Palliative prognostic scores for survival prediction of cancer patients: a systematic review and meta-analysis.

Cancer patients and their families seek information on prognosis for a variety of reasons, such as making informed medical decisions and addressing psychological and social considerations like facilitating farewells to loved ones and organizing personal affairs ( 1 ). Clinical expertise and knowledge are used to estimate patient survival ( 2 ). It is well known that such predictions tend to be inaccurate and overoptimistic ( 3-5 ). The accurate prediction of outcomes for cancer patients holds significance in various aspects, including their eligibility for palliative care services ( 6 ) and clinical trials ( 7 ) and the guidance it provides to clinicians in determining suitable treatment approaches. This includes assessing the potential benefits of further curative or palliative interventions, considering the patients’ projected life span and their ability to derive advantages from the proposed treatment ( 8 , 9 ).

However, available evidence for the superiority of prognostic prediction by any particular group of health-care professionals is currently inconsistent ( 10-12 ). Clinicians may be trained to convey bad news, but most are not trained to formulate or use prognostic estimations ( 13-15 ). The palliative prognostic score is one of many prognostic measures developed and validated to give clinicians a more objective and standardized prediction.

The palliative prognostic score was developed in Italy. Dyspnea, anorexia, Karnofsky Performance Status, clinician estimated survival, total white blood cells, and lymphocyte percentage are scored ( 16 ). Patients are categorized into 3 risk groups, which indicate their probability of surviving 30 days: group A (score < 5.5; 30-day survival probability > 70%), group B (score = 5.6-11.0; 30-day survival probability = 30%-70%), and group C (score = 11.1-17.5, 30-day survival probability < 30%) ( 16 ). A subsequent study confirmed the accuracy of groups A, B, and C with survival probabilities of 86.6%, 51.6%, and 16.9%, respectively ( 17 ).

The palliative prognostic score is the most widely validated palliative prognostic scale, tested in different populations and settings ( 13 ) and recommended by the European Association of Palliative Care ( 18 ). The European Society for Medical Oncology has suggested palliative prognostic score as a multivariable predictive model for clinicians to evaluate advanced cancer patients with a weeks-to-month prognosis and no longer receiving disease-modifying treatment ( 19 ). Research has shown that clinicians were generally better at identifying patients who are imminently dying because of the horizon effect ( 8 , 20 ) and likely to live for more than 1 year compared with predicting those with survival of weeks and months, suggesting that there is considerable prognostic uncertainty for those with survival of weeks to months ( 21 ). The use of prognostic tools such as the palliative prognostic score may be a valuable adjunct to enhance the accuracy of clinician-prediction survival.

After demonstrating that delirium and palliative prognostic score are independent prognostic factors for worse survival outcomes in advanced cancer patients, researchers thought combining them would improve survival prediction ( 22 ). The delirium–palliative prognostic score has the same components and partial scores as palliative prognostic score, but delirium adds 2 points, reaching a maximum score of 19.5 ( 23 ). Survival probabilities are the same as palliative prognostic score, but score boundaries are altered (group A: 0.0-7.0; group B: 7.1-12.5; group C: 12.6-19.5) ( 23 ). Some studies have modified the palliative prognostic score, for example, by excluding clinician prediction of survival ( 24 , 25 ). As clinician prediction often considers objective variables like dyspnea, anorexia, and Karnofsky Performance Status, it is unknown if combining them with a subjective variable like clinician-prediction survival could improve or worsen prediction accuracy ( 25 ). However, it remains unclear which version performs better as studies that directly compare tool performance are rare ( 1 , 26 , 27 ).

The only systematic review on the palliative prognostic score compared its performance with clinician survival predictions but did not conduct a meta-analysis ( 27 ). Areas that have yet to be adequately explored are the evaluation of the palliative prognostic score as a prognostic model and its accuracy in predicting actual survival. This information is crucial for physicians as it would provide insights into the reliability of the palliative prognostic score in assisting them with estimating patient survival. Whether the palliative prognostic score is an independent prognostic factor has yet to be established in a systematic review. Furthermore, no systematic review has compared the performance of the different palliative prognostic score tools. It is critical to evaluate whether one tool is better than the other so that clinicians can make an informed choice in selecting the appropriate prognostic tool.

This systematic review and meta-analysis aimed to 1) evaluate the model performance of the palliative prognostic score tools (ie, palliative prognostic score, delirium–palliative prognostic score, and palliative prognostic score without clinician prediction of survival) in predicting 30-day survival of cancer patients; 2) compare the performance of the different palliative prognostic score tools in predicting 30-day survival; and 3) evaluate the prognostic utility of these tools.

