Median overall survival for patients starting first‐line chemotherapy for metastatic gastric cancer is less than 1 year. This study is a follow‐up to a study that determined that most cancer patients preferred to receive survival estimates for worst‐case, typical, and best‐case scenarios rather than a single‐point estimate of median survival. This article describes the applicability of these findings in a population with advanced gastric cancer.
Keywords: Prognosis in gastric cancer, Estimating survival times
Abstract
Background.
Worst‐case, typical, and best‐case scenarios for survival, based on simple multiples of an individual's expected survival time (EST), estimated by their oncologist, are a useful way of formulating and explaining prognosis. We aimed to determine the accuracy and prognostic significance of oncologists’ estimates of EST, and the accuracy of the resulting scenarios for survival time, in advanced gastric cancer.
Materials and Methods.
Sixty‐six oncologists estimated the EST at baseline for each of the 152 participants they enrolled in the INTEGRATE trial. We hypothesized that oncologists’ estimates of EST would be unbiased (∼50% would be longer or shorter than the observed survival time [OST]); imprecise (<33% within 0.67–1.33 times the OST); independently predictive of overall survival (OS); and accurate at deriving scenarios for survival time with approximately 10% of patients dying within a quarter of their EST (worst‐case scenario), 50% living within half to double their EST (typical scenario), and 10% living three or more times their EST (best‐case scenario).
Results.
Oncologists’ estimates of EST were unbiased (45% were shorter than the OST, 55% were longer); imprecise (29% were within 0.67–1.33 times observed); moderately discriminative (Harrell's C‐statistic 0.62, p = .001); and an independently significant predictor of OS (hazard ratio, 0.89; 95% confidence interval, 0.83–0.95; p = .001) in a Cox model including performance status, number of metastatic sites, neutrophil‐to‐lymphocyte ratio ≥3, treatment group, age, and health‐related quality of life (EORTC‐QLQC30 physical function score). Scenarios for survival time derived from oncologists’ estimates were remarkably accurate: 9% of patients died within a quarter of their EST, 57% lived within half to double their EST, and 12% lived three times their EST or longer.
Conclusion.
Oncologists’ estimates of EST were unbiased, imprecise, moderately discriminative, and independently significant predictors of OS. Simple multiples of the EST accurately estimated worst‐case, typical, and best‐case scenarios for survival time in advanced gastric cancer.
Implications for Practice.
Results of this study demonstrate that oncologists’ estimates of expected survival time for their patients with advanced gastric cancer were unbiased, imprecise, moderately discriminative, and independently significant predictors of overall survival. Simple multiples of the expected survival time accurately estimated worst‐case, typical, and best‐case scenarios for survival time in advanced gastric cancer.
摘要
背景。肿瘤学家根据个人预期生存时间 (EST) 的简单倍数估计的最坏生存情况设想、一般生存情况设想和最佳生存情况设想是制定和解释预后的有效方法。我们的目的在于确定肿瘤学家估计的晚期胃癌EST的准确性和预后意义，以及由此产生的生存时间设想的准确性。
材料和方法。六十六位肿瘤学家估计了 152 名参与 INTEGRATE 试验的参与者中每位参与者的基线EST。我们假设肿瘤学家估计的EST是中肯的 [~50% 患者的预期生存时间将比观察生存时间 (OST)长或短]；不精确(<33%在OST的 0.67–1.33 倍之间)；可独立预测总生存期 (OS)；并且准确得出生存时间的设想，其中，约 10% 的患者在其EST的四分之一内死亡(最坏情况设想)，50% 患者的生存时间在EST的一半至两倍之间(一般情况设想)，10% 患者的生存时间为EST的三倍或以上(最佳情况设想)。
结果。肿瘤学家估计的EST是中肯的(45% 的患者比OST短，55% 的患者比OST长)；不精确(29%在OST的 0.67–1.33 倍之间)；比较具有判别性(Harrell 的 C 统计量为 0.62，p = 0.001)；并且是 Cox 模型中OS的独立重要的预测因子(风险比， 0.89；95% 置信区间，0.83–0.95；p = 0.001)，包括体能状况、转移灶数量、中性粒细胞与淋巴细胞比率 ≥3、治疗小组、年龄和健康相关的生活质量(EORTC‐QLQC30 身体机能评分)。根据肿瘤学家的估计得出的生存时间设想非常准确:9% 的患者在EST的四分之一内死亡，57% 患者的生存时间在EST的一半至两倍之间，12% 患者的生存时间为EST的三倍或以上。
结论。肿瘤学家估计的ESY是中肯的、不精确、比较具有判别性，并且是OS独立重要的预测因子。简单的EST倍数准确估计了晚期胃癌生存时间的最坏情况、一般情况和最佳情况设想。
实践意义:本研究的结果表明，肿瘤学家估计的晚期胃癌患者的预期生存时间是中肯的、不精确、比较具有判别性，并且是总生存期独立重要的预测因子。简单的预期生存时间倍数准确估计了晚期胃癌生存时间的最坏情况、一般情况和最佳情况设想。
Introduction
The median overall survival (OS) for patients starting first‐line chemotherapy for metastatic gastric cancer is less than 1 year and has remained so over the last 2 decades [1], [2], [3]. Oncologists caring for patients with metastatic gastric cancer are frequently asked for estimates of expected survival time by patients and their family members. Conversations about life expectancy may occur any time after a diagnosis of cancer, but a common trigger is when the cancer progresses on systemic therapy.
In a previous study, we showed that most patients with cancer or survivors who wanted numerical information about their prognosis reported that they would prefer to receive estimates for worst‐case, typical, and best‐case scenarios rather than a single point estimate of the median survival [4]. As a consequence, we recommended that oncologists explain survival time by using ranges for these scenarios [4], [5]. We have also shown that simple multiples of an oncologist's estimate of expected survival time (EST) in individual patients with a range of advanced cancers provided accurate estimates of ranges for these three scenarios [5], [6]. For example, in two cohorts of patients with mixed advanced cancers in which oncologists estimated each patient's EST at baseline, we found that approximately 10% of patients died within one quarter of their EST (worst‐case scenario), approximately 50% survived for between half to double their EST (typical scenario), and 10% lived three or more times their EST (best‐case scenario) [6], [7]. This distribution is approximately what would be expected given an exponential distribution for survival time.
In the current study we sought to test the applicability of our previous findings in a more homogenous population with advanced gastric cancer. Specifically, we aimed to determine the accuracy and prognostic significance of oncologists’ estimates of EST and the resulting scenarios for survival time based on simple multiples of these estimates. As an exploratory objective we sought to determine if time to progression (TTP) on last‐line therapy could be used to estimate postprogression survival time (PPST).
Subjects, Materials, and Methods
We studied participants in a previously reported, randomized, placebo‐controlled, phase II trial of regorafenib in advanced gastric cancer (INTEGRATE) [8]. Key eligibility criteria included an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1, locally recurrent or metastatic gastric or esophago‐gastric cancer progressing after one or two lines of chemotherapy, measurable disease according to RECIST 1.1, and an estimated life expectancy of at least 12 weeks.
Between November 2012 and February 2014, 66 medical oncologists from four countries (Australia, New Zealand, Canada, and South Korea) recruited 152 patients to participate in the randomized, phase II INTEGRATE trial. The protocol was approved by human research ethics committees at all participating institutions, and all participants gave written, informed consent. At the baseline assessment for each participant recruited, we asked the treating oncologist to record their “estimate of this patient's expected survival time in weeks, in other words, your estimate of the median survival of a group of identical patients.”
Hypotheses and criteria for accuracy were specified a priori. As per our previous studies, we hypothesized that oncologists’ estimates of EST would be unbiased, with approximately equal proportions being longer and shorter than the observed survival time (OST). [6], [7] In keeping with previous studies, point estimates of EST were deemed precise if they were within 0.67–1.33 times the OST, and we expected less than a third of estimates would meet this criterion [6], [7], [9], [10]. To determine if the accuracy of oncologist estimates varied by region, we compared OST/EST in Korean oncologists with OST/EST for oncologists in Australia, New Zealand, and Canada, in keeping with the stratification by region in the original trial.
To determine the accuracy of the scenarios for survival we calculated the ratio of the OST divided by the expected survival time (OST/EST) and used the Kaplan‐Meier distribution of OST/EST to account for participants who were still alive at the end of the study and therefore had censored observations for OST. Based on our previous work we hypothesized that simple multiples of the oncologist's estimate of EST could be used to estimate scenarios for observed survival as follows: (a) approximately 10% of participants would die within one quarter of their oncologists’ estimate (worst‐case scenario, OST/EST ≤0.25); (b) approximately 50% of participants would survive from half to double their oncologists’ estimate (typical scenario, 0.5 ≤ OST/EST ≤2); and (c) Approximately 10% of participants would live for three or more times their oncologist's estimate (best‐case scenario, OST/EST ≥3).
We also hypothesized that oncologists’ estimates of EST would be moderately discriminative, with a lower 95% confidence interval for Harrell's C‐statistic >0.5. [11] The C‐statistic is the probability that for a randomly chosen pair of participants, the participant predicted to have a longer survival actually lived longer. A value of 1.0 indicates perfect predictions and a value of 0.5 indicates completely random predictions.
Cox proportional hazards regression was used to assess the association between EST and OST, accounting for other covariates that had clinical relevance and were found to be associated with OS in a previous analysis of the INTEGRATE trial (EORTC QLQC30 physical function score, neutrophil to lymphocyte ratio ≥3, treatment group ECOG performance status, primary site of disease, and number of metastatic sites) [8], [12]. The effect of adjustment for each covariate was evaluated individually, and a multivariate model was constructed using a backward elimination approach.
In our planned exploratory analyses, we sought to determine a relationship between time to progression in an individual (TTPi) and their postprogression survival time (PPSTi). For each patient we calculated the ratio of PPSTi/TTPi and compared the median of these individual ratios with the ratio of the median PPST and median TTP for the trial overall. We explored whether there was a simple method of using an individual's TTPi to estimate their PPSTi.
Results
Estimates of EST were available for all 152 participants enrolled in the INTEGRATE trial. The 66 oncologists were all site investigators for the trial and had expertise in upper gastrointestinal cancer; most practiced in university affiliated teaching hospitals (80%), 78% were male, 60% were based in Australia, 20% in Canada, 18% in Korea, and 3% in New Zealand. Each oncologist enrolled from 1 to 11 participants (median, 1; interquartile range [IQR], 1–3).
Participant characteristics are summarized in Table 1. There were 121 deaths (82%) during the median follow‐up time of 17 months (range, 15–19). The majority of deaths were attributed to disease progression (117/121, 97%). [8] The four deaths attributed to other causes included abdominal pain of uncertain cause in the regorafenib arm and liver failure in the regorafenib arm, death of unspecified cause in the placebo group, and anorexia in the placebo group.
Table 1. Patient characteristics.
Abbreviation: ECOG, Eastern Cooperative Oncology Group
The median OST was 5.0 months (IQR, 2.6–9.5). The median estimate of EST was 5.5 months, (IQR, 4.6–6.6). Kaplan‐Meier curves of observed and estimated survival times are shown in Figure 1. The most frequent estimate of EST was 6 months (34%; Fig. 2).
Figure 1.
Kaplan‐Meier curves of observed and expected survival.
Figure 2.
Frequency distribution of oncologists’ estimates of expected survival time.
Oncologists’ estimates were unbiased, with approximately equal proportions being shorter (45%) or longer (55%) than the OST, as illustrated in Figure 3. Oncologists from Korea tended to underestimate survival, with a median OST/EST of 1.25 (95% confidence interval [CI], 0.71–1.68), whereas oncologists from Australia, New Zealand, and Canada tended to overestimate survival, with a median OST/EST of 0.79 (95% CI, 0.69–0.96), p = .03.
Figure 3.
Observed versus expected survival times for each participant. Points on the 45‐degree line represent participants who lived exactly as long as predicted, points above the line represent participants who lived longer than predicted, and points below the line represent participants who lived shorter than predicted.
As hypothesized, approximately one third of oncologists’ estimates of EST were precise (29% were within 0.67–1.33 times OST). Oncologists’ estimates were moderately discriminative, with a Harrell's C‐statistic of 0.62 (95% CI, 0.57–0.68, p = .001).
Scenarios for survival time based on simple multiples of the EST were as hypothesized, with 9% of participants dying within a quarter of their EST (a priori hypothesis 10%), 57% living between half to double their EST (a priori hypothesis 50%), and 12% living three or more times their EST (a priori hypothesis 10%). The Kaplan‐Meier curve for the distribution of OST/EST is shown in Figure 4.
Figure 4.
Kaplan‐Meier distribution of observed‐to‐expected survival time ratios (OST/EST).
Abbreviations: EST, expected survival time; OST, observed survival time.
Oncologists’ estimates of EST were independently significant predictors of OS (HR, 0.89; 95% CI, 0.83–0.95, p = .001) in a multivariable model accounting for other prognostic factors. Table 2 summarizes the factors associated with OST.
Table 2. Prognostic factors associated with observed survival time.
Abbreviations: CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; EORTC, ; HR, hazard ratio.
A total of 100 of the 152 participating patients had a date of progression documented before death: 62 allocated regorafenib and 38 allocated placebo. For the 62 participants allocated regorafenib, the median time to progression (TTPm) for the group was 1.8 months, and the median postprogression survival time PPST (PPSTm) for the group was 3.6 months, giving a PPSTm/TTPm ratio of 2.0. The median of the 62 ratios of each individual's postprogression survival time to their time to progression (PPSTi/TTPi) was similar at 1.9 but widely dispersed (IQR, 1.5–6.0). An individual's TTPi was not strongly associated with their PPSTi (HR, 0.89, p = .08), but using simple multiples of “2.0 × TTPi” to calculate scenarios for an individual's PPSTi was reasonably accurate. That is, 10% of participants who progressed then died within a quarter of (2 × TTPi), 50% lived within half to double (2 × TTPi), and 23% lived three or more times (2 × TTPi).
Of the 50 patients randomized to placebo, 29 received regorafenib following progression. For the 38 patients on placebo with a documented date of progression, the TTPm for the group was 0.9 months, PPSTm for the group was 3.8 months, and the ratio PPSTm/TTPm was 4.1. The median of the 38 individual PPSTi/TTPi ratios was similar at 4.0 but widely dispersed (IQR, 1.5–6).
Discussion
Oncologists’ estimates of EST in patients with advanced gastric cancer were unbiased, imprecise, and independently significant predictors of OST after accounting for conventional prognostic factors. Although point estimates of survival time were imprecise, simple multiples of oncologists’ estimates of EST were accurate for estimating worst‐case (less than one quarter times the EST), typical (half to double the EST), and best‐case (three or more times the EST) scenarios for survival.
Our previous work has been in heterogeneous populations with mixed advanced cancers, in both clinical trial and real world settings, with median OS times of approximately 12 months [6], [7]. This study comprised a more homogeneous population of patients with advanced gastric cancer in a trial of last‐line systemic therapy and a median survival time of only 5 months. Despite these differences, the results were consistent with our previous studies [6], [7], supporting the broader applicability of our method for estimating scenarios for survival time in advanced cancer.
We found that point estimates of EST were precise (within 0.67–1.33 times the OST) in approximately one third of participants, similar to the finding in our previous study (29%) [7] and that in a single‐center study of 75 patients with mixed advanced cancers by Taniyama et al. (36%) [13]. The ranges for three scenarios for survival were more accurate than the single point estimates of expected survival time on which they were based.
In the current study, oncologists’ estimates overall were unbiased with no systematic tendency to underestimate or overestimate survival time. However, when comparing oncologists by region, those from Australia, New Zealand, and Canada tended to overestimate survival, whereas those from Korea tended to underestimate survival. A larger study would be required to determine if this difference is real and, if so, to explore the reasons for it. A tendency to overestimate survival has been reported in studies of patients in the terminal phase of their illness with median survival times of 1 month or less [9], [14], [15]. Oncologists’ estimates of EST were moderately discriminative, with a C‐statistic of 0.62. We also found oncologists’ estimates of EST had independent prognostic value after accounting for established prognostic factors, similar to the finding in our previous study of patients with mixed advanced cancers [7].
Although the median survival time of the participants in this study was 5 months, the individual survival times ranged from 3 weeks to 2 years, demonstrating the wide variability of survival times even in a highly selected and homogeneous population. Interestingly, 39 participants (26%) died within 12 weeks of study entry, despite a life expectancy of <12 weeks being an exclusion criterion, highlighting the vagaries of prognostication. Estimating and explaining prognosis as ranges for possible worst‐case, typical, and best‐case scenarios accurately conveys this inherent uncertainty and also conveys realistic hope.
Questions about prognosis are often triggered at the time of progression on anticancer therapy and require oncologists to estimate and explain postprogression survival. For the 62 patients allocated regorafenib who had a documented date of progression, we found the median postprogression survival time was double the median time to progression. The median of each individual's PPSTi/TTPi was also 2. Simple multiples of this ratio of two were useful for estimating worst‐case and typical scenarios for an individual's PPSTi from their TTPi but underestimated the best‐case scenario. For example, for patients in this study with a TTP of 1 month and an expected PPST of 2 months, approximately 10% died within 2 weeks, (1/4 × 2 months), 50% survived 1 to 4 months (half to double 2 months), but 20% (not 10%) survived 6 months or longer (three or more times 2 months). Further studies are needed to determine if simple multiples of the ratio of median PPST to median TTP from a trial can be used to estimate an individual's PPSTi from their TTPi.
The median ratio of PPSTi/TTPi in the 38 participants allocated placebo with a documented date of progression before death was 4. This larger ratio reflects a shorter TTP in these patients randomized to placebo and the fact that many received regorafenib following progression. These observations are consistent with regorafenib having a greater effect on progression free survival than on overall survival.
The strengths of this study include its prospective design and extended follow‐up. This study was conducted as part of a rigorous, multinational, clinical trial, with questions and hypotheses specified a priori. Follow‐up was sufficient for a date of death in 81% of participants.
The main limitation of our study is that it was confined to participants in a clinical trial who may not be representative of unselected patients in routine clinical practice where the survival times would likely be shorter. In our previous work of “real world” patients with mixed advanced cancers, we have demonstrated that an estimated worst‐case scenario of one sixth of the EST would be more appropriate, compared with one quarter in the clinical trial setting. [6] In this study of patients receiving last‐line therapy for advanced gastric cancer, the median OS was 5 months, so the difference between using multiples of one‐sixth or one‐quarter of the EST was typically less than 2 weeks and was not clinically important. Another limitation was that the clinicians were site investigators with expertise and interest in gastric cancer, and were predominantly from academic centers, so may not be representative of oncologists in general.
This study supports our three scenarios framework for oncologists trying to estimate and explain survival time to patients with advanced gastric cancer. When such a patient requests information about their survival time, we recommend oncologists start by estimating “the median survival in a group of identical patients” and then use simple multiples of this estimate to calculate the worst‐case, typical, and best‐case scenarios for survival time. For example, if an oncologist estimates the expected survival for an individual as 6 months, they can explain to that individual that they would expect about 10 of 100 similar patients to die within 6 weeks (one quarter × 6 months, the worst‐case scenario), the middle 50 of the 100 to live between 3 and 12 months (half to double 6 months, the typical scenario), and about 10 of the 100 to live beyond 18 months (≥3 × 6 months, the best‐case scenario). In previous work we have shown that patients find survival information formatted as worst‐case, typical, and best‐case scenarios more helpful, easier to understand, and less upsetting than a single point estimate of the median survival [4].
Conclusion
In this study of participants in a trial of last‐line anticancer treatment for advanced gastric cancer, we have shown that oncologists’ estimates of EST were unbiased, moderately discriminative, and independently significant predictors of OS. Simple multiples of the EST were remarkably accurate at estimating worst‐case, typical, and best‐case scenarios for survival, a format preferred by patients. Further research is needed to determine if estimates of postprogression survival can be based on time to progression on last‐line therapy, but the ratio of median PPST/TTP from the group of patients in a trial multiplied by the TTP in an individual may be a reasonable starting point for estimating scenarios. Helping oncologists with their estimation and explanations of survival time should improve their patients’ understanding of prognosis and lead to better informed treatment decisions, better plans for the future, better directives for end of life care, and more timely referrals to palliative care.
Acknowledgments
We thank the patients and investigators for their participation in this study. This study was conducted by the Australasian Gastro‐Intestinal Trials Group in collaboration with the National Health and Medical Research Council Clinical Trials Centre, University of Sydney. This work was supported by an NHMRC Clinical Trials Centre, PhD Scholarship.
Author Contributions
Conception/design: Anuradha Vasista, Martin Stockler, Belinda Kiely
Providing study material: Anuradha Vasista, Martin Stockler, Andrew Martin, Nick Pavlakis, Katrin Sjoquist, David Goldstein, Sanjeev Gill, Vikram Jain, Geoffrey Liu, George Kannourakis, Yeul Hong Kim, Louise Nott, Stephanie Snow, Matthew Burge, Dean Harris, Derek Jonker, Yu Jo Chua, Richard Epstein, Antony Bonaventura, Belinda Kiely
Collection and/or assembly of data: Anuradha Vasista, Martin Stockler, Andrew Martin, Belinda Kiely
Data analysis and interpretation: Anuradha Vasista, Martin Stockler, Andrew Martin, Nick Pavlakis, Katrin Sjoquist, David Goldstein, Sanjeev Gill, Vikram Jain, Geoffrey Liu, George Kannourakis, Yeul Hong Kim, Louise Nott, Stephanie Snow, Matthew Burge, Dean Harris, Derek Jonker, Yu Jo Chua, Richard Epstein, Antony Bonaventura, Belinda Kiely
Manuscript writing: Anuradha Vasista, Martin Stockler, Andrew Martin, Nick Pavlakis, Katrin Sjoquist, David Goldstein, Belinda Kiely
Final approval of the manuscript: Anuradha Vasista, Martin Stockler, Andrew Martin, Nick Pavlakis, Katrin Sjoquist, David Goldstein, Sanjeev Gill, Vikram Jain, Geoffrey Liu, George Kannourakis, Yeul Hong Kim, Louise Nott, Stephanie Snow, Matthew Burge, Dean Harris, Derek Jonker, Yu Jo Chua, Richard Epstein, Antony Bonaventura, Belinda Kiely
Disclosures
Geoffrey Liu: AstraZeneca, Roche, Takeda, Novartis, Abbive, Bayer, Pfizer, Merck, Bristol‐Myers Squibb (H, RF, advisory board); Nick Pavlakis: Roche, AstraZeneca, Amgen, Novartis, Pfizer, Boerhinger Ingelheim, Merck Sharp & Dohme, Bristol‐Myers Squibb, Merck KgA, Takeda, Ipsen (C/A), Bayer (RF); Katrin Sjoquist: Pfizer (H), Ipsen, Amgen (Travel/conference support); Belinda Kiely: Novartis, Roche (H), Roche (SAB). The other authors indicated no financial relationships.
(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board
References
- 1.Wagner AD, Syn NL, Moehler M et al. Chemotherapy for advanced gastric cancer. Cochrane Database Syst Rev 2017;8:CD004064. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Yang D, Hendifar A, Lenz C et al. Survival of metastatic gastric cancer: Significance of age, sex and race/ethnicity. J Gastrointest Oncol 2011;2:77–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bernards N, Creemers GJ, Nieuwenhuijzen GA et al. No improvement in median survival for patients with metastatic gastric cancer despite increased use of chemotherapy. Ann Oncol 2013;24:3056–3060. [DOI] [PubMed] [Google Scholar]
- 4.Kiely BE, McCaughan G, Christodoulou S et al. Using scenarios to explain life expectancy in advanced cancer: Attitudes of people with a cancer experience. Support Care Cancer 2013;21:369–376. [DOI] [PubMed] [Google Scholar]
- 5.Kiely BE, Soon YY, Tattersall MH et al. How long have I got? Estimating typical, best‐case, and worst‐case scenarios for patients starting first‐line chemotherapy for metastatic breast cancer: A systematic review of recent randomized trials. J Clin Oncol 2011;29:456–463. [DOI] [PubMed] [Google Scholar]
- 6.Stockler MR, Tattersall MH, Boyer MJ et al. Disarming the guarded prognosis: Predicting survival in newly referred patients with incurable cancer. Br J Cancer 2006;94:208–212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kiely BE, Martin AJ, Tattersall MH et al. The median informs the message: Accuracy of individualized scenarios for survival time based on oncologists’ estimates. J Clin Oncol 2013;31:3565–3571. [DOI] [PubMed] [Google Scholar]
- 8.Pavlakis N, Sjoquinst KM, Martin AJ et al. Regorafenib for the treatment of advanced gastric cancer (INTEGRATE): A multinational placebo‐controlled phase II trial. J Clin Oncol 2016;34:2728–2735. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Christakis NA, Lamont EB. Extent and determinants of error in doctors’ prognoses in terminally ill patients: Prospective cohort study. BMJ 2000;320:469–472. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Llobera J, Esteva M, Rifà J et al. Terminal cancer. Duration and prediction of survival time. Eur J Cancer 2000;36:2036–2043. [DOI] [PubMed] [Google Scholar]
- 11.Harrell, FE Jr, Lee KL, Mark DB. Multivariable prognostic models: Issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361–387. [DOI] [PubMed] [Google Scholar]
- 12.Martin AJ, Gibbs E, Sjoquist K et al. Health‐related quality of life associated with regorafenib treatment in refractory advanced gastric adenocarcinoma. Gastric Cancer 2018;21:473–480. [DOI] [PubMed] [Google Scholar]
- 13.Taniyama TK, Hashimoto K, Katsumata N et al. Can oncologists predict survival for patients with progressive disease after standard chemotherapies? Curr Oncol 2014;21:84–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Gripp, S, Moeller S, Bölke E et al. Survival prediction in terminally ill cancer patients by clinical estimates, laboratory tests, and self‐rated anxiety and depression. J Clin Oncol 2007;25:3313–3320. [DOI] [PubMed] [Google Scholar]
- 15.Glare P, Virik K, Jones M et al. A systematic review of physicians’ survival predictions in terminally ill cancer patients. BMJ 2003;327:195–198. [DOI] [PMC free article] [PubMed] [Google Scholar]