How can we make cancer survival statistics more useful for patients and clinicians: An illustration using localized prostate cancer in Sweden

Impact of Breast Cancer Screening on 10-Year Net Survival in Canadian Women Age 40-49 Years

PURPOSE

In Canada, some provincial/territorial mammography screening programs include women age 40-49 years, whereas others do not. This study examines the impact of this dichotomy on the 10-year breast cancer (BC) net survival (NS) among women age 40-49 years and 50-59 years at diagnosis.

METHODS

Using the Canadian Cancer Registry data record linked to death information, we evaluated the cohort of Canadian women age 40-49 years and 50-59 years diagnosed with BC from 2002 to 2007. We compared 10-year NS estimates in the jurisdictions with organized screening programs that included women age 40-49 years, designated as screeners (Northwest Territories, British Columbia, Alberta, Nova Scotia, and Prince Edward Island), with comparator programs that did not (Yukon, Manitoba, Saskatchewan, Ontario, Quebec, New Brunswick, and Newfoundland and Labrador).

RESULTS

BC was the primary cause of 10-year mortality in women age 40-49 years diagnosed with BC (90.7% of deaths). Among these women, the 10-year NS for screeners (84.8%; 95% CI, 83.8 to 85.8) was 1.9 percentage points (pp) higher than that for comparators (82.9%; 95% CI, 82.3 to 83.5; P = .001). The difference in favor of screeners was significant among women age 45-49 years (2.6 pp; P = .001) but not among women age 40-44 years (0.9 pp; P = .328). Similarly, the incidence-based BC mortality rate was significantly lower in screener jurisdictions among women age 40-49 years and 45-49 years, but not for 40-44 years. Provincial/territorial NS increased significantly with higher mammography screening participation (P = .003). The BC incidence rate was virtually identical in screener and comparator jurisdictions among women age 40-49 years (P = .976) but was significantly higher for comparators among women age 50-59 years (P < .001).

CONCLUSION

Screening programs that included women in their 40s were associated with a significantly higher BC 10-year NS in women age 40-49 years, but not an increased rate of BC diagnosis. These results may inform screening guidelines for women age 40-49 years.

Estimating causal effects in the presence of competing events using regression standardisation with the Stata command standsurv

BACKGROUND

When interested in a time-to-event outcome, competing events that prevent the occurrence of the event of interest may be present. In the presence of competing events, various estimands have been suggested for defining the causal effect of treatment on the event of interest. Depending on the estimand, the competing events are either accommodated or eliminated, resulting in causal effects with different interpretations. The former approach captures the total effect of treatment on the event of interest while the latter approach captures the direct effect of treatment on the event of interest that is not mediated by the competing event. Separable effects have also been defined for settings where the treatment can be partitioned into two components that affect the event of interest and the competing event through different causal pathways.

METHODS

We outline various causal effects that may be of interest in the presence of competing events, including total, direct and separable effects, and describe how to obtain estimates using regression standardisation with the Stata command standsurv. Regression standardisation is applied by obtaining the average of individual estimates across all individuals in a study population after fitting a survival model.

RESULTS

With standsurv several contrasts of interest can be calculated including differences, ratios and other user-defined functions. Confidence intervals can also be obtained using the delta method. Throughout we use an example analysing a publicly available dataset on prostate cancer to allow the reader to replicate the analysis and further explore the different effects of interest.

CONCLUSIONS

Several causal effects can be defined in the presence of competing events and, under assumptions, estimates of those can be obtained using regression standardisation with the Stata command standsurv. The choice of which causal effect to define should be given careful consideration based on the research question and the audience to which the findings will be communicated.

Reference adjusted loss in life expectancy for population-based cancer patient survival comparisons - with an application to colon cancer in Sweden

BACKGROUND

The loss in life expectancy, LLE, is defined as the difference in life expectancy between cancer patients and that of the general population. It is a useful measure for summarising the impact of a cancer diagnosis on an individual's life expectancy. However, it is less useful for making comparisons of cancer survival across groups or over time, since the LLE is influenced by both mortality due to cancer and other causes and the life expectancy in the general population.

METHODS

We present an approach for making LLE estimates comparable across groups and over time by using reference expected mortality rates with flexible parametric relative survival models. The approach is illustrated by estimating temporal trends in LLE of colon cancer patients in Sweden.

RESULTS

The life expectancy of Swedish colon cancer patients has improved, but the LLE has not decreased to the same extent since the life expectancy in the general population has also increased. When using a fixed population and other-cause mortality, i.e. a reference-adjusted approach, the LLE decreases over time. For example, using 2010 mortality rates as the reference, the LLE for females diagnosed at age 65 decreased from 11.3 if diagnosed in 1976 to 7.2 if diagnosed in 2010, and from 3.9 to 1.9 years for women 85 years old at diagnosis.

CONCLUSION

The reference-adjusted LLE is useful for making comparisons across calendar time, or groups, since differences in other cause mortality are removed.

IMPACT

The reference-adjusted approach enhances the use of LLE as a comparative measure.

Socio-economic inequalities in cancer survival: how do they translate into Number of Life-Years Lost?

BACKGROUND

We aimed to investigate the impact of socio-economic inequalities in cancer survival in England on the Number of Life-Years Lost (NLYL) due to cancer.

