Skip to main content
NCBI home page
American Journal of Epidemiology logoLink to American Journal of Epidemiology
. 2025 Aug 19;194(11):3316–3324. doi: 10.1093/aje/kwaf148

The impact of cancer survivors’ extra risk of noncancer mortality on net survival estimation

Laura Botta 1,2,✉, Riccardo Capocaccia 3, Alice Bernasconi 4, Silvia Rossi 5, Jaume Galceran 6,7, Luigino Dal Maso 8, Come Lepage 9,10, Florence Molinié 11,12,13, Anne-Marie Bouvier 14,15,16, Rafael Marcos-Gragera 17,18, Claudia Vener 19,20, Marcela Guevara 21,22,23, Deirdre Murray 24,25, Rosalia Ragusa 26, Gemma Gatta 27, Valerie Jooste 28,29,30; the EUROCARE-6 WG
PMCID: PMC12634115  PMID: 40826891

Abstract

Relative survival (RS) with the general population as the reference is commonly used to estimate net survival (NS). However, cancer patients may face an increased risk of noncancer death compared to cancer-free individuals. We evaluate the impact of considering this relative risk (RR) on NS estimation. First, we compared selected NS values to generated RS under various theoretical scenarios, considering different RR, NS, age at diagnosis, time since diagnosis, and sex. Then, differences between NS and RS for 3 cancers were analyzed from cure model-based estimates using EUROCARE-6 data. We observed differences between RS and the true value of NS, larger for longer time since diagnosis, older patients, and higher NS. For head and neck cancer, the smallest differences were for young female patients at 5 years from diagnosis (4%) and the highest (32%) for older patients. For colorectal cancer, differences were <7% for all ages, both sexes, and times since diagnosis and for breast cancer, differences were <5% except for older patients after 5 years. If RR > 1, RS underestimates NS. Our findings aim to correctly interpret the differences between RS and NS, and contextualize the possible biases of assuming RS as a proxy for cancer-specific survival.

Keywords: relative survival, net survival, population-based cancer registry, relative risk of noncancer death for cancer patients compared to general population

Introduction

In the framework of population-based studies, net survival (NS) has been set to estimate cancer-specific survival among cancer patients. Net survival was defined as the probability that patients survive their cancer in the absence of other causes of death, that is, “the hypothetical situation where the disease under study would be the only possible cause of death”.1 Being unaffected by the mortality from other causes, NS is the indicator of choice for cancer survival comparisons between cohorts of patients with different backgound mortality, for example, comparison between countries.1–3 Net survival can be estimated using the ratio of the observed survival (all-cause survival) of the patients’ cohort to the expected survival of a set of cancer-free individuals matched for age, sex, and calendar year (RS) or using the excess mortality framework (Pohar Perme—PP). Both these methods for estimating NS do not require knowledge on cause of death but expected survival in absence of cancer is required. As this is difficult to obtain, life tables representing survival of the general population to which the patients belong are used instead. Those methods are considered estimators of NS under the assumption of equal risk of death from other causes in the patients and the general population.4 All NS estimators used so far (eg, Ederer 2, Pohar Perme) consider the general population lifetables as reference mortality rates in absence of cancer. The aim of this paper is to address the bias resulting from this last assumption.

There is increasing evidence that cancer patients may have a different and often higher risk of dying from other diseases compared to the general population,5–8 whether caused (eg, adverse effects of treatments) or not (eg, independent second cancer, chronic disease related to the cancer risk factors) by the diagnosed cancer. In this case, NS estimators taking the general population as the reference group tends to diverge from NS.9,10 This divergence depends on the cancer type, the length of follow-up, age at diagnosis, and sex.

The concept is illustrated in Figure 1, when the total death hazard of cancer patients at a given instant is conceptually broken down into 4 components. The bar A represents the hazard of a general population matched to the patients by age, calendar year, residence area, and sex; the B and C bars represent the hazard of death from other causes (including other cancers) associated, respectively, to patients’ additional risk factors (such as smoke, obesity, genetical, etc.) and to side effects of cancer treatment. The bar D is the specific death due to the cancer progression. The total hazard is considered in the calculation of overall survival. The NS estimators used so far are usually based on the hazard components B + C + D (ie, after removal of the general population hazard) and provide the most appropriate, objective, and easily estimable measure available for survival comparisons between population groups. Hazard components C and D refer to deaths that would not have occurred in absence of the considered cancer, and their combination gives then the most appropriate epidemiological indicators for estimating cancer burden. However, the clinical pathways deriving from risks C and D, for example, cardiac disorders due to treatment versus relapse and metastatic spread of the originally diagnosed cancer, are very different. Net survival considers only the risk of death due to cancer progression (D) and is therefore the most clinically oriented survival indicator and the most direct measure to consider when planning post-treatment clinical surveillance, and organization of palliative care structures.

Figure 1.

Figure 1

Open in a new tab

Total hazard of death by cause across three populations: general population, matched cancer-free population, and cancer patients. This figure illustrates the overall hazard of death and its internal composition: A) Hazard of death in the general population matched to cancer patients by age, calendar year, geographic area, and sex. B) Hazard of death from other causes (including other cancers) attributable to additional risk factors in cancer patients (e.g., smoking, obesity, genetic predispositions). C) Hazard of death from other causes related to side effects of cancer treatments. D) Hazard of death directly due to cancer progression.

