigenous and non-Indigenous Australian cancer patients in the presence and absence of competing risks. Popul Health Metr 2017, 15: 1.
    OpenUrlCrossRef
  24. 24.↵
    Australian Bureau of Statistics: Life tables for Aboriginal and Torres Strait Islander Australians: 2010-2012. Canberra: ABS; 2013.
  25. 25.↵
    Suwanrungruang K, Sriplung H, Temiyasathit S, Waisri N, Daoprasert K, Kamsa-Ard S, Tasanapitak C, McNeil E: Appropriateness of the Standard Mortality/Incidence Ratio in Evaluation of Completeness of Population-Based Cancer Registry Data. Asian Pacific Journal of Cancer Prevention 2011, 12: 3283–3288.
    OpenUrl
  26. 26.↵
    Asadzadeh Vostakolaei F, Karim-Kos HE, Janssen-Heijnen MLG, Visser O, Verbeek ALM, Kiemeney LALM: The validity of the mortality to incidence ratio as a proxy for sitespecific cancer survival. European Journal of Public Health 2011, 21: 573–577.
    OpenUrlCrossRefPubMedWeb of Science
  27. 27.↵
    Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M, Parkin DM, Forman D, Bray F: Cancer incidence and mortality worldwide: Sources, methods and major patterns in GLOBOCAN 2012. International Journal of Cancer 2015, 136: E359–E386.
    OpenUrlCrossRefPubMedWeb of Science
  28. 28.↵
    Choi E, Lee S, Nhung BC, Suh M, Park B, Jun JK, Choi KS: Cancer mortality-to-incidence ratio as an indicator of cancer management outcomes in Organization for Economic Cooperation and Development countries. Epidemiology and Health 2017, 39: e2017006.
    OpenUrl
  29. 29.↵
    World Health Organization: The World Health Report 2000: Health systems - Improving performance. pp. 1-177. Geneva: WHO; 2000:1–177.
  30. 30.↵
    Sunkara V, Hébert JR: The colorectal cancer mortality-to-incidence ratio as an indicator of global cancer screening and care. Cancer 2015, 121: 1563–1569.
    OpenUrl
  31. 31.
    Doubova SV, Pérez-Cuevas R: Going further to measure improvements in health-care access and quality. The Lancet 2018, 391: 2190–2192.
    OpenUrl
  32. 32.
    Fullman N, Yearwood J, Abay SM, Abbafati C, Abd-Allah F, Abdela J, Abdelalim A, Abebe Z, Abebo TA, Aboyans V, et al: Measuring performance on the Healthcare Access and Quality Index for 195 countries and territories and selected subnational locations: a systematic analysis from the Global Burden of Disease Study 2016. The Lancet 2018, 391: 2236–2271.
    OpenUrl
  33. 33.↵
    Chen S-L, Wang S-C, Ho C-J, Kao Y-L, Hsieh T-Y, Chen W-J, Chen C-J, Wu P-R, Ko J-L, Lee H, Sung W-W: Prostate Cancer Mortality-To-Incidence Ratios Are Associated with Cancer Care Disparities in 35 Countries. 2017, 7: 40003.
    OpenUrl
  34. 34.↵
    Cancer Australia: National Cancer Control Indicators (NCCI) Framework. Sydney: Cancer Australia; 2017.
  35. 35.↵
    Hébert JR, Daguise VG, Hurley DM, Wilkerson RC, Mosley CM, Adams SA, Puett R, Burch JB, Steck SE, Bolick-Aldrich SW: Mapping Cancer Mortality-to-Incidence Ratios to Illustrate Racial and Sex Disparities in a High-risk Population. Cancer 2009, 115: 2539–2552.
    OpenUrlCrossRefPubMedWeb of Science
  36. 36.
    Hashim D, Manczuk M, Holcombe R, Lucchini R, Boffetta P: Cancer mortality disparities among New York City’s Upper Manhattan neighborhoods. European Journal of Cancer Prevention 2017, 26: 453–460.
    OpenUrl
  37. 37.↵
    Babatunde OA, Adams SA, Eberth JM, Wirth MD, Choi SK, Hebert JR: Racial disparities in endometrial cancer mortality-to-incidence ratios among Blacks and Whites in South Carolina. Cancer Causes Control 2016, 27: 503–511.
    OpenUrl
  38. 38.↵
    Australian Cancer Incidence and Mortality (ACIM) books: All cancers combined [http://www.aihw.gov.au/acim-books]
  39. 39.↵
    Australian Bureau of Statistics: Estimates and Projections, Aboriginal and Torres Strait Islander Australians: 2001 to 2026. Canberra: ABS; 2014.