METHODS

We analysed 1.2 million patients diagnosed with one of the 23 most common cancers (92.3% of all incident cancers in England) between 2010 and 2014. Socio-economic deprivation of patients was based on the income domain of the English Index of Deprivation. We estimated the NLYL due to cancer within 3 years since diagnosis for each cancer and stratified by sex, age and deprivation, using a non-parametric approach. The relative survival framework enables us to disentangle death from cancer and death from other causes without the information on the cause of death.

RESULTS

The largest socio-economic inequalities were seen mostly in adults <45 years with poor-prognosis cancers. In this age group, the most deprived patients with lung, pancreatic and oesophageal cancer lost up to 6 additional months within 3 years since diagnosis than the least deprived. For most moderate/good prognosis cancers, the socio-economic inequalities widened with age.

CONCLUSIONS

More deprived patients and particularly the young with more lethal cancers, lose systematically more life-years than the less deprived. To reduce these inequalities, cancer policies should systematically encompass the inequities component.

Direct modeling of the crude probability of cancer death and the number of life years lost due to cancer without the need of cause of death: a pseudo-observation approach in the relative survival setting

In population-based cancer studies, net survival is a crucial measure for population comparison purposes. However, alternative measures, namely the crude probability of death (CPr) and the number of life years lost (LYL) due to death according to different causes, are useful as complementary measures for reflecting different dimensions in terms of prognosis, treatment choice, or development of a control strategy. When the cause of death (COD) information is available, both measures can be estimated in competing risks setting using either cause-specific or subdistribution hazard regression models or with the pseudo-observation approach through direct modeling. We extended the pseudo-observation approach in order to model the CPr and the LYL due to different causes when information on COD is unavailable or unreliable (i.e., in relative survival setting). In a simulation study, we assessed the performance of the proposed approach in estimating regression parameters and examined models with different link functions that can provide an easier interpretation of the parameters. We showed that the pseudo-observation approach performs well for both measures and we illustrated their use on cervical cancer data from the England population-based cancer registry. A tutorial showing how to implement the method in R software is also provided.

Cancer-specific survival by stage of bladder cancer and factors collected by Mallorca Cancer Registry associated to survival

Background

Information about survival by stage in bladder cancer is scarce, as well as about survival of non-invasive bladder cancer. The aims of this study are: 1) to find out the distribution of bladder cancer by stage; 2) to determine cancer-specific survival by stage of bladder cancer; 3) to identify factors that explain and predict the likelihood of survival and the risk of dying from these cancers.

Methods

Incident bladder cancer cases diagnosed between 2006 and 2011 were identified through the Mallorca Cancer Registry. Inclusion criteria: cases with code C67 according to the ICD-O 3rd edition with any behaviour and any histology, except lymphomas and small cell carcinomas. Cases identified exclusively through the death certificate were excluded. We collected the following data: sex; age; date and method of diagnosis; histology according to the ICD-O 3rd edition; T, N, M and stage at the time of diagnosis; and date of follow-up or death. End point of follow-up was 31 December 2015. Multiple imputation (MI) was performed to estimate cases with unknown stage. Cases with benign or indeterminate behaviour were excluded for the survival analysis. Actuarial and Kaplan-Meier methods and Cox regression models were used for survival analysis.

Results

One thousand nine hundred fourteen cases were identified. 14% were women and 65.4% were 65 years or older. 3.9% had no stage (benign or undetermined behaviour) and 11.5% had unknown stage. After MI, 37.5% were in stage Ta (non-invasive papillary carcinoma), 3.2% in stage Tis (carcinoma in situ), 34.3% in stage I, 11.7% in Stage II, 4.3% in stage III, and 9.0% in stage IV. Survival was 76% at 5 years. Survival by stage: 98% at stage Ta, 90% at stage Tis, 85% at stage I, 45% at stage II, 35% at stage III, and 7% at stage IV. The Cox model showed that age, histology, and stage, but not sex, were associated with survival.

Conclusion

Bladder cancer survival vary greatly with stage, among both non-invasive and invasive cases. The percentage of non-invasive cancers is high. Stage, age, and histology are associated to survival.

Net survival differences of breast cancer between stages at diagnosis and age groups in the east coast region of West Malaysia: a retrospective cohort study

OBJECTIVE

Estimation of the net survival of breast cancer helps in assessing breast cancer burden at a population level. Thus, this study aims to estimate the net survival of breast cancer at different cancer staging and age at diagnosis in the east coast region of West Malaysia.

SETTING

Kelantan, Malaysia.

PARTICIPANTS

All breast cancer cases diagnosed in 2007 and 2011 identified from Kelantan Cancer Registry.

DESIGN

This retrospective cohort study used a relative survival approach to estimate the net survival of patients with breast cancer. Thus, two data were needed; breast cancer data from Kelantan Cancer Registry and general population mortality data for Kelantan population.

PRIMARY AND SECONDARY OUTCOME MEASURES

Net survival according to stage and age group at diagnosis at 1, 3 and 5 years following diagnosis.

RESULTS

The highest net survival was observed among stage I and II breast cancer cases, while the lowest net survival was observed among stage IV breast cancer cases. In term of age at diagnosis, breast cancer cases aged 65 and older had the best net survival compared with the other age groups.

CONCLUSION

The age at diagnosis had a minimal impact on the net survival compared with the stage at diagnosis. The finding of this study is applicable to other populations with similar breast cancer profile.