The multicomponent structure of the cancer patients’ hazard has only been addressed sporadically, probably due to the practical difficulty of disentangling mortality from cancer and mortality from other causes in population-based setting. Several authors concluded that the bias caused by the noncomparability assumption using RS in the estimates of NS is negligible, for many cancer sites.10–12 However, recent findings based on a modeling approach suggested a greater impact than previously deemed.13,14 These papers provided for the first time the NS estimates using a new approach aimed at excluding the extra risk of death from other causes that affect cancer patients.

The purpose of the present work is to explore the bias occurring in NS estimation when using RS under a wrong assumption of equal risk of death from other causes in the patients and in the general population. We explore the dimension of this bias, first, under a set of theoretical but plausible scenarios and, second, from a real-world example derived from a model-based analysis of head and neck (H&N), colorectal and breast cancer in Europe. The final aim is to draw attention on the difference between RS and NS and to contextualize possible biases arising from the assumption of RS as a proxy for cancer-specific survival.

Methods

So far in the NS framework, it is assumed that instantaneous risk of dying (ie, hazard) of cancer patients is given by the sum of the hazard of death due to cancer and the expected hazard (due to other causes), which is given by the hazard observed in the general population. Suppose however, that, unlike to the standard approach, the latter is given by the hazard of the general population multiplied by a relative risk (RR) of death from causes other than the primary cancer, undergone by cancer patients compared to the general population. In symbols:

graphic file with name DmEquation1.gif

Where h are mortality hazards, respectively: hO observed, hC due to cancer, hE expected (due to other causes) and hP population hazards. The expression is intended to hold for a specific pattern of covariates, such as age at diagnosis or sex, that do not affect the subsequent considerations and are omitted for simplicity of notation.

On the survival scale, we write

graphic file with name DmEquation2.gif

Where OS(t), NS(t), and PS(t) are respectively the observed, net, and population survival functions of time t since diagnosis. NS(t) is the true estimation of NS, under the assumption of a RR different than 1. In terms of relative survival RS(t), we have:

graphic file with name DmEquation3.gif (1)

The above expressions define the relationship between NS (ie, survival associated to cancer death after accounting for excess mortality by other causes than cancer) and RS (ie, survival relative with respect to the general population).

Theoretical scenario

The right side of expression (1) will be used to calculate the RS values estimated in different scenarios generated combining several values of RR (0.9, 1.1, 1.5, 2, 3), levels of NS(t) (20%, 40%, 60%, 80%), age at diagnosis (50, 60, and 70), sex, and follow-up time (5, 10, and 15 years). The RR values were chosen as being sensible and plausible in relation to the real data analyzed so far.14,15 PS(t) used for this theoretical evaluation was the European population 2017 lifetable as provided by EUROSTAT (https://ec.europa.eu/eurostat/databrowser/view/demo_mlifetable__custom_11976961/default/table?lang=en accessed June 9, 2024). Note that when we refer to theoretical RS in this work, we mean the RS calculated using the listed NS, RR, and PS theoretical values.

Real data

The difference between NS and RS will be also highlighted using real-world data, from a recently published survival analysis of Eurocare-6 colorectal, H&N, and female breast cancers survival.14 The RS estimates were obtained by a mixture cure model described in detail by Botta et al.16 on data from patients aged 40-79, diagnosed in 1998-2002 and followed up to December 31, 2014. Data were provided by 65 population-based cancer registry from 20 European countries (Austria, Bulgaria, Czechia, Denmark, Estonia, Finland, Ireland, Lithuania, Malta, Netherlands, Slovenia, Slovakia, Iceland, Norway, United Kingdom, Italy, Spain, Switzerland, France, and Germany) participating in the EUROCARE-6 project.

In summary, we calculated for each sex the actuarial life tables with 15 yearly follow-up intervals and by 5-year age classes. Point estimated of observed RS were obtained with the Ederer 2 methods2 from overall survival and using the general population life tables as providing the expected mortality of the cancer patients’ cohort. Model-based RS was obtained by fitting the RS point estimates with a new mixture cure model16:

graphic file with name DmEquation4.gif (2)

where t is time from diagnosis, x is age at diagnosis, π(x) = [1+ exp(−ϕ(x))]−1 indicates the cure fraction (CF), that is, the proportion of statistically cured patients, as a function of age (−ϕ(x) with the logistic link), and Su(Inline graphic) represents the RS function of the uncured patients, PS(Inline graphic) the expected survival of a comparable group in the general population between ages x and Inline graphic, and α is an estimator of RR, the relative risk of death of patients, compared to the general population, from causes other than the primary cancer. Su(Inline graphic) follows a Weibull distribution for colon and rectum and female breast and Lognormal for H&N cancers.14 The expression in square brackets in the equation represents NS, and is the model-based equivalent of expression (1). This indicator is the main finding of the paper.

All analyses and predictions were performed using Stata version 17.