  40. 40.↵
    1. Jensen O,
    2. Parkin D,
    3. MacLennan R,
    4. Muir C,
    5. Skeet R. Lyon
    Parkin D, Hakulinen T: Analysis of survival. In Cancer Registration: Principles and Methods Edited by Jensen O, Parkin D, MacLennan R, Muir C, Skeet R. Lyon: International Agency for Research on Cancer; 1991: 159–176
  41. 41.↵
    Hostetter M, Klein S: Segmenting Populations to Tailor Services, Improve Care. The Commonwealth Fund; 2015.
  42. 42.↵
    Australian Bureau of Statistics: Estimates of Aboriginal and Torres Strait Islander Australians, June 2016. Released 31 August, 2018 edition. Canberra: ABS; 2018.
  43. 43.↵
    South Australian Cancer Registry: Cancer in South Australia 2015 - with projections to 2018. Adelaide, SA: South Australian Department for Health and Wellbeing; 2018.
  44. 44.↵
    Canadian partnership against cancer: Cancer Stage in Performance Measurement: A First Look. Toronto: The Partnership; 2015.
  45. 45.↵
    Chadder J, Dewar R, Shack L, Nishri D, Niu J, Lockwood G: A first look at relative survival by stage for colorectal and lung cancers in Canada. Current Oncology 2016, 23: 119–124.
    OpenUrl
  46. 46.↵
    Murray C: Quantifying the burden of disease: the technical basis for disability-adjusted life years. Bulletin of the World Health Organization 1994, 72: 429–445.
    OpenUrlPubMedWeb of Science
  47. 47.↵
    Murray CJ, Acharya AK: Understanding DALYs (disability-adjusted life years). J Health Econ 1997, 16: 703–730.
    OpenUrlCrossRefPubMedWeb of Science
  48. 48.↵
    Abajobir AA, Abbafati C, Abbas KM, Abd-Allah F, Abera SF, Aboyans V, Adetokunboh O, Afshin A, Agrawal A, Ahmadi A, et al: Global, regional, and national age-sex specific mortality for 264 causes of death, 1980–2016: a systematic analysis for the Global Burden of Disease Study 2016. The Lancet 2017, 390: 1151–1210.
    OpenUrl
  49. 49.↵
    Diaz A, Moore SP, Martin JH, Green AC, Garvey G, Valery PC: Factors Associated With Cancer-Specific and Overall Survival Among Indigenous and Non-Indigenous Gynecologic Cancer Patients in Queensland, Australia: A Matched Cohort Study. International Journal of Gynecological Cancer 2015, 25: 542–547.
    OpenUrlAbstract/FREE Full Text
  50. 50.↵
    South Australian Cancer Registry: Cancer in South Australia 2014 - with projections to 2017. Adelaide, SA: South Australian Department for Health and Ageing; 2017.
  51. 51.↵
    Yerrell PH, Roder D, Cargo M, Reilly R, Banham D, Micklem JM, Morey K, Stewart HB, Stajic J, Norris M, et al: Cancer Data and Aboriginal Disparities (CanDAD)—developing an Advanced Cancer Data System for Aboriginal people in South Australia: a mixed methods research protocol. BMJ Open 2016, 6.
  52. 52.↵
    Fritz A, Percy C, Jack A, Shanmugaratnam K, Sobin L, Parkin D, Whelan S (Eds.): International Classification of Diseases for Oncology, Third Edition edition. Geneva: World Health Organization; 2000.
  53. 53.↵
    Young JJ, Roffers S, Ries L, Fritz A, Hurlbut A, National Cancer Institute (Eds.): SEER Summary Staging Manual - 2000: Codes and Coding Instructions,. Bethesda, MD: National Cancer Institute; 2001.
  54. 54.↵
    GBD 2016 Causes of Death Collaborators: Global, regional, and national age-sex specific mortality for 264 causes of death, 1980–2016: a systematic analysis for the Global Burden of Disease Study 2016. The Lancet 2017, 390: 1151–1210.
    OpenUrl
  55. 55.↵
    Brustugun OT, Moller B, Helland A: Years of life lost as a measure of cancer burden on a national level. Br J Cancer 2014, 111: 1014–1020.
    OpenUrlCrossRefPubMed
  56. 56.↵
    Zhao Y, Malyon R: Life Years at Risk: A Population Health Measure from a Prevention Perspective. International Journal of Environmental Research and Public Health 2009, 6: 2387–2396.
    OpenUrl
  57. 57.↵
    Papke LE, Wooldridge JM: Econometric methods for fractional response variables with an application to 401(k) plan participation rates. Journal of Applied Econometrics 1996, 11: 619–632.
    OpenUrlCrossRefWeb of Science
  58. 58.↵
    StataCorp: Stata Statistical Software: Release 15.1. College Station, TX: StataCorp LP; 2017.
  59. 59.↵
    Sarfati D, Garvey G, Robson B, Moore S, Cunningham R, Withrow D, Griffiths K, Caron N, Bray F: Measuring cancer in Indigenous populations. Annals of Epidemiology 2018.