Crude probability of death for cancer patients by spread of disease in New South Wales, Australia 1985 to 2014

BACKGROUND

To estimate trends in the crude probability of death for cancer patients by sex, age and spread of disease over the past 30 years in New South Wales, Australia.

METHODS

Population-based cohort of 716,501 people aged 15-89 years diagnosed with a first primary cancer during 1985-2014 were followed up to 31 December 2015. Flexible parametric relative survival models were used to estimate the age-specific crude probability of dying from cancer and other causes by calendar year, sex and spread of disease for all solid tumours combined and cancers of the colorectum, lung, female breast, prostate and melanoma.

RESULTS

Estimated 10-year sex, age and spread-specific crude probabilities of cancer death generally decreased over time for most cancer types, although the magnitude of the decrease varied. For example, out of 100 fifty-year old men with localized prostate cancer, 12 would have died from their cancer if diagnosed in 1985 and 3 in 2014. Greater degree of spread was consistently associated with higher probability of dying from cancer, although outcomes for lung cancer were consistently poor. For both males and females, the probability of non-cancer deaths was higher among older patients, those diagnosed with localized cancers and where cancer survival was higher.

CONCLUSION

Crude probabilities presented here may be useful in helping clinicians and their patients better understand prognoses and make informed decisions about treatment. They also provide novel insights into the relative contributions that early detection and improved treatments have on the observed temporal patterns in cancer survival.

Excess deaths in treatment-resistant depression

BACKGROUND

Patients with treatment-resistant depression (TRD) have an increased mortality risk compared with other patients with depression, but it is not known how this translates into absolute numbers of excess deaths.

METHODS

Swedish national registers were used to identify a cohort of 118,774 antidepressant initiators 18-69 years old with a depression diagnosis. Patients who initiated a third consecutive treatment trial were classified as having TRD. Flexible parametric survival models were used to estimate the mortality risk due to all causes and external causes (suicides and accidents), comparing TRD patients with patients with other depression while adjusting for clinical and sociodemographic covariates and including interactions with TRD, age, and Charlson comorbidity index (CCI) for a number of somatic comorbidities. Standardized survival was estimated, as were numbers of excess deaths among TRD patients within each age and comorbidity category.

RESULTS

Compared with the mortality risk of other depressed patients, patients with TRD experienced excess deaths in most age and comorbidity categories in the range of 7-16 deaths per 1000 patients during 5 years. Highest numbers for all-cause excess deaths were found among patients 18-29 years old with CCI 1, where 16 [95% confidence interval 5-28] of the expected 37 [25-48] deaths per 1000 patients were excess deaths. The majority of the excess deaths were due to external causes.

CONCLUSION

Patients with TRD experience significant numbers of excess deaths compared with other patients with depression.

Cancer survival statistics for patients and healthcare professionals - a tutorial of real-world data analysis

Monitoring survival of cancer patients using data collected by population-based cancer registries is an important component of cancer control. In this setting, patient survival is often summarized using net survival, that is survival from cancer if there were no other possible causes of death. Although net survival is the gold standard for comparing survival between groups or over time, it is less relevant for understanding the anticipated real-world prognosis of patients. In this review, we explain statistical concepts targeted towards patients, clinicians and healthcare professionals that summarize cancer patient survival under the assumption that other causes of death exist. Specifically, we explain the appropriate use, interpretation and assumptions behind statistical methods for competing risks, loss in life expectancy due to cancer and conditional survival. These concepts are relevant when producing statistics for risk communication between physicians and patients, planning for use of healthcare resources, or other applications when consideration of both cancer outcomes and the competing risks of death is required. To reinforce the concepts, we use Swedish population-based data of patients diagnosed with cancer of the breast, prostate, colon and chronic myeloid leukaemia. We conclude that when choosing between summary measures of survival it is critical to characterize the purpose of the study and to determine the nature of the hypothesis under investigation. The choice of terminology and style of reporting should be carefully adapted to the target audience and may range from summaries for specialist readers of scientific publications to interactive online tools aimed towards lay persons.

Reference-adjusted and standardized all-cause and crude probabilities as an alternative to net survival in population-based cancer studies

BACKGROUND

In population-based cancer survival studies, the most common measure to compare population groups is age-standardized marginal relative survival, which under assumptions can be interpreted as marginal net survival; the probability of surviving if it was not possible to die of causes other than the cancer under study (if the age distribution was that of a common reference population). The hypothetical nature of this definition has led to confusion and incorrect interpretation. For any measure to be fair in terms of comparing cancer survival, then differences between population groups should depend only on differences in excess mortality rates due to the cancer and not differences in other-cause mortality rates or differences in the age distribution.

METHODS

We propose using crude probabilities of death and all-cause survival which incorporate reference expected mortality rates. This makes it possible to obtain marginal crude probabilities and all-cause probability of death that only differ between population groups due to excess mortality rate differences. Choices have to be made regarding what reference mortality rates to use and what age distribution to standardize to.

RESULTS

We illustrate the method and some potential choices using data from England for men diagnosed with melanoma. Various marginal measures are presented and compared.

CONCLUSIONS

The new measures help enhance understanding of cancer survival and are a complement to the more commonly used measures.