Results

Table 1 shows the theoretical RS values by sex deriving from the European general population life tables and 4 selected theoretical NS values (20%, 40%, 60%, and 80%). Relative survival was calculated by expression (1) at 5, 10, and 15 years after diagnosis, by patients aged 50, 60, and 70 years at diagnosis, and exposed to various RR of death from other causes with respect to the population. The shades of gray indicate the difference between the theoretical values of RS and NS. For example, if RR = 1.5, the theoretical RS after 5 years of a male patient diagnosed with a cancer at age 70 was 56% compared to an NS of 60%. Furthermore, if the same 60% level of NS was reached at 10 and 15 years, the corresponding theoretical RS was as low as 49% and 40%, respectively. The absolute differences between the 2 indicators increased with increasing NS level, age, time from diagnosis, and RR value. It was lower for female cancer patients than for male cancer patients. When RR was as small as 1.1 or 0.9, little differences, less than 5 in absolute values, appeared at 5 years since diagnosis from 70 years of age or at 10-15 years since diagnosis for younger patients. Differences between 5 and 15 percentage points were present with RR = 1.5 and were seen in oldest men and in long-term survival. Differences of 15% points or more were found for RR = 2 and even more for RR = 3. In these cases, theoretical RS could become, for advanced ages and long time since diagnosis, as low as half of the corresponding NS. It should be noted that, in our experience, relative risks greater than 2 were never observed in real data situation for middle-aged or elderly patients.

Table 1.

Theoretical relative survival (RS) estimates at 5, 10, and 15 years since diagnosis according to selected levels of net survival (NS) attained at the same times, by different relative risk (RR), sex, and age at diagnosis.

graphic file with name kwaf148fx1.jpg
Open in a new tab

Figures 2–6 show the differences between RS and NS estimated through the modeling approach applied to real-world data. The RRs estimated by the models are reported in Table S1. For H&N cancer patients they stood for males at 4.0, 2.6, and 1.6 at ages 40-59, 60-69, and 70-79, respectively. The corresponding estimates for female patients were 4.5, 2.9, and 1.8. Figures 2 and 3 clearly show the large resulting differences between the age-specific cumulative RS curves, steeply decreasing without any tendency to leveling off, and the corresponding NS curves reaching a plateau 8-10 years after diagnosis.

Figure 2.

Figure 2

Open in a new tab

Model-based relative survival (RS) (gray) and net survival (NS) (black) using new mixture cure models by age group and sex for male head and neck (H&N) cancer patients.

Figure 6.

Figure 6

Open in a new tab

Model-based RS (gray) and NS (black) using new mixture cure models by age group and sex for female breast cancer patients.

Figure 3.

Figure 3

Open in a new tab

Model-based RS (gray) and NS (black) using new mixture cure models by age group and sex for female H&N cancer patients.

Estimations of RR were very close to 1 for colorectal cancer (Table S1), with small differences between the considered age classes. The difference between RS and NS was very small for all ages, sexes, and times since diagnosis (Figures 4 and 5). The highest values (about 6% points of absolute difference between RS and NS) were observed for older patients after 15 years of follow-up.

Figure 4.

Figure 4

Open in a new tab

Model-based RS (gray) and NS (black) using new mixture cure models by age group and sex for male colorectum cancer patients.

Figure 5.

Figure 5

Open in a new tab

Model-based RS (gray) and NS (black) using new mixture cure models by age group and sex for female colorectum cancer patients.

The RR was estimated for female breast cancer as 1.3 (95% CI, 1.22-1.40) at ages 40-59, 1.4 (95% CI, 1.36-1.43) at ages 60-69, and 1.2 (95% CI, 1.22-1.27) at ages 70-79. The RS and NS curves substantially differed only for the oldest ages at diagnosis due the higher level of other causes mortality present at these ages in the general population (Figure 6). However, some flattening appeared for the NS curves indicating, differently from those of RS, that they tended to a constant level with no or negligible remaining risk of cancer death.

Discussion

When cancer patients have different mortality rates from other causes with respect to the general population, RS estimated from the population life tables tends, in the long term, to diverge from NS, defined as the patients’ hypothetical survival in absence of death from other diseases. More specifically, RS underestimates NS when the RR is greater than 1 (as for most of the cancer sites analyzed from real data so far), overestimates NS when RR is smaller than one, and equals NS when RR = 1. We have shown that the bias at 10 years from diagnosis was detectable when the RR was small, for example, equal to 1.1, only for elderly patients, became relevant for RR = 1.5, and was large at all ages when the RR was greater than 2. Since mortality from causes other than cancer increases sharply with age, RS also tends to exhibit a steeper decreasing age trend with respect to NS (see Table 1). For instance, if the NS was 60% at all ages and the RR was 1.5, at 5 years the RS decreased in men to 59% for younger patients, to 58% for patients aged 60 years at diagnosis, and to 56% for older patients. In fact, our results show that by using RS, part of the mortality from other causes can be wrongly attributed to cancer, and there is a possible overestimation of the impact of age on cancer survival, which results in a lower RS in the older age groups of patients. To note, results are not expected to change using PP estimator as the difference between RS and PP are negligible for 10-year age classes.17

The divergence between RS and NS, which affected both model-based and observed RS (Table S1), does not come as a surprise, as the 2 indicators point to different concepts. Relative survival discards excess mortality with respect to the general population whereas NS discards other cause mortality with respect to a more comparable general population (including common risk factors) and accounts for adverse effects of treatment. It focuses on the chance of surviving cancer itself, and has a clear clinical meaning. For instance, heart failure resulting from cancer treatment should not be counted as a cancer death in NS, even though it contributes to the excess risk of death for cancer patients compared to the general population. Net survival is the basic measure on which effective shaping of long-term clinical follow-up of cancer survivors depends. From the patient’s perspective it is also important to distinguish between risks due to the cancer and to other threats, and the concepts of cure fraction and time to cure should refer to the former.