  60. 60.↵
    SCRGSP (Steering Committee for the Review of Government Service Provision): Report on Government Services 2018. Canberra: Productivity Commission; 2018.
  61. 61.↵
    Whitehead M: The concepts and principles of equity and health. Copenhagen: World Health Organization; 1990.
  62. 62.↵
    Lokuge K, Thurber K, Calabria B, Davis M, McMahon K, Sartor L, Lovett R, Guthrie J, Banks E: Indigenous health program evaluation design and methods in Australia: a systematic review of the evidence. Australian and New Zealand Journal of Public Health 2017, 41: 480–482.
    OpenUrl
  63. 63.↵
    Health Performance Council: What’s working, what’s not: Review of the South Australian Health System Performance for 2011–2014. Adelaide: Health Performance Council; 2014.
  64. 64.↵
    Murray CJL, Lopez AD: Measuring global health: motivation and evolution of the Global Burden of Disease Study. The Lancet 2017, 390: 1460–1464.
    OpenUrl
  65. 65.↵
    Harper S, Lynch J: Methods for Measuring Cancer Disparities: Using Data Relevant to Healthy People 2010 Cancer-Related Objectives.. Bethesda, MD: National Cancer Institute; 2005.
  66. 66.↵
    Blakely T, Soeberg M, Carter K, Costilla R, Atkinson J, Sarfati D: Bias in relative survival methods when using incorrect life-tables: Lung and bladder cancer by smoking status and ethnicity in New Zealand. International Journal of Cancer 2012, 131: E974–E982.
    OpenUrlCrossRefPubMed
  67. 67.↵
    Grafféo N, Giorgi R: Non-Matching Life Tables and Impact. In Satellite Symposium European Partnership for Action Against Cancer (EPAAC). Ispra; 2014.
  68. 68.↵
    Dickman PW, Auvinen A, Voutilainen ET, Hakulinen T: Measuring social class differences in cancer patient survival: is it necessary to control for social class differences in general population mortality? A Finnish population-based study. Journal of Epidemiology and Community Health 1998, 52: 727–734.
    OpenUrlAbstract
  69. 69.↵
    Withrow DR, Racey CS, Jamal S: A critical review of methods for assessing cancer survival disparities in indigenous population. Annals of Epidemiology 2016, 26: 579–591.
    OpenUrl
  70. 70.
    Withrow DR, Pole JD, Nishri ED, Tjepkema M, Marrett LD: Choice of relative or causespecific approach to cancer survival analysis impacts estimates differentially by cancer type, population, and application: evidence from a Canadian population-based cohort study. Population Health Metrics 2017, 15: 24.
    OpenUrl
  71. 71.↵
    Howlader N, Ries LAG, Mariotto AB, Reichman ME, Ruhl J, Cronin KA: Improved Estimates of Cancer-Specific Survival Rates From Population-Based Data. JNCI: Journal of the National Cancer Institute 2010, 102: 1584–1598.
    OpenUrl
  72. 72.↵
    Dwyer-Lindgren L, Stubbs RW, Bertozzi-Villa A, Morozoff C, Callender C, Finegold SB, Shirude S, Flaxman AD, Laurent A, Kern E, et al: Variation in life expectancy and mortality by cause among neighbourhoods in King County, WA, USA, 1990–2014: a census tractlevel analysis for the Global Burden of Disease Study 2015. Lancet Public Health 2017, 2: 400–410.
    OpenUrl
  73. 73.↵
    Sheppard AJ, Chiarelli AM, Marrett LD, Nishri ED, Trudeau ME: Stage at Diagnosis and Comorbidity Influence Breast Cancer Survival in First Nations Women in Ontario, Canada. Cancer Epidemiology Biomarkers & Prevention 2011, 20: 2160–2167.
    OpenUrlAbstract/FREE Full Text
  74. 74.↵
    Santorelli ML, Hirshfield KM, Steinberg MB, Lin Y, Rhoads GG, Bandera EV, Demissie K: Racial differences in the effects of comorbidity on breast cancer-specific survival. Cancer Causes & Control 2017, 28: 809–817.
    OpenUrl
  75. 75.↵
    Wardliparingga Aboriginal Research Unit: South Australian Aboriginal Health Research Accord: Companion Document. Adelaide: South Australian Health and Medical Research Institute; 2014.
Back to top
Posted October 18, 2021.
Download PDF
Data/Code
Email
The premature mortality to incidence ratio (PREMIER): a person-centred measure of cancer burden
David Banham, Jonathan Karnon, Alex Brown, David Roder, John Lynch
medRxiv 2021.10.17.21265122; doi: https://doi.org/10.1101/2021.10.17.21265122
Twitter logo Facebook logo LinkedIn logo Mendeley logo
Citation Tools

Subject Area