Time-range based sequential mining for survival prediction in prostate cancer

BACKGROUND AND OBJECTIVE

Metastatic prostate cancer has a higher mortality rate than localized cancers. There is a need to investigate the survival outcome of metastatic prostate cancers separately. Also, the treatments undertaken by the patients affect their overall survival. The present study tries to analyze the sequence of treatments given to the patients, along with the time intervals between each set of treatments. The time when medication needs to be changed may provide some useful insights into the survival outcome of the patients.

MATERIALS AND METHODS

A total of 407 metastatic prostate cancer patients' data was collected and analyzed from an Indian tertiary care center. Popular sequence mining algorithms with exact order constraint have been applied to the treatment data. Appropriate time intervals were added in the resulted frequent sequences and fed to machine learning techniques along with other clinical data.

RESULTS

The study suggests that the proposed methodology of the time range based sequence mining approach gave better results than the existing methods with 84.5% accuracy and 0.89 AUC. The time intervals in the existing sequence mining algorithms can give the clinicians some useful insights into the survival analysis and in determining the best lines of treatments for a particular patient.

Estimation of the adjusted cause-specific cumulative probability using flexible regression models for the cause-specific hazards

In competing risks setting, we account for death according to a specific cause and the quantities of interest are usually the cause-specific hazards (CSHs) and the cause-specific cumulative probabilities. A cause-specific cumulative probability can be obtained with a combination of the CSHs or via the subdistribution hazard. Here, we modeled the CSH with flexible hazard-based regression models using B-splines for the baseline hazard and time-dependent (TD) effects. We derived the variance of the cause-specific cumulative probabilities at the population level using the multivariate delta method and showed how we could easily quantify the impact of a covariate on the cumulative probability scale using covariate-adjusted cause-specific cumulative probabilities and their difference. We conducted a simulation study to evaluate the performance of this approach in its ability to estimate the cumulative probabilities using different functions for the cause-specific log baseline hazard and with or without a TD effect. In the scenario with TD effect, we tested both well-specified and misspecified models. We showed that the flexible regression models perform nearly as well as the nonparametric method, if we allow enough flexibility for the baseline hazards. Moreover, neglecting the TD effect hardly affects the cumulative probabilities estimates of the whole population but impacts them in the various subgroups. We illustrated our approach using data from people diagnosed with monoclonal gammopathy of undetermined significance and provided the R-code to derive those quantities, as an extension of the R-package mexhaz.

Temporal trends in net and crude probability of death from cancer and other causes in the Australian population, 1984-2013

BACKGROUND

While net probabilities of death in the relative survival framework ignore competing causes of death, crude probabilities allow estimation of the real risk of cancer deaths. This study quantifies temporal trends in net and crude probabilities of death.

METHODS

Australian population-based cohort of 2,015,903 people aged 15-89 years, diagnosed with a single primary invasive cancer from 1984 to 2013 with mortality follow-up to 31 December 2014. Survival was analyzed with the cohort method. Flexible parametric relative survival models were used to estimate both probability measures by diagnosis year for all cancers and selected leading sites.

RESULTS

For each site, excess mortality rates reduced over time, especially for prostate cancer. While both the 10-year net and crude probability of cancer deaths decreased over time, specific patterns varied. For example, the crude probability of lung cancer deaths for males aged 50 years decreased from 0.90 (1984) to 0.79 (2013); whereas the corresponding probabilities for kidney cancer were 0.64 and 0.18 respectively. Patterns for crude probabilities of competing deaths were relatively constant. Although for younger patients, both net and crude measures were similar, crude probability of competing deaths increased with age, hence for older ages net and crude measures were different except for lung and pancreas cancers.

CONCLUSIONS

The observed reductions in probabilities of death over three decades for Australian cancer patients are encouraging. However, this study also highlights the ongoing mortality burden following a cancer diagnosis, and the need for continuing efforts to improve cancer prevention, diagnosis and treatment.

Summarizing and communicating on survival data according to the audience: a tutorial on different measures illustrated with population-based cancer registry data

Survival data analysis results are usually communicated through the overall survival probability. Alternative measures provide additional insights and may help in communicating the results to a wider audience. We describe these alternative measures in two data settings, the overall survival setting and the relative survival setting, the latter corresponding to the particular competing risk setting in which the cause of death is unavailable or unreliable. In the overall survival setting, we describe the overall survival probability, the conditional survival probability and the restricted mean survival time (restricted to a prespecified time window). In the relative survival setting, we describe the net survival probability, the conditional net survival probability, the restricted mean net survival time, the crude probability of death due to each cause and the number of life years lost due to each cause over a prespecified time window. These measures describe survival data either on a probability scale or on a timescale. The clinical or population health purpose of each measure is detailed, and their advantages and drawbacks are discussed. We then illustrate their use analyzing England population-based registry data of men 15-80 years old diagnosed with colon cancer in 2001-2003, aiming to describe the deprivation disparities in survival. We believe that both the provision of a detailed example of the interpretation of each measure and the software implementation will help in generalizing their use.

InterPreT cancer survival: A dynamic web interactive prediction cancer survival tool for health-care professionals and cancer epidemiologists

BACKGROUND

There are a variety of ways for quantifying cancer survival with each measure having advantages and disadvantages. Distinguishing these measures and how they should be interpreted has led to confusion among scientists, the media, health care professionals and patients. This motivates the development of tools to facilitate communication and interpretation of these statistics.