Net survival hazard is also a component of a measure of practical interest as cumulative incidence of death due to cancer and to other causes accounting for the competing mortality. This is derived from crude cumulative survival, estimated by combining the overall survival with the correct net hazard: D + C if we intend “death caused by the cancer”, only D if “death from the cancer” (Figure 1).

The major divergence between NS and RS was observed for H&N tumors. This result is plausible, considering the strong impact of their risk factors, mainly smoke, alcohol, and HPV infection, on many other chronic diseases and cancers, often intermediated by low socioeconomic status. This is not the case of colorectal and breast cancers that are more related to a high socioeconomic status.18

Our intention with this paper was also to present our model-based method, which allows, by including the RR, the estimation of the extra risk of death by other causes. We would like to show a possible way to analyze such RR and provide an estimate closer to NS than those proposed so far. Being a complex measure, the RR is not easy to calculate, it can be estimated through modeling that has assumptions and limitations that are not without cost, all of which are debatable and questionable, but perhaps provides one more tool to describe cancer survival.

This new model-based approach is not intended to be used in place of RS that is an easily available epidemiological indicator for performing comparative studies on survival between populations and time periods. The real-world estimates of RR used in this paper came from a cure model analysis of EUROCARE-6 data.14 Model estimates of cure fraction were interpreted considering as cured those patients who are “cured from the original cancer regardless of any potential for, or presence of, remaining disabilities or side effects of treatment”.19 This definition is, when RR > 1, broader than the usual definition of patients that at some point after diagnosis reach the same survival and risk of death as the general population.20–22 The advantage that has determined the “success” of this last definition so far has been its conceptual simplicity and relative reproducibility.22 From our examples, H&N data failed to show evidence of the RS curve flattening out and no cure is then envisaged according to the usual definition. This contradicts the clinical evidence and is not of help to patients who, being in any case informed on their risks of side effects and of other diseases, could neither consider themselves safe from relapse and progression of their original cancer even when NS is not decreasing anymore. On the other hand, the considered model has proved16 to be robust also in true “no cure” scenarios with steadily decreasing cumulative NS. In these cases, the model correctly provided zero estimates of cure proportion parameter.

Previous applications of survival models not focused on cure13,23,24 have shown that an increased patients’ mortality risk due to other causes exists for several cancers. Other estimations of the excess mortality from other causes have been also based on the analysis of death certificates. They showed higher risks of other causes mortality for patients diagnosed with H&N, lung, stomach, and kidney cancers.25 On the contrary, a lower risk was found for prostate cancer patients, while conflicting results were reported for breast and colon cancers.25–28

Despite the robustness displayed in simulation analysis16 by the model considered for the empirical analysis, results from real-world data require caution. The parameters used in this work come from models that included age as covariate but did not consider interactions or time-dependent variables. In particular, the RR was estimated as a fixed parameter and not dependent on follow-up time. Instead, it could behave as a random effect within the patients’ population, with a consequent time-decreasing trend due to the selection of individuals at lower risk of death from other causes. The random effect case was explored in the previous simulation analysis finding, if ignored, a moderate degree (ie, less than 10% of the true RR value at the beginning) of underestimation.16 All model estimates are also sensitive to the considered parametric survival distribution. The results shown in this paper come after testing several survival distributions and choosing the one that best fitted the observed data.14 However, since all survival distributions have constraints, caution is recommended. Having information on relapse, treatment, risk factors, and comorbidities would be useful for disentangling the excess risk of dying from other causes, distinguishing between excess death due to treatment side effects versus shared risk factors, and validating model results through the application of a new multistate model.29 Unfortunately, this information is not routinely collected by all population-based cancer registries. In any case, the RR estimates considered in this paper should only be taken as plausible values aimed to underline the conceptual and practical difference between NS and RS.

Net survival should coincide with cause-specific survival estimated using cause of death. Unfortunately, cause of death is often not always collected in population-based settings, mainly because its reliability is not well known.10 Defining the cause of death for a patient with comorbidities is not easy, as it may be unclear whether the patient has died from the progression of cancer, comorbidities, or complications related to the cancer treatment. Model-based net hazard provides an alternative way to separate cancer from noncancer mortality risks.

A precise evaluation of the additional risks of diseases and mortality of cancer survivors is crucial, especially given the discrimination they often face in health insurance, employment, and loan access. While general health conditions related to risk factors or to side effects of treatments are already independently considered when assessing their future health, these factors should not influence the estimated probability of cancer relapse and progression. This issue arises when RS is used instead of NS as an indicator of cancer-specific mortality. In conclusion, we believe that overestimating cancer mortality risk may result in an undue additional burden on cancer survivors’ quality of life.

Supplementary Material

Web_Material_kwaf148
web_material_kwaf148.docx (25.4KB, docx)

Acknowledgments

We thank Joanne Mary Fleming for English language editing.