Methods

This review was reported according to the Transparent Reporting of a Multivariable Prediction model for Individual Prognosis or Diagnosis–Systematic Reviews and Meta-analyses guidelines ( 28 ) and other related guidelines ( 29-31 ). Its protocol was registered in the International Prospective Registry of Systematic Reviews (PROSPERO; CRD42023447139).

Eligibility criteria

Detailed criteria are presented in Table 1 . Studies in English must evaluate any of the 3 prognostic scores in adults aged 18 years and older with cancer at any stage and of any type. The performance or prognostic utility must be reported for 30-day survival using discrimination, calibration measures, hazard ratios (HRs), and so forth. Other survival durations are excluded. The studies can be conducted in any setting, and prospective and retrospective designs are eligible. Peer-reviewed articles and dissertations and theses were included.

Search strategy

We searched PubMed, Embase, ScienceDirect, Web of Science, ProQuest Theses and Dissertations, and Google Scholar for relevant articles from inception to June 2, 2023 ( Supplementary Methods , available online). The reference lists of relevant research and reviews were examined. We used a 3-step search strategy. First, terms such as palliative prognostic score , predict , and cancer were searched in PubMed to generate keywords and Medical Subject Headings. Second, the terms were then used to search other databases. Finally, we searched Google Scholar and ProQuest for grey literature. The records generated were imported into Rayyan, and duplicates were deleted. Two authors (SK and PB) independently identified potential papers using titles and abstracts. They independently evaluated full-text publications for eligibility. Discrepancies were resolved with a third author (SQY).

Data extraction

A data extraction form was piloted with 5 studies and revised as necessary. Data extraction was conducted by PB, and SK, SQY, and DP verified the data. Information such as year of publication, country, study design, diagnoses, and participant characteristics were extracted. Calibration, discrimination, and reclassification measures were also extracted. Any inconsistencies were discussed until everyone agreed on the data to be extracted.

Data analysis

Random effects meta-analysis of C-statistics

The C-statistic is an estimated conditional probability that for any pair of participants, the predicted risk of an event is higher for the subject who experienced the outcome compared with another subject who did not experience the outcome ( 32 ). For binary outcomes, C-statistic is equivalent to the area under curve (AUC) (ie, alive or dead at 30 days) ( 33 ). Pooled AUC is the primary outcome of this review. For time-to-event outcomes, the C-index is used, which examines a continuous outcome and measures the proportion of patient pairs where predicted, and observed survival outcomes are concordant ( 29 ).

C-statistics were pooled in Revman 5.4 using random effects generic inverse variance meta-analysis. The value of the C-statistic indicates the level of discrimination (0.5, no discrimination; 0.7-0.8, acceptable; 0.8-0.9, excellent; >0.9, outstanding) ( 34 ). A value of 0.5 indicates that the model is no better than chance, while 1 indicates perfect discrimination ( 33 ). Statistical significance of effect sizes was set at a P value less than .05. Test performance was compared using subgroup analysis based on the tool used (eg, palliative prognostic score vs delirium–palliative prognostic score).

Cochran Q test and I 2 statistic were used to evaluate heterogeneity. Statistically significant heterogeneity was set at a P value less than .10. Heterogeneity was unimportant when I 2 equaled 0%-40%, moderate when I 2 equaled 30%-60%, substantial when I 2 equaled 50%-90%, and considerable when I 2 equaled 75%-100%. Statistically significant subgroup difference was defined as a P value less than .10 ( 35 ). If at least 10 studies were in a meta-analysis, heterogeneity was explored using subgroup analysis based on study design, methodological quality, and study setting. If at least 10 studies were in a meta-analysis, publication bias was assessed using visual inspection of funnel plots and Begg and Egger linear regression tests using Jamovi version 2.3.21 ( 36 ).

Diagnostic test accuracy meta-analysis

When possible, we reconstructed a 2x2 table using formulas from Taylor ( 37 ) ( Supplementary Methods , available online) to calculate true positive, true negative, false positive, and false negative values required for meta-analysis. Using MetaDTA, the pooled diagnostic odds ratio with 95% confidence interval (CI) were obtained using the hierarchical summary receiver operating characteristic model ( 38 ). The hierarchical summary receiver operating characteristic model estimates a summary receiver operating characteristic curve across different thresholds and is useful for evaluating test performance across a range of thresholds ( 39 ). Diagnostic odds ratio ranges from 0 to infinity; higher values indicate better test performance ( 40 ). When studies investigated multiple thresholds, the most commonly used threshold was selected. Heterogeneity was quantified using variances of logit-transformed sensitivity and specificity. Although heterogeneity can be quantified, it is not easily interpreted as it represents variation in parameters expressed on logit scales, and methods for describing heterogeneity in diagnostic test meta-analyses are still being developed ( 41 , 42 ).