METHODS

"InterPreT Cancer Survival" is a newly developed, publicly available, online interactive cancer survival tool targeted towards health-care professionals and epidemiologists (http://interpret.le.ac.uk). It focuses on the correct interpretation of commonly reported cancer survival measures facilitated through the use of dynamic interactive graphics. Statistics presented are based on parameter estimates obtained from flexible parametric relative survival models using large population-based English registry data containing information on survival across 6 cancer sites; Breast, Colon, Rectum, Stomach, Melanoma and Lung.

RESULTS

Through interactivity, the tool improves understanding of various measures and how survival or mortality may vary by age and sex. Routine measures of cancer survival are reported, however, individualised estimates using crude probabilities are advocated, which is more appropriate for patients or health care professionals. The results are presented in various interactive formats facilitating understanding of individual risk and differences between various measures.

CONCLUSIONS

"InterPreT Cancer Survival" is presented as an educational tool which engages the user through interactive features to improve the understanding of commonly reported cancer survival statistics. The tool has received positive feedback from a Cancer Research UK patient sounding board and there are further plans to incorporate more disease characteristics, e.g. stage.

Gonadotropin-releasing Hormone Agonists, Orchiectomy, and Risk of Cardiovascular Disease: Semi-ecologic, Nationwide, Population-based Study

BACKGROUND

In observational studies, men with prostate cancer treated with gonadotropin-releasing hormone (GnRH) agonists had a higher risk of cardiovascular disease (CVD) compared to men who had undergone orchiectomy. However, selection bias may have influenced the difference in risk.

OBJECTIVE

To investigate the association of type of androgen deprivation therapy (ADT) with risk of CVD while minimising selection bias.

DESIGN, SETTING, AND PARTICIPANTS

Semi-ecologic study of 6556 men who received GnRH agonists and 3330 men who underwent orchiectomy as primary treatment during 1992-1999 in the Prostate Cancer Database Sweden 3.0.

OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS

We measured the proportion of men who received GnRH agonists as primary treatment in 580 experimental units defined by healthcare provider, diagnostic time period, and age at diagnosis. Incident or fatal CVD events in units with high and units with low use of GnRH agonists were compared. Net and crude probabilities were also analysed.

RESULTS AND LIMITATIONS

The risk of CVD was similar between units with the highest and units with the lowest proportion of GnRH agonist use (relative risk 1.01, 95% confidence interval [CI] 0.93-1.11). Accordingly, there was no difference in the net probability of CVD after GnRH agonist compared to orchiectomy (hazard ratio 1.02, 95% CI 0.96-1.09). The 10-yr crude probability of CVD was 0.56 (95% CI 0.55-0.57) for men on GnRH agonists and 0.52 (95% CI 0.50-0.54) for men treated with orchiectomy. The main limitation was the nonrandom allocation to treatment, with younger men with lower comorbidity and less advanced cancer more likely to receive GnRH agonists.

CONCLUSION

Our data do not support previous observations that GnRH agonists increase the risk of CVD in comparison to orchiectomy.

PATIENT SUMMARY

We found a similar risk of cardiovascular disease between medical and surgical treatment as androgen deprivation therapy for prostate cancer.

Potential overtreatment among men aged 80 years and older with localized prostate cancer in Japan

Despite treatment guidelines recommending observation for men with low‐risk prostate cancer with life expectancy <10 years, a majority of elderly patients choose active treatment, which may result in overtreatment. Given the growing burden of prostate cancer among men aged ≥80 years (super‐elderly men), accumulation of survival data for evaluation of overtreatment among super‐elderly patients is imperative. Here, we report results of a population‐based cohort study to clarify potential overtreatment of super‐elderly men with localized prostate cancer. We used cancer registry data from the Monitoring of Cancer Incidence in Japan project, which covers 47% of the Japanese population. The subjects were men diagnosed with prostate cancer between 2006 and 2008. Follow‐up period was 5 years. We calculated 5‐year relative survival rates among the active treatment and observation groups after imputation for missing values. Of the 48 782 patients with prostate cancer included in the analysis, 15.1% were super‐elderly men. The 5‐year relative survival rates of super‐elderly men with localized cancer were 105.9% and 104.1% among the active treatment and observation groups, respectively. This excellent relative survival rate in the observation group remained consistent even after stratification by tumor grade. Of the 2963 super‐elderly men with localized cancer, 252 (8.5%) with curative treatment and 1476 (49.8%) with hormone therapy were assumed to have been overtreated. The proportion of overtreatment was estimated to reach 80% after imputation. These specific survival data in super‐elderly men in the observation group can be useful in shared decision‐making for these patients and may lead to a reduction in overtreatment.

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

Background

Net survival is the most common measure of cancer prognosis and has been used to study differentials in cancer survival between ethnic or racial population subgroups. However, net survival ignores competing risks of deaths and so provides incomplete prognostic information for cancer patients, and when comparing survival between populations with different all-cause mortality. Another prognosis measure, “crude probability of death”, which takes competing risk of death into account, overcomes this limitation. Similar to net survival, it can be calculated using either life tables (using Cronin-Feuer method) or cause of death data (using Fine-Gray method). The aim of this study is two-fold: (1) to compare the multivariable results produced by different survival analysis methods; and (2) to compare the Cronin-Feuer with the Fine-Gray methods, in estimating the cancer and non-cancer death probability of both Indigenous and non-Indigenous cancer patients and the Indigenous cancer disparities.