EUROCARE-6 WG: Austria: M. Hackl (National Cancer Registry [CR]); Belgium: E. Van Eycken; N. Van Damme (National CR); Bulgaria: Z. Valerianova (National CR); Croatia: M. Sekerija (National CR); Cyprus: V. Scoutellas; A. Demetriou (National CR); Czechia: L. Dušek; D. Krejici (National CR); Denmark: H. Storm (National CR); Estonia: M. Mägi; K. Innos* (National CR); Finland: J. Pitkäniemi (National CR); France: M. Velten (Bas Rhin CR); X. Troussard (Basse Normandie, Haematological Malignancies CR); A.M. Bouvier; V. Jooste* (Burgundy, Digestive CR); A.V. Guizard (Calvados, General CR); G. Launoy (Calvados, Digestive CR); S. Dabakuyo Yonli (Cote d'Or, Gynaecological (Breast) CR); M. Maynadié (Cote d'Or, Haematological Malignancies CR); A.S. Woronoff (Doubs CR); J.B. Nousbaum (Finistère, Digestive CR); G. Coureau (Gironde, General CR); A. Monnereau* (Gironde, Haematological Malignancies CR); I. Baldi (Gironde, Central Nervous System CR); K. Hammas (Haut-Rhin CR); B. Tretarre (Herault CR); M. Colonna (Isere CR); S. Plouvier (Lille Area CR); T. D'Almeida (Limousin CR); F. Molinié; A. Cowppli-Bony (Loire-Atlantique/Vendée CR); S. Bara (Manche CR); A. Debreuve (Marne-Ardennes, Thyroid CR); G. Defossez (Poitou-Charentes CR); B. Lapôtre-Ledoux (Somme CR); P. Grosclaude; L. Daubisse-Marliac (Tarn CR); Germany: S. Luttmann; A. Eberle (Bremen CR); R. Stabenow (Common CR of 4 Federal States (Brandenburg, Mecklenburg-West Pomerania, Saxony-Anhalt, Thüringen)); A. Nennecke (Hamburg CR); J. Kieschke (Lower Saxony CR); S. Zeissig (Rhineland-Palatinate CR); B. Holleczek (Saarland CR); A. Katalinic* (Schleswig-Holstein CR); Iceland: H. Birgisson (National CR); Ireland: D. Murray; P.M. Walsh (National CR); Italy: G. Mazzoleni; F. Vittadello (Alto Adige CR); F. Cuccaro (Barletta-Andria-Trani CR); R. Galasso (Basilicata CR); G. Sampietro (Bergamo CR); S. Rosso (Biella CR); C. Gasparotti; G. Maifredi (Brescia CR); M. Ferrante; R. Ragusa (Catania-Messina-Enna CR); A. Sutera Sardo (Catanzaro CR); M.L. Gambino; M. Lanzoni (Province of Varese and Como CR); P. Ballotari; E. Giacomazzi (Cremona and Mantova CR); S. Ferretti (Ferrara CR); A. Caldarella; G. Manneschi (Firenze-Prato CR); G. Gatta*; M. Sant*; P. Baili*; F. Berrino*; L. Botta; A. Trama; R. Lillini; A. Bernasconi; S. Bonfarnuzzo; C. Vener; F. Didonè; P. Lasalvia; L. Buratti; G. Tagliabue (Fondazione IRCCS Istituto Nazionale dei Tumori, Milan); D. Serraino; L. Dal Maso (Centro di Riferimento Oncologico, IRCCS, Aviano for the Friuli Venezia Giulia CR); R. Capocaccia* (Epidemiologia & Prevenzione Board); R. De Angelis*; E. Demuru; F. Cerza; F. Di Mari; C. Di Benedetto; S. Rossi*; M. Santaquilani; S. Venanzi; M. Tallon (Istituto Superiore di Sanità, Rome); L. Boni (Genova CR); S. Iacovacci (Latina CR); V. Gennaro (Liguria, mesotheliomas CR); A.G. Russo; F. Gervasi (Province of Milan and Lodi CR); G. Spagnoli (Modena CR); L. Cavalieri d'Oro (Monza and Brianza CR); M. Fusco; M.F. Vitale (Napoli 3 South CR); P.Pinna (Nuoro CR); W. Mazzucco (Palermo CR); M. Michiara (Parma CR); G. Chiranda (Piacenza CR); G. Cascone; M.C. Giurdanella (Ragusa CR); L. Mangone (Reggio Emilia CR); F. Falcini (Romagna CR); R. Cavallo (Salerno CR); D. Piras (Sassari CR); A. Madeddu; F. Bella (Siracusa CR); A.C. Fanetti (Sondrio CR); S. Minerba (Taranto CR); G. Candela; T. Scuderi (Trapani CR); R.V. Rizzello (Trento CR); F. Stracci (Umbria CR); M. Zorzi, S. Guzzinati(Veneto CR); A. Brustolin (Viterbo CR); Latvia: S. Pildava (National CR); Lithuania: I. Vincerzevskiene (National CR); Malta: M. Azzopardi (National CR); Norway: T.B. Johannesen* (National CR); Poland: J. Didkowska; U. Wojciechowska (National CR); M. Bielska-Lasota*; Portugal: A. Pais (Central Portugal CR); M.J. Bento; C. Alves-Rodrigues (Northern Portugal CR); A. Lourenço ; A. Mayer (Southern Portugal CR); Slovakia: C. Safaei Diba (National CR); Slovenia: V. Zadnik; T. Zagar (National CR); Spain: C. Sánchez-Contador Escudero ; P. Franch Sureda (Balearic Islands, Mallorca CR); A. Lopez de Munain; M. De-La-Cruz (Basque Country CR); M.D. Rojas; A. Aleman (Canary Islands CR); A. Vizcaino (Castellon CR); R. Marcos-Gragera; A. Sanvisens (Girona CR); M.J. Sanchez (Granada CR); M.D. Chirlaque Lopez; A. Sanchez-Gil (Murcia CR); M. Guevara*; E. Ardanaz (Navarra CR, CIBERESP); J. Galceran; M. Carulla (Tarragona CR); Switzerland: Y. Bergeron (Fribourg CR); E. Rapiti; R. Schaffar (Geneva CR); S. Mohsen Mousavi; P. Went (Graubünden and Glarus CR); S. Mohsen Mousavi; M. Blum (Eastern Switzerland CR); A. Bordoni (Ticino CR); Netherlands: O. Visser*; S. Siesling; (National CR); UK-England: S. Stevens; J. Broggio (National CR); UK-Northern Ireland: D. Bennett (National CR); A. Gavin*; UK-Scotland: D. Morrison (National CR); UK-Wales: D. W. Huws*(National CR);*EUROCARE Steering Committe