If at least 10 studies were in a meta-analysis, heterogeneity was explored using meta-regression and sensitivity analysis. Meta-regression was performed using Meta-Disc 2.0 ( 43 ) to examine the effect of covariates (study design, methodological quality, and study setting) on model performance. As recommended by the Preferred Reporting Items for Systemic Reviews and Meta-Analyses guidelines for diagnostic test accuracy meta-analyses, publication bias would not be assessed ( 31 ).

To compare the performance of the different tools, because of the limitations of Meta-Disc 2.0, we could only utilize the bivariate model by including the type of tool as a covariate to assess differences in pooled sensitivity and specificity. We conducted indirect comparisons (between-study comparisons) as a limited number of studies compared the tools directly (within the same study) ( 44 ).

Survival analysis

Because the palliative prognostic score categorizes patients into 3 risk groups with different 30-day survival probabilities (group A: >70% chance of survival; group B: 30%-70%; group C: <30%), we descriptively evaluated whether the actual survival of the patients corresponded to the survival probabilities as suggested by palliative prognostic score.

Constructing Kaplan–Meier survival curves for risk groups is an informal way to assess discrimination, and a wider separation of curves indicates better discrimination ( 45 ). Although log-rank or Cox tests between risk groups are often conducted, they are not recommended to analyze discrimination, and evaluating hazard ratios across risk groups is preferred. However, hazard ratios are rarely reported ( 45 ). Hence, we extracted and narratively described the findings of the survival curves, log-rank, and Cox tests when studies reported them and meta-analyzed hazard ratios where possible. Random effects meta-analysis of hazard ratios was conducted using Revman 5.4. We planned to synthesize 1) adjusted and unadjusted effect measures, 2) effect measures within similar categories, and 3) continuous effect measures separately ( 46 ).

Narrative synthesis of other analyses

Findings that could not be included in the meta-analysis (eg, dissimilar outcomes, inadequate data) were narratively summarized and/or presented in tables. The calibration performance of the tools was narratively summarized as well.

Quality assessment

The risk of bias was evaluated independently by SQY and PB, and any discrepancies were discussed until a consensus was reached. Figures were visualized using Robvis ( 47 ).

For studies evaluating palliative prognostic score as a prognostic model, the Prediction Model Risk of Bias Assessment Tool was used to assess the risk of bias and applicability using 4 domains (participants, predictors, outcomes, and analysis) ( 48 ).

Studies that evaluated palliative prognostic scores solely as a prognostic factor were assessed using the Quality In Prognosis Studies tool. It considers 6 domains: 1) study participation, 2) study attrition, 3) prognostic factor measurement, 4) outcome measurement,( 5 ) study confounding, and 6) statistical analysis and reporting, which were rated as high, moderate, or low risk of bias ( 49 , 50 ). An overall risk of bias was rated for each study. If all 6 domains had a low risk of bias or up to 1 moderate domain, the study would be rated as low risk of bias. If there were at least 1 high risk of bias domain or 3 moderate risk of bias domains, the study would be at high risk of bias. Studies in between would be rated as moderate risk of bias ( 49 ).

Results

Search results

The initial search yielded 409 articles. After removing duplicates, 329 articles were screened using title and abstract, and 48 full texts were assessed against the eligibility criteria, resulting in 39 studies included in the review. The Preferred Reporting Items for Systemic Reviews and Meta-Analyses diagram depicting the study selection process is shown in Figure 1 .

Characteristics of included studies

The studies were published from 1999 to 2023 ( 16 , 17 , 23-25 , 51-84 ) ( Table 2 ). They were conducted in 13 countries, most commonly in Japan (n = 12), Italy (n = 9), and Australia (n = 4). Of the studies 26 were prospective, and 13 were retrospective. One study was a dissertation ( 51 ), and the rest were peer-reviewed. Four studies ( 23 , 52 , 70 , 83 ) investigated delirium–palliative prognostic score, while others investigated palliative prognostic score with or without clinician-predicted survival.

Most studies were conducted in palliative care units and hospices or patients were seen by the palliative care team (home-based care or inpatient). Two were in an oncology ward ( 57 , 76 ), and some studies included outpatients ( 54 , 69 , 72 ). Three studies did not provide details on the settings ( 23 , 24 , 51 ) or merely reported that it was in the hospital ( 66 ). Most studies included patients with various cancer diagnoses, while the other studies focused on a particular cancer type, such as lung cancer ( 24 , 66 , 71 , 74 ). Most patients were receiving palliative care, while some were concurrently receiving palliative treatment such as chemotherapy. There were 10 617 patients included in this systematic review. The sample size ranged from 44 to 2072 patients. The age of patients ranged from 18 to 102 years.