Methods

Cancer survival was investigated for 9,595 people (18.5% Indigenous) diagnosed with cancer in the Northern Territory of Australia between 1991 and 2009. The Cox proportional hazard model along with Poisson and Fine-Gray regression were used in the multivariable analysis. The crude probabilities of cancer and non-cancer methods were estimated in two ways: first, using cause of death data with the Fine-Gray method, and second, using life tables with the Cronin-Feuer method.

Results

Multivariable regression using the relative survival, cause-specific survival, and competing risk analysis produced similar results. In the presence of competing risks, the Cronin-Feuer method produced similar results to Fine-Gray in the estimation of cancer death probability (higher Indigenous cancer death probabilities for all cancers) and non-cancer death probabilities (higher Indigenous non-cancer death probabilities for all cancers except lung cancer and head and neck cancers). Cronin-Feuer estimated much lower non-cancer death probabilities than Fine-Gray for non-Indigenous patients with head and neck cancers and lung cancers (both smoking-related cancers).

Conclusion

Despite the limitations of the Cronin-Feuer method, it is a reasonable alternative to the Fine-Gray method for assessing the Indigenous survival differential in the presence of competing risks when valid and reliable subgroup-specific life tables are available and cause of death data are unavailable or unreliable.

Electronic supplementary material

The online version of this article (doi:10.1186/s12963-016-0118-9) contains supplementary material, which is available to authorized users.

Crude mortality and loss of life expectancy of patients diagnosed with urothelial carcinoma of the urinary bladder in Norway

OBJECTIVE

Reports from cancer registries often lack clinically relevant information, which would be useful in estimating the prognosis of individual patients with urothelial carcinoma of the urinary bladder (UCB). This article presents estimates of crude probabilities of death due to UCB and the expected loss of lifetime stratified for patient characteristics.

MATERIALS AND METHODS

In Norway, 10,332 patients were diagnosed with UCB between 2001 and 2010. The crude probabilities of death due to UCB were estimated, stratified by gender, age and T stage, using flexible parametric survival models. Based on these models, the loss in expectation of lifetime due to UCB was also estimated for the different strata.

RESULTS

There is large variation in the estimated crude probabilities of death due to UCB (from 0.03 to 0.76 within 10 years since diagnosis) depending on age, gender and T stage. Furthermore, the expected loss of life expectancy is more than a decade for younger patients with muscle-invasive UCB and between a few months and 5 years for nonmuscle-invasive UCB.

CONCLUSIONS

The suggested framework leads to clinically relevant prognostic risk estimates for individual patients diagnosed with UCB and the consequence in terms of loss of lifetime expectation. The published probability tables can be used in clinical praxis for risk communication.

Analysing population-based cancer survival - settling the controversies

BACKGROUND

The relative survival field has seen a lot of development in the last decade, resulting in many different and even opposing suggestions on how to approach the analysis.

METHODS

We carefully define and explain the differences between the various measures of survival (overall survival, crude mortality, net survival and relative survival ratio) and study their differences using colon and prostate cancer data extracted from the national population-based cancer registry of Slovenia as well as simulated data.

RESULTS

The colon and prostate cancer data demonstrate clearly that when analysing population-based data, it is useful to split the overall mortality in crude probabilities of dying from cancer and from other causes. Complemented by net survival, it provides a complete picture of cancer survival in a given population. But when comparisons of different populations as defined for example by place or time are of interest, our simulated data demonstrate that net survival is the only measure to be used.

CONCLUSIONS

The choice of the method should be done in two steps: first, one should determine the measure of interest and second, one should choose among the methods that estimate that measure consistently.

Model Comparison for Breast Cancer Prognosis Based on Clinical Data

We compared the performance of several prediction techniques for breast cancer prognosis, based on AU-ROC performance (Area Under ROC) for different prognosis periods. The analyzed dataset contained 1,981 patients and from an initial 25 variables, the 11 most common clinical predictors were retained. We compared eight models from a wide spectrum of predictive models, namely; Generalized Linear Model (GLM), GLM-Net, Partial Least Square (PLS), Support Vector Machines (SVM), Random Forests (RF), Neural Networks, k-Nearest Neighbors (k-NN) and Boosted Trees. In order to compare these models, paired t-test was applied on the model performance differences obtained from data resampling. Random Forests, Boosted Trees, Partial Least Square and GLMNet have superior overall performance, however they are only slightly higher than the other models. The comparative analysis also allowed us to define a relative variable importance as the average of variable importance from the different models. Two sets of variables are identified from this analysis. The first includes number of positive lymph nodes, tumor size, cancer grade and estrogen receptor, all has an important influence on model predictability. The second set incudes variables related to histological parameters and treatment types. The short term vs long term contribution of the clinical variables are also analyzed from the comparative models. From the various cancer treatment plans, the combination of Chemo/Radio therapy leads to the largest impact on cancer prognosis.

The impact of comorbid disease history on all-cause and cancer-specific mortality in myeloid leukemia and myeloma - a Swedish population-based study

Background

Comorbidity increases overall mortality in patients diagnosed with hematological malignancies. The impact of comorbidity on cancer-specific mortality, taking competing risks into account, has not been evaluated.