Contributor Information

Laura Botta, Evaluative Epidemiology Unit, Department of Epidemiology and Data Science, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy; INSERM CTM UMR 1231 EPICAD, University of Burgundy, Dijon, France.

Riccardo Capocaccia, Editorial Board, Epidemiologia e Prevenzione, Milan, Italy.

Alice Bernasconi, Evaluative Epidemiology Unit, Department of Epidemiology and Data Science, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy.

Silvia Rossi, Department of Oncology and Molecular Medicine, Istituto Superiore di Sanità, Rome, Italy.

Jaume Galceran, Tarragona Cancer Registry, Cancer Epidemiology and Prevention Service, Hospital Universitari Sant Joan de Reus, Reus, Spain; Pere Virgili Health Research Institute, Reus, Spain.

Luigino Dal Maso, Cancer Epidemiology Unit, Centro di Riferimento Oncologico di Aviano (CRO), IRCCS, Aviano, Italy.

Come Lepage, INSERM CTM UMR 1231 EPICAD, University of Burgundy, Dijon, France; Digestive oncology department, Dijon University Hospital, Dijon, France.

Florence Molinié, FRANCIM Network, Toulouse F-31073, France; Loire-Atlantique/Vendée Cancer Registry, Nantes, France; UMR 1295, Université Toulouse III, Inserm, Equipe EQUITY, Equipe constitutive du CERPOP, Toulouse, France.

Anne-Marie Bouvier, INSERM CTM UMR 1231 EPICAD, University of Burgundy, Dijon, France; FRANCIM Network, Toulouse F-31073, France; Digestive Cancer Registry of Burgundy, Dijon University Hospital, Dijon, France.

Rafael Marcos-Gragera, Epidemiology Unit and Girona Cancer Registry, Oncology Coordination Plan, Catalan Institute of Oncology (ICO), Girona Biomedical Research Institute (IdiBGi-CERCA), University of Girona, Josep Carreras Leukemia Research Institute (IJC), Girona, Spain; Centro de Investigación Biomédica en Red de Epidemiología y Salud Pública (CIBERESP), Madrid, Spain.

Claudia Vener, Epidemiology and Prevention Unit, Department of Epidemiology and Data Science, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy; Analytical Epidemiology and Health Impact Unit, Department of Epidemiology and Data Science, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy.

Marcela Guevara, Centro de Investigación Biomédica en Red de Epidemiología y Salud Pública (CIBERESP), Madrid, Spain; Instituto de Salud Pública y Laboral de Navarra, Pamplona, Spain; Navarra Institute for Health Research (IdiSNA), Pamplona, Spain.

Deirdre Murray, National Cancer Registry Ireland, Cork, Ireland; School of Public Health, University College Cork, Cork, Ireland.

Rosalia Ragusa, Catania-Messina-Enna CR, Azienda Ospedaliero Universitaria Policlinico, Catania, Italy.

Gemma Gatta, Evaluative Epidemiology Unit, Department of Epidemiology and Data Science, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy.

Valerie Jooste, INSERM CTM UMR 1231 EPICAD, University of Burgundy, Dijon, France; FRANCIM Network, Toulouse F-31073, France; Digestive Cancer Registry of Burgundy, Dijon University Hospital, Dijon, France.

the EUROCARE-6 WG:

M Hackl, E Van Eycken, N Van Damme, Z Valerianova, M Sekerija, V Scoutellas, A Demetriou, L Dušek, D Krejici, H Storm, M Mägi, K Innos, J Pitkäniemi, M Velten, X Troussard, A M Bouvier, V Jooste, A V Guizard, G Launoy, S Dabakuyo Yonli, M Maynadié, A S Woronoff, J B Nousbaum, G Coureau, A Monnereau, I Baldi, K Hammas, B Tretarre, M Colonna, S Plouvier, T D'Almeida, F Molinié, A Cowppli-Bony, S Bara, A Debreuve, G Defossez, B Lapôtre-Ledoux, P Grosclaude, L Daubisse-Marliac, S Luttmann, A Eberle, R Stabenow, A Nennecke, J Kieschke, S Zeissig, B Holleczek, A Katalinic, H Birgisson, D Murray, P M Walsh, G Mazzoleni, F Vittadello, F Cuccaro, R Galasso, G Sampietro, S Rosso, C Gasparotti, G Maifredi, M Ferrante, R Ragusa, A Sutera Sardo, M L Gambino, M Lanzoni, P Ballotari, E Giacomazzi, S Ferretti, A Caldarella, G Manneschi, G Gatta, M Sant, P Baili, F Berrino, L Botta, A Trama, R Lillini, A Bernasconi, S Bonfarnuzzo, C Vener, F Didonè, P Lasalvia, L Buratti, G Tagliabue, D Serraino, L Dal Maso, R Capocaccia, R De Angelis, E Demuru, F Cerza, F Di Mari, C Di Benedetto, S Rossi, M Santaquilani, S Venanzi, M Tallon, L Boni, S Iacovacci, V Gennaro, A G Russo, F Gervasi, G Spagnoli, L Cavalieri d'Oro, M Fusco, M F Vitale, P Pinna, W Mazzucco, M Michiara, G Chiranda, G Cascone, M C Giurdanella, L Mangone, F Falcini, R Cavallo, D Piras, A Madeddu, F Bella, A C Fanetti, S Minerba, G Candela, T Scuderi, R V Rizzello, F Stracci, M Zorzi, S Guzzinati, A Brustolin, S Pildava, I Vincerzevskiene, M Azzopardi, T B Johannesen, J Didkowska, U Wojciechowska, M Bielska-Lasota, A Pais, M J Bento, C Alves-Rodrigues, A Lourenço, A Mayer, C Safaei Diba, V Zadnik, T Zagar, C Sánchez-Contador Escudero, P Franch Sureda, A Lopez de Munain, M De-La-Cruz, M D Rojas, A Aleman, A Vizcaino, R Marcos-Gragera, A Sanvisens, M J Sanchez, M D Chirlaque Lopez, A Sanchez-Gil, M Guevara, E Ardanaz, J Galceran, M Carulla, Y Bergeron, E Rapiti, R Schaffar, S Mohsen Mousavi, P Went, S Mohsen Mousavi, M Blum, A Bordoni, O Visser, S Siesling, S Stevens, J Broggio, D Bennett, A Gavin, D Morrison, and D W Huws

Supplementary material

Supplementary material is available at American Journal of Epidemiology online.

Funding

Funding from the European Commission (Work Programme 2017, Grant Agreement 801520 HP-JA-2017, “Innovative Partnership for Action Against Cancer”) was used to prepare and clean the EUROCARE-6 database. The salary of L.B. was granted by the Italian Association of Cancer Research (AIRC) under Investigator Grant 2020 ID.24933 (Principal Investigator G.G.). The work of V.J. was supported by the French Institut National du Cancer (INCa grant number 2018-178). The work of LDM was supported by the Italian Association for Cancer Research (AIRC) (Grant no. 28893).

Conflict of interest

The authors have no relevant financial or nonfinancial interests to disclose.

Disclaimer

The sponsor has no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the article for publication.

Data availability

We are not permitted to share individual data. Aggregated-level data, in the form of counts, rates, or survival proportions, can be only shared after express permission from the participating registries. These data should be requested by contacting the corresponding author or Eurocare Secretariat (eurocare.secretariat@istitutotumori.mi.it).

References

  • 1. Perme  MP, Stare  J, Estève  J. On estimation in relative survival. Biometrics.  2012;68(1):113–120. 10.1111/j.1541-0420.2011.01640.x [DOI] [PubMed] [Google Scholar]
  • 2. Ederer  F, Axtell  LM, Cutler  SJ. The relative survival rate: a statistical methodology. Nat Cancer Inst Monogr.  1961;6:101–121 [PubMed] [Google Scholar]
  • 3. Hakulinen  T. On long-term relative survival rates. J Chron Dis.  1977;30(7):431–443. 10.1016/0021-9681(77)90036-4 [DOI] [PubMed] [Google Scholar]
  • 4. Pohar Perme  M, Estève  J, Rachet  B. Analysing population-based cancer survival—settling the controversies. BMC Cancer.  2016;16(1):933. 10.1186/s12885-016-2967-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Aziz  NM. Cancer survivorship research: state of knowledge, challenges and opportunities. Acta Oncol.  2007;46(4):417–432. 10.1080/02841860701367878 [DOI] [PubMed] [Google Scholar]
  • 6. Armenian  SH, Xu  L, Ky  B, et al.  Cardiovascular disease among survivors of adult-onset cancer: a community-based retrospective cohort study. J Clin Oncol.  2016;34(10):1122–1130. 10.1200/JCO.2015.64.0409 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Capocaccia  R, Gatta  G, Dal Maso  L. Life expectancy of colon, breast, and testicular cancer patients: an analysis of US-SEER population-based data. Ann Oncol.  2015;26(6):1263–1268. 10.1093/annonc/mdv131 [DOI] [PubMed] [Google Scholar]
  • 8. Clèries  R, Ameijide  A, Buxó  M, et al.  Ten-year probabilities of death due to cancer and cardiovascular disease among breast cancer patients diagnosed in North-Eastern Spain. Int J Environ Res Public Health.  2022;20(1):405. 10.3390/ijerph20010405 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Ellis  L, Coleman  MP, Rachet  B. The impact of life tables adjusted for smoking on the socio-economic difference in net survival for laryngeal and lung cancer. Br J Cancer.  2014;111(1):195–202. 10.1038/bjc.2014.217 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10. Bright  CJ, Brentnall  AR, Wooldrage  K, et al.  Errors in determination of net survival: cause-specific and relative survival settings. Br J Cancer.  2020;122(7):1094–1101. 10.1038/s41416-020-0739-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Hinchliffe  SR, Rutheford  MJ, Crowter  MJ, et al.  Should relative survival be used with lung cancer data. Br J Cancer.  2012;106(11):11854–11859. 10.1038/bjc.2012.182 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Withrow  DR, Pole  JD, Nishri  ED, et al.  Choice of relative or cause-specific approach to cancer survival analysis impacts estimates differentially by cancer type, population, and application: evidence from a Canadian population-based cohort study. Popul Health Metr.  2017;15(1):24. 10.1186/s12963-017-0142-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Rubio  FJ, Rachet  B, Giorgi  R, et al.  On models for the estimation of the excess mortality hazard in case of insufficiently stratified life tables. Biostatistics.  2019;22(1):51–67. 10.1093/biostatistics/kxz017 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. Botta  L, Capocaccia  R, Bernasconi  A, et al.  Estimating cure and risk of death from other causes of cancer patients: EUROCARE-6 data on head & neck, colorectal, and breast cancers. Eur J Cancer.  2024;508:114187. 10.1016/j.ejca.2024.114187 [DOI] [PubMed] [Google Scholar]
  • 15. Botta  L, Gatta  G, Trama  A, et al.  Excess risk of dying of other causes of cured cancer patients. Tumori J.  2019;105(3):199–204. 10.1177/0300891619837896 [DOI] [PubMed] [Google Scholar]
  • 16. Botta  L, Goungounga  J, Capocaccia  R, et al.  A new cure model that corrects for increased risk of non-cancer death: analysis of reliability and robustness, and application to real-life data. BMC Med Res Methodol.  2023;23(1):70. 10.1186/s12874-023-01876-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Roche  L, Danieli  C, Belot  A, et al.  Cancer net survival on registry data: use of the new unbiased Pohar-Perme estimator and magnitude of the bias with the classical methods. Int J Cancer.  2013;132(10):2359–2369. 10.1002/ijc.27830 [DOI] [PubMed] [Google Scholar]
  • 18. Launoy  G, Coleman  MP, Zadnik  V. Social Environment and Cancer in Europe. Cham: Springer; 2021. [Google Scholar]
  • 19. Haupt  R, Spinetta  JJ, Ban  I, et al.  International Berlin-Frankfurt-Münster Study Group Early and Late Toxicity Educational Committee (I-BFM-SG ELTEC). Long term survivors of childhood cancer: cure and care. The Erice statement. Eur J Cancer.  2007;43(12):1778–1780. 10.1016/j.ejca.2007.04.015 [DOI] [PubMed] [Google Scholar]
  • 20. Lambert  PC, Thompson  JR, Weston  CL, et al.  Estimating and modeling the cure fraction in population-based cancer survival analysis. Biostatistics.  2007;8(3):576–594. 10.1093/biostatistics/kxl030 [DOI] [PubMed] [Google Scholar]
  • 21. Romain  G, Boussari  O, Bossard  N, et al.  Time-to-cure and cure proportion in solid cancers in France. A population based study. Cancer Epidemiol.  2019;60:93–101. 10.1016/j.canep.2019.02.006 [DOI] [PubMed] [Google Scholar]
  • 22. Dal Maso  L, Panato  C, Tavilla  A, et al.  EUROCARE-5 Working Group. Cancer cure for 32 cancer types: results from the EUROCARE-5 study. Int J Epidemiol.  2020;49(5):1517–1525. 10.1093/ije/dyaa128 [DOI] [PubMed] [Google Scholar]
  • 23. Goungounga  JA, Touraine  C, Grafféo  N, et al.  CENSUR working survival group. Correcting for misclassification and selection effects in estimating net survival in clinical trials. BMC Med Res Methodol.  2019;19(1):104. 10.1186/s12874-019-0747-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Touraine  C, Graffeo  N, Giorgi  R. More accurate cancer-related excess mortality through correcting background mortality for extra variables. Stat Meth Med Res.  2020;29(1):122–136. 10.1177/0962280218823234 [DOI] [PubMed] [Google Scholar]
  • 25. Howlader  N, Ries  LAG, Mariotto  AB, et al.  Improved estimates of cancer-specific survival rates from population-based data. J Natl Cancer Inst.  2010;102(20):1584–1598. 10.1093/jnci/djq366 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Chen  J, Zheng  Y, Haihong Wang  H, et al.  Cause of death among patients with colorectal cancer: a population-based study in the United States. AGING.  2020;12(2):104022. 10.18632/aging.104022 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Almejide  A, Clèries  R, Carulla  M, et al.  Cause-specific mortality after a breast cancer diagnosis: a cohort study of 10,195 women in Girona and Tarragona. Clin Transl Oncol.  2019;21(8):1014–1025. 10.1007/s12094-018-02015-5 [DOI] [PubMed] [Google Scholar]
  • 28. Izci  H, Tambuyzer  T, Vandeven  J, et al.  Cause of death for patients with breast cancer: discordance between death certificates and medical files, and impact on survival estimates. Arch Public Health.  2021;79(1):111. 10.1186/s13690-021-00637-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Manevski  D, Putter  H, Pohar Perme  M, et al.  Integrating relative survival in multi-state models—a non-parametric approach. Stat Methods Med Res.  2022;31(6):997–1012. 10.1177/09622802221074156 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Web_Material_kwaf148
web_material_kwaf148.docx (25.4KB, docx)

Data Availability Statement

We are not permitted to share individual data. Aggregated-level data, in the form of counts, rates, or survival proportions, can be only shared after express permission from the participating registries. These data should be requested by contacting the corresponding author or Eurocare Secretariat (eurocare.secretariat@istitutotumori.mi.it).

Articles from American Journal of Epidemiology are provided here courtesy of Oxford University Press

RESOURCES