Results of risk of bias assessment

Most studies assessed using Prediction Model Risk of Bias Assessment Tool were at high risk of bias (32 of 33 studies), except for 1 that had low risk of bias ( 61 ), and all studies had low applicability concerns ( Supplementary Figure 1 , available online). Most studies did not specify who scored the palliative prognostic score and/or the characteristics of clinicians who did the scoring. This is important information that should have been included as clinician characteristics may influence the accuracy of clinician prediction of survival and, thus, palliative prognostic score. Although most studies assessed discrimination, calibration was only evaluated in 1 study using a calibration curve where most data points lay near the reference line, showing satisfactory agreement between the observed and predicted probabilities ( 61 ).

Studies assessed using Quality In Prognosis Studies mainly were at high risk of bias (4 of 6 studies). Most studies did not adequately address if there was study attrition, nor did they account for confounders in survival analysis.

Synthesis results

The main findings of the review are found in Box 1 .

Palliative prognostic score vs palliative prognostic score without clinician-predicted survival

Meta-analysis of 11 studies ( 25 , 51 , 55 , 58 , 61 , 63 , 67 , 73 , 79 , 82 , 83 ) found that pooled AUC of palliative prognostic score was 0.82 (95% CI = 0.79 to 0.84; P < .00001), which indicated acceptable-excellent discrimination. However, heterogeneity was substantial ( P < .00001, I 2 = 85%) ( Figure 2, A ).

Meta-analysis of 4 studies ( 24 , 25 , 61 , 82 ) found that pooled AUC of palliative prognostic score without clinician-predicted survival was slightly lower at 0.74 (95% CI = 0.71 to 0.78; P < .00001), which shows acceptable discrimination. Similarly, heterogeneity was substantial ( P < .0001, I 2 = 78%) ( Figure 2, A ).

Subgroup analysis found a statistically significant subgroup difference ( P = .001, I 2 = 90.4%) ( Figure 2, A ), showing that the discrimination performance of palliative prognostic score was significantly better than palliative prognostic score without clinician-predicted survival, highlighting that clinician-predicted survival is actually beneficial in improving palliative prognostic score performance. However, clinician-predicted survival tends to overestimate survival in general ( 85 ). In contrast, 1 included study found that through net reclassification index and integrated discrimination improvement analysis, palliative prognostic score without clinician-predicted survival was more accurate than palliative prognostic score ( 25 ).

As the meta-analysis for palliative prognostic score had statistically significant heterogeneity and at least 10 studies, subgroup analysis was conducted. There was no statistically significant subgroup difference based on the study setting ( P = .55, I 2 = 0%) ( Figure 2, B ) and study design ( P = .54, I 2 = 0%) ( Figure 2, C ). Subgroup analysis based on methodological quality was not conducted as most studies had a high risk of bias. Publication bias was unlikely, as Kendall τ ( P = .113) and Kendall regression ( P = .530) were nonsignificant, and the funnel plot appeared symmetric ( Figure 2, D ).

Meta-analysis of 6 studies ( 25 , 63 , 69 , 70 , 72 , 82 ) reporting C index for palliative prognostic score was 0.72 (95% CI = 0.68 to 0.76; P < .00001) ( Figure 2, E ). Heterogeneity was considerable ( P < .00001, I 2 = 93%). Meta-analysis of 2 studies ( 25 , 82 ) for C index of palliative prognostic score without clinician-predicted survival was 0.71 (95% CI = 0.68 to 0.73; P < .00001), and heterogeneity was not statistically significant ( P = .14, I 2 = 46%). Subgroup analysis found no statistically significant differences in C index between the 2 tools ( P = .63, I 2 = 0%).

Palliative prognostic score vs delirium-palliative prognostic score

Seven studies ( 51-53 , 67 , 69 , 70 , 83 ) in which the 2x2 table could be extracted were pooled in the comparative diagnostic test meta-analysis ( Figure 3, A ). Findings that could not be included in the meta-analysis are presented in Supplementary Table 1 (available online). Hierarchical summary receiver operating characteristic meta-analysis found that the pooled diagnostic odds ratio for palliative prognostic score was 7.514 (95% CI = 3.045 to 18.544). Logit(sensitivity) was 0.284 (95% CI = 0.673 to 1.241), while logit(specificity) was 1.733 (95% CI = 1.209 to 2.256) ( Figure 3, B ). In contrast, the pooled diagnostic odds ratio for delirium–palliative prognostic score was much higher at 13.507 (95% CI = 10.252 to 17.797). Logit(sensitivity) was 1.165 (95% CI = 0.662 to 1.668), and logit(specificity) was 1.438 (95% CI = 0.946 to 1.930) ( Figure 3, B ).

When conducting a meta-regression using the bivariate model, for the interpretation of a summary sensitivity and specificity to be clinically meaningful, only studies that used similar thresholds should be included in the meta-analysis ( 41 ). Hence, Kim et al. ( 67 ), using a threshold of 10, and Abraham ( 51 ), who did not specify the threshold used, were excluded from the analysis. The type of tool used did not affect sensitivity and specificity, as the relative sensitivity level (estimate = 0.784, 95% CI = 0.505 to 1.216; P = .281) and relative specificity level (estimate = 1.059, 95% CI = 0.920 to 1.242; P = .476) of palliative prognostic score and delirium–palliative prognostic score were not statistically significant ( Figure 3, C ). The global test comparison also found no statistically significant differences between palliative prognostic score and delirium–palliative prognostic score ( P = .544) ( Figure 3, C ), suggesting neither test outperformed the other.

Survival analyses as reported by studies

Most studies found that the palliative prognostic score accurately predicted 30-day survival probabilities, except for Glare and Virik ( 56 ) group A; Lee et al. ( 68 ) group A; Maltoni et al. ( 69 ) groups B and C; Yoon et al. ( 83 ) groups A and B; and Tassinari et al. ( 81 ) group B. Median survival ranged from 29 days ( 83 ) to 11 months ( 66 ) for group A, 18 ( 70 ) to 95 days ( 69 ) for group B, and 0.6 weeks ( 72 ) to 48 days ( 69 ) for group C ( Table 3 ).

Hazard ratios were not meta-analyzed because it was unclear if they were univariate or multivariate analyses, and there were several inconsistent findings that placed doubt on the studies’ data. Results from meta-analyses may be misleading. Most studies reported effect sizes confirming higher-risk groups had shorter survival than lower-risk groups ( Table 4 ). However, Chung et al. ( 53 ) reported that palliative prognostic score for group B (univariate HR = 0.61, 95% CI = 0.44 to 0.84) was a statistically significant predictor of 30-day survival, which seemed to suggest palliative prognostic score was a protective factor, contradictory to the increased risks of death in groups B and C in the palliative prognostic score model. It is unclear if palliative prognostic scores for group B were analyzed as a categorial or continuous variable. The variables adjusted in the multivariate analysis and the reference category used were not stated ( 53 ). Hyodo et al. ( 65 ) reported that hazard ratios for group B vs A was 0.536 (95% CI = 0.360 to 0.799), which again contradicts the palliative prognostic score model. No explanations were provided for the surprising results in both articles. Details on adjusted and unadjusted effect sizes reported by the studies are presented in Table 4 . Categorical hazard ratios were most commonly reported, and only 1 study analyzed palliative prognostic score without clinician-predicted survival ( 24 ).

Discussion

This systematic review and meta-analysis found that the palliative prognostic score (pooled AUC = 0.82, 95% CI = 0.79 to 0.84; pooled C index = 0.72, 95% CI = 0.68 to 0.76) and the palliative prognostic score without clinician-predicted survival (pooled AUC = 0.74, 95% CI = 0.71 to 0.78; pooled C index = 0.71, 95% CI = 0.68 to 0.73) are accurate prognostic tools. However, palliative prognostic score performed better than palliative prognostic score without clinician-predicted survival in terms of AUC, suggesting that palliative prognostic score should be used in its original form wherever possible. There were no statistically significant differences between palliative prognostic score and delirium–palliative prognostic score performance. Most studies reported survival probabilities corresponding to the palliative prognostic score risk groups, and higher risk groups were significantly associated with shorter survival.

The advantage of palliative prognostic score over palliative prognostic score without clinician-predicted survival could be because subjective clinical judgement is important to integrate the more intangible elements into the task of formulating prognoses. Predicting patient survival depends on many factors clinicians assess that the objective variables of palliative prognostic score may not capture, as evidenced by the ever-growing literature on prognostic factors associated with survival in cancer patients. Some credit must be given to the experience clinicians have gained over the years in having some intuition on the patient’s potential poor prognosis through pattern recognition, such as physical or psychological signals ( 86 ), deterioration rate, and symptom burden ( 87 ). This supports the use of palliative prognostic score together with clinician prediction to improve accuracy ( 11 ).

There were no statistically significant differences in performance for delirium–palliative prognostic score and palliative prognostic score in the comparative diagnostic meta-analysis. This could be limited by the sample sizes as fewer patients were assessed using delirium–palliative prognostic score (n = 3004) compared with palliative prognostic score (n = 4029) ( Figure 3, A ). In addition, the meta-analysis assessed relative sensitivity and specificity regardless of risk groups. In the development study, where 27.4% had delirium, the survival difference between patients with or without delirium was highest in group C (group A: 0.9%; group B: 4.4%; group C: 6.2%), and the presence of delirium was highest in group C (group A: 10.1%; group B: 25.9%; group C: 60.8%). The 30-day survival probabilities were very similar for palliative prognostic score and delirium–palliative prognostic score in groups A (87% vs 83%) and B (51% vs 50%) but not group C (16% vs 9%), respectively ( 16 ). Hence, it is likely that delirium–palliative prognostic score could better classify patients in group C, although there were marginal differences in its overall performance compared with palliative prognostic score. In Maltoni et al. ( 70 ) where 28.2% had delirium, this relationship was not verified, as the 30-day survival probability was the same in group C for both tools. They commented that as both tools had comparable performance, palliative prognostic score modification was unnecessary ( 70 ). Yoon et al. ( 83 ), with 19.1% having delirium, reported lower survival probabilities for all 3 risk categories in delirium–palliative prognostic score compared with palliative prognostic score.

Delirium is a recognized risk factor for increased mortality in cancer ( 22 , 88 ). Still, it may be difficult to distinguish between reversible delirium and delirium due to nearing the end of life. Delirium is frequently underdiagnosed in cancer ( 89 ), and accurate assessment using validated tools requires training and standardization ( 89 , 90 ), which could have affected the results in the included studies. As only 4 studies assessed delirium–palliative prognostic score, with 3 included in the meta-analysis, more research is needed to investigate which tool is more useful and under what circumstances.

Thirteen of the 18 studies that reported survival probabilities confirmed the probabilities suggested by the palliative prognostic score tool ( Table 3 ), suggesting that palliative prognostic score generally can discriminate patients between the 3 risk groups. For Lee et al. ( 68 ) and Glare et al. ( 56 ), in group A, the survival probabilities (69.4% and 66%, respectively) were slightly lower than expected (>70%). Yoon et al. ( 83 ) groups A and B had much smaller survival probabilities, which can be explained by the lower median survival time compared with the development study ( 16 ). Actual survival probabilities for Maltoni et al. ( 69 ) groups B (76%) and C (100%) were much higher than expected, likely because of the long median survival and that the patients being evaluated for suitability for palliative radiotherapy, who are likely to have longer survival than those near the end of life in the development study ( 16 ). Tassinari et al. ( 81 ) similarly recruited patients with advanced solid tumors receiving chemotherapy, and their median survival was much higher, with no patients in group C. These suggest that palliative prognostic score is accurate in discriminating risk groups when the median survival of patients is 2 months or less, following the development study ( 16 ). Palliative prognostic score may have limited utility in patients with much longer or shorter life expectancy.

There are several reasons for promoting accurate prognostication from the perspectives of the patient and the clinician. Prognostic uncertainty for patients is a known cause of distress and poorer quality of life, which is particularly prevalent in hospitalized cancer patients before consultation with palliative care and with late enrollment in hospice or less hospice use ( 5 ). This means that all the benefits of specialist support for the end of life are withheld from many people and families who could benefit. Lewis et al. ( 91 ) extend this reasoning much further. They describe delays in identifying the risk of death and not informing patients who are nearing the end of life and their families as health-related harm to health services worldwide.

For the clinician, lack of accuracy and overestimated prognosis can lead to the initiation or continuation of futile treatments of patients within the last months of life ( 24 ). It is obvious and understandable that it is difficult to make the call to stop treatments focused on life prolongation and to focus on comfort, which makes the utility of the 30-day survival prediction such an important quality benchmark for assessing aggressive interventions near the end of life with a consequent focus on quality of life. A study concluded that early intervention in patients with a history of hospitalization before their last month may prevent aggressive interventions during the last month of life and help caregivers be more prepared for death during bereavement ( 92 ). A reduction in medical costs through futile treatment then follows naturally ( 93 , 94 ).

The palliative prognostic score, therefore, has a important place in assisting the clinician through confirming clinical judgement and instilling confidence to communicate with patients with cancer and their families. However, the reliance on blood tests makes it less attractive for those already under the care of palliative care or hospice services. Phlebotomy is avoided in palliative settings to ensure patient comfort ( 95 , 96 ), and it may be necessary only when it can change patient symptom management ( 96 , 97 ). During the assessment of patients in palliative care and who are close to the end of life, these patients do not usually have their blood taken or undergo any other invasive tests ( 56 ). This suggests that other prognostic decision aids such as the Palliative Prognostic Index should be used instead ( 98 ). Furthermore, the palliative prognostic score requires clinician-predicted survival, which many clinicians may not be confident in estimating ( 99-101 ). Other objective prognostic tools, which are easily scored and require minimal experience, could be used instead, such as the Palliative Prognostic Index ( 98 ) or the Palliative Performance Scale ( 102 ).

Using a prognostic tool can help clinicians through the confidence of having their clinical judgement confirmed or validated, and this leads to the initiation of conversations with patients and their families to plan well for end-of-life care ( 103 ). The question now remains as to how to encourage the use of prognostic tools and explore the experiences of clinicians and patients in giving and receiving a prognosis using a tool so that best practices can be established in translating the evidence to practice.

Future studies should investigate if there are differences in the prognostic accuracy of the palliative prognostic score depending on clinician characteristics (eg, years of experience in palliative care or oncology, rank). This would further inform whether the palliative prognostic score would suit clinicians of different experiences and backgrounds.

Most studies in this review were at high risk of bias, primarily because calibration was not assessed, similar to previous systematic reviews on prediction models ( 98 , 104 ). The findings of this review should thus be interpreted with caution. Future validation studies should adhere to the Transparent Reporting of a Multivariable Prediction model for Individual Prognosis or Diagnosis guidelines to ensure a comprehensive analysis of palliative prognostic score performance and improve the quality of the studies. Reporting adjusted hazard ratios is also important to assess whether palliative prognostic score is an independent predictor of survival, which was rarely analyzed or reported adequately by the included studies. Future studies should follow the Prognosis Research Strategy framework to facilitate validation of palliative prognostic score as a prognostic factor and for future meta-analyses ( 105 ).

Finally, understanding from the clinician and patient perspectives how prognostic information can be given using a tool will lead to the establishment of best practices for prognosis in cancer and better end-of-life care planning and transition to palliative and hospice services.

In this review, AUC was primarily used to evaluate the performance of palliative prognostic score tools. This is because the studies included in the meta-analysis commonly reported only AUC, and a limited number of studies reported adequate data for diagnostic test meta-analysis. However, AUCs summarize test accuracies across all thresholds, and they do not provide information on error rates such as false positives and false negatives, unlike other diagnostic test measures such as specificities and sensitivities ( 39 ).

The studies that evaluated palliative prognostic score without clinician-predicted survival also did not assess if the partial scores would remain the same as the original palliative prognostic score prediction model. They either calculated the scores as a sum of the other variables with a total score of 9 ( 24 , 25 , 82 ) or scaled it up to the range of the original scale ( 84 ), essentially reporting results of a new prediction model that has yet to be formally developed ( 48 ). Hence, the evidence available to compare palliative prognostic score and palliative prognostic score without clinician-predicted survival is limited, and the findings of this comparison should be interpreted with caution ( Figure 2, A ). Future researchers should conduct a development and validation study on palliative prognostic score without clinician-predicted survival to assess its accuracy compared with palliative prognostic score.

Most studies (37 of 39) were conducted in palliative settings, and clinician-predicted survival is likely to be estimated by palliative care clinicians. Hence, the findings of this review may not be generalizable to cancer patients in general and clinicians in other specialties.

This review provided important evidence to show that palliative prognostic score is an accurate tool to estimate 30-day survival of cancer patients. Palliative prognostic score is not intended to replace clinician-predicted survival, but rather clinicians may consider using palliative prognostic score together with clinical judgement to enhance confidence in their prognostic estimates and accuracy. The included studies show that the palliative prognostic score can reliably discriminate between risk groups and usually corresponds to the survival probabilities as suggested. The findings of this review suggest that palliative prognostic score is most suitable for use, as it is more accurate than palliative prognostic score without clinician-predicted survival, and relatively fewer studies have evaluated delirium–palliative prognostic score.

Supplementary Material

Acknowledgements

Not applicable.

Data availability

The data underlying this article are available in the article and its online supplementary material .

Author contributions

Si Qi Yoong, BSN (Honors) (Conceptualization; Formal analysis; Investigation; Methodology; Visualization; Writing—original draft; Writing—review & editing), Priyanka Bhowmik, MBBS (Formal analysis; Investigation; Writing—review & editing), Sreerag Kapparath, MBBS (Investigation), and Davina Porock, PhD (Writing – original draft; Writing—review & editing).

Funding

This research received no specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Conflicts of interest

The authors declare no conflicts of interest.

Main findings of this review a

1. Performance of palliative prognostic score

Pooled AUC = 0.82 (95% CI = 0.79 to 0.84)

Pooled C-index = 0.72 (95% CI = 0.68 to 0.76)

Performance of palliative prognostic score without clinician-predicted survival

Pooled AUC = 0.74 (95% CI = 0.71 to 0.78)

Pooled C-index = 0.71 (95% CI = 0.68 to 0.73)

Palliative prognostic score and palliative prognostic score without clinician-predicted survival are accurate prognostic tools. However, palliative prognostic score performed better in terms of AUC, suggesting that palliative prognostic score should be used in its original form.

2. There were no statistically significant differences between palliative prognostic score and delirium–palliative prognostic score performance in the comparative diagnostic test meta-analysis.

3. Most studies reported survival probabilities that corresponded to the palliative prognostic score risk groups, and higher-risk groups were statistically significantly associated with shorter survival.

a AUC = area under curve; CI = confidence interval.

References

1 Chu C , Anderson R , White N , Stone P. Prognosticating for adult patients with advanced incurable cancer: a needed oncologist skill . Curr Treat Options Oncol . 2020 ; 21 ( 1 ): 5 . doi: 10.1007/s11864-019-0698-2 31950387 PMC6965075

2 Martin EJ , Widera E. Prognostication in serious illness . Med Clin North Am . 2020 ; 104 ( 3 ): 391 - 403 . doi: 10.1016/j.mcna.2019.12.002 32312405

3 Chow E , Harth T , Hruby G , Finkelstein J , Wu J , Danjoux C. How accurate are physicians’ clinical predictions of survival and the available prognostic tools in estimating survival times in terminally ill cancer patients? A systematic review . Clin Oncol (R Coll Radiol) . 2001 ; 13 ( 3 ): 209 - 218 . doi: 10.1053/clon.2001.9256 11527298

4 Glare P , Virik K , Jones M , et al A systematic review of physicians’ survival predictions in terminally ill cancer patients . BMJ . 2003 ; 327 ( 7408 ): 195 - 198 . 12881260 10.1136/bmj.327.7408.195 PMC166124

5 Gramling R , Gajary-Coots E , Cimino J , et al Palliative care clinician overestimation of survival in advanced cancer: disparities and association with end-of-life care . J Pain Symptom Manage . 2019 ; 57 ( 2 ): 233 - 240 . doi: 10.1016/j.jpainsymman.2018.10.510 30391655

6 Grądalski T. Medical referral criteria for palliative care in adults: a scoping review . Pol Arch Intern Med . 2022 ; 132 ( 3 ): 16223 . doi: 10.20452/pamw.16223 35243858

7 Simms L , Barraclough H , Govindan R. Biostatistics primer: what a clinician ought to know—prognostic and predictive factors . J Thorac Oncol . 2013 ; 8 ( 6 ): 808 - 813 . doi: 10.1097/JTO.0b013e318292bdcd 23676560

8 Chu C , White N , Stone P. Prognostication in palliative care . Clin Med (Lond) . 2019 ; 19 ( 4 ): 306 - 310 . doi: 10.7861/clinmedicine.19-4-306 31308109 PMC6752241

9 Kutzko JH , Dadwal P , Holt T , Rahman MA , Zahir SF , Hickey B. Defining the expected 30-day mortality for patients undergoing palliative radiotherapy: a meta-analysis . Radiother Oncol . 2022 ; 168 : 147 - 210 . doi: 10.1016/j.radonc.2022.01.030 35101462

10 White N , Reid F , Harris A , Harries P , Stone P. A systematic review of predictions of survival in palliative care: how accurate are clinicians and who are the experts? PLoS One . 2016 ; 11 ( 8 ): e0161407 . doi: 10.1371/journal.pone.0161407 27560380 PMC4999179

11 Tavares T , Oliveira M , Gonçalves J , et al Predicting prognosis in patients with advanced cancer: A prospective study . Palliat Med . 2018 ; 32 ( 2 ): 413 - 416 . doi: 10.1177/0269216317705788 28488922

12 Urahama N , Sono J , Yoshinaga K. Comparison of the accuracy and characteristics of the prognostic prediction of survival of identical terminally ill cancer patients by oncologists and palliative care physicians . Jpn J Clin Oncol . 2018 ; 48 ( 7 ): 695 - 698 . doi: 10.1093/jjco/hyy080 29850870

13 Glare P , Sinclair CT , Stone P , Clayton JM. Predicting survival in patients with advanced disease . In: Cherny N , Fallon M , Kaasa S , Portenoy RK , Currow DC , eds. Oxford Textbook of Palliative Medicine . Oxford : Oxford University Press; 2015 :65-76. doi: 10.1093/med/9780199656097.003.0007

14 Namavar AA , Vangala