Methods

Using the Swedish Cancer Register, we identified patients aged >18 years with a first diagnosis of acute myeloid leukemia (AML, N = 2,550), chronic myeloid leukemia (CML, N = 1,000) or myeloma (N = 4,584) 2002–2009. Comorbid disease history was assessed through in- and out-patient care as defined in the Charlson comorbidity index. Mortality rate ratios (MRR) were estimated through 2012 using Poisson regression. Probabilities of cancer-specific death were computed using flexible parametric survival models.

Results

Comorbidity was associated with increased all-cause as well as cancer-specific mortality (cancer-specific MRR: AML = 1.27, 95 % CI: 1.15–1.40; CML = 1.28, 0.96–1.70; myeloma = 1.17, 1.08–1.28) compared with patients without comorbidity. Disorders associated with higher cancer-specific mortality were renal disease (in patients with AML, CML and myeloma), cerebrovascular conditions, dementia, psychiatric disease (AML, myeloma), liver and rheumatic disease (AML), cardiovascular and pulmonary disease (myeloma). The difference in the probability of cancer-specific death, comparing patients with and without comorbidity, was largest among AML patients <70 years, whereas in myeloma the difference did not vary by age among the elderly. The probability of cancer-specific death was generally higher than other-cause death even in older age groups, irrespective of comorbidity.

Conclusion

Comorbidities associated with organ failure or cognitive function are associated with poorer prognosis in several hematological malignancies, likely due to lower treatment tolerability. The results highlight the need for a better balance between treatment toxicity and efficacy in comorbid and elderly AML, CML and myeloma patients.

Electronic supplementary material

The online version of this article (doi:10.1186/s12885-015-1857-x) contains supplementary material, which is available to authorized users.

Use of relative survival to evaluate non-ST-elevation myocardial infarction quality of care and clinical outcomes

Survival after non-ST-elevation myocardial infarction (NSTEMI) is high and non-cardiovascular death has become more frequent. Observational studies typically quantify quality of care and clinical outcomes using all-cause mortality, which nowadays may not reflect the impact of index NSTEMI. We review and investigate relative survival for quantifying longer term outcomes after NSTEMI. National cohort study of hospitalized NSTEMI (Myocardial Ischaemia National Audit Project; patients: n = 346 546, hospitals: n = 243, countries: England and Wales). Mortality rates derived from two relative survival techniques were compared with all-cause mortality, and the impact of relative survival adjusted patient characteristics compared with those from Cox proportional estimates. Cox proportional hazards models provide lower survival estimates because they include deaths from all causes, overestimate the impact of increasing age on survival, and underestimate temporal improvements in care. The Royston-Parmar model allows more accurate estimation of relative survival because it is flexible to the high early hazard of death after hospitalized NSTEMI. All-cause mortality gives an overall assessment of survival for a cohort of patients. Relative survival provides a more accurate and informed estimation of the impact of an index cardiovascular event and, if necessary, patient characteristics on survival.

Prostate cancer, prostate cancer death, and death from other causes, among men with metabolic aberrations

Supplemental Digital Content is available in the text.

Background:

Few previous studies of metabolic aberrations and prostate cancer risk have taken into account the fact that men with metabolic aberrations have an increased risk of death from causes other than prostate cancer. The aim of this study was to calculate, in a real-life scenario, the risk of prostate cancer diagnosis, prostate cancer death, and death from other causes.

Methods:

In the Metabolic Syndrome and Cancer Project, prospective data on body mass index, blood pressure, glucose, cholesterol, and triglycerides were collected from 285,040 men. Risks of prostate cancer diagnosis, prostate cancer death, and death from other causes were calculated by use of competing risk analysis for men with normal (bottom 84%) and high (top 16%) levels of each factor, and a composite score.

Results:

During a mean follow-up period of 12 years, 5,893 men were diagnosed with prostate cancer, 1,013 died of prostate cancer, and 26,328 died of other causes. After 1996, when prostate-specific antigen testing was introduced, men up to age 80 years with normal metabolic levels had 13% risk of prostate cancer, 2% risk of prostate cancer death, and 30% risk of death from other causes, whereas men with metabolic aberrations had corresponding risks of 11%, 2%, and 44%.

Conclusions:

In contrast to recent studies using conventional survival analysis, in a real-world scenario taking risk of competing events into account, men with metabolic aberrations had lower risk of prostate cancer diagnosis, similar risk of prostate cancer death, and substantially higher risk of death from other causes compared with men who had normal metabolic levels.

Cancer survival: global surveillance will stimulate health policy and improve equity

Millions of people will continue to be diagnosed with cancer every year for the foreseeable future. These patients all need access to optimum health care. Population-based cancer survival is a key measure of the overall effectiveness of health systems in management of cancer. Survival varies very widely around the world. Global surveillance of cancer survival is needed, because unless these avoidable inequalities are measured, and reported on regularly, nothing will be done explicitly to reduce them.

A comparison of relative and cause-specific survival by cancer site, age and time since diagnosis

Relative survival (RS) estimates are widely used by cancer registries, mainly because they do not rely on the well-documented deficiencies of cause of death information. The aim of our study was to compare 5-year cause-specific survival (CSS) estimates and 5-year RS estimates for different cancer sites by age and time since diagnosis, and discuss possible reasons for observed differences. Using data from the Cancer Registry of Norway, we identified 200,008 patients diagnosed with cancer at one of the 48 sites included in this analysis during the period 1996-2005, and followed them up until the end of 2010. CSS estimates were calculated (i) considering cause of death to be the cancer that was originally diagnosed and (ii) considering the cause of death to be a cancer within the same organ system. For most cancer sites the difference between CSS and RS estimates was small (<5%). The greatest differences were seen for rarer cancers such as mediastinum and Kaposi sarcoma. Including deaths from the same organ system in the calculation of CSS further reduced the differences for many sites. For younger age groups and shorter time since diagnosis, RS and CSS estimates tended to be similar, whereas CSS estimates tended to be lower than RS estimates with longer time since diagnosis in the oldest age groups. When compared to RS estimates CSS estimates were reliable for most of the cancer sites included in our analysis. There are, however, some exceptions where CSS estimates may not be recommended, including for rarer cancers and for patients aged 85 and above.

Disparate survival and risk of secondary non-Hodgkin lymphoma in histologic subtypes of Hodgkin lymphoma: a population-based study

We compared survival outcomes and rates of secondary non-Hodgkin lymphoma (NHL) in 28 323 patients with nodular lymphocyte predominant (NLPHL) and classical Hodgkin lymphoma (HL) from the Surveillance, Epidemiology and End Results database, diagnosed between 1995 and 2010. In a multivariate analysis NLPHL demonstrated a significantly better relative survival (5-year risk of lymphoma-related death 5.7%, hazard ratio [HR] 0.46, p < 0.0001) than the reference nodular sclerosis (NSHL) subtype (5-year risk 12.7%). Lymphocyte-rich classical HL had outcomes comparable to NSHL (5-year risk 14.3%, HR 0.84, p = 0.11). Exceptionally poor outcomes were observed in lymphocyte depleted HL (5-year risk 48.8%, HR 2.26, p < 0.0001). The risk of secondary NHL was increased in NLPHL (HR 2.81, p < 0.001) and lymphocyte-rich classical HL (HR 2.27, p = 0.002), but not in other subtypes compared with NSHL. In conclusion, the histologic classification retains a significant prognostic value in HL and the disparities between the subtypes warrant customized treatment and surveillance strategies.

Comparative outcomes of rituximab-based systemic therapy and splenectomy in splenic marginal zone lymphoma

Despite diagnostic and therapeutic advances, the majority of patients with splenic marginal zone lymphoma (SMZL) are still treated with splenectomy. We analyzed survival outcomes after surgery or rituximab-based systemic therapy in the Surveillance Epidemiology and End Results-Medicare database, using inverse probability of treatment weighting to minimize treatment selection bias. From the 657 recorded cases diagnosed between 2000 and 2007, with a median age of 77 years, we selected 227 eligible patients treated with splenectomy (68 %), rituximab alone (23 %), or in combination with chemotherapy (9 %) within 2 years from diagnosis. No significant difference between the groups was observed in the cumulative incidence of lymphoma-related death (LRD) at 3 years (19.6 % with systemic therapy and 17.3 % with splenectomy; hazard ratio [HR], 1.04; 95 % confidence interval [CI], 0.56-1.92; P = 0.90) or in the overall survival (HR, 1.01; 95 % CI, 0.66-1.55; P = 0.95). The 90-day mortality after splenectomy was 7.1 %. The rates of hospitalizations, infections, transfusions, and cardiovascular or thromboembolic events were higher after combination chemoimmunotherapy than after splenectomy. Conversely, there was no significant difference in most complications between groups treated with splenectomy or rituximab alone. The cumulative incidence of LRD after single-agent rituximab at 3 years was 18.7 % (95 % CI, 8.6-31.7). In conclusion, in SMZL patients over the age of 65 years, the risk of LRD and overall survival are similar with systemic therapy or splenectomy as initial therapy. Single-agent rituximab may offer the most favorable risk/benefit ratio in this population.

Population-based prognostic factors for survival in patients with Burkitt lymphoma: an analysis from the Surveillance, Epidemiology, and End Results database

BACKGROUND

Burkitt lymphoma (BL) is an aggressive but potentially curable lymphoma, previously described in small, single-institution studies. This study evaluated prognostic factors for survival in adult patients with BL and a potential outcome improvement over the past decade in a population-based cohort.

METHODS

Adult patients with BL diagnosed between 1998 and 2009 were selected from the Surveillance, Epidemiology, and End Results (SEER) database. Prognostic factors were identified in a multivariate model for relative survival (RS), and trends in survival were evaluated using period analysis.

RESULTS

The study cohort included 2284 patients, with a median age of 49 years and male predominance (2.6:1). Gastrointestinal tract and the head and neck were the most common sites of extranodal disease. Older age, black race/ethnicity, and advanced stage were associated with a worse survival. In the period analysis, trends in improved survival between 1998 and 2009 were seen, except for patients older than 60 years and black patients, whose survival did not improve. A prognostic score divided patients into 4 groups: low-risk (5-year RS: 71%), low-intermediate (5-year RS: 55%), high-intermediate (5-year RS: 41%), and high-risk (5-year RS: 29%; P  < .001).

CONCLUSIONS

The outcome of patients younger than 60 years with BL improved over the past decade. Age, race, and stage have a prognostic role for survival. The proposed score can help inform prognosis in newly diagnosed patients and stratify participants in future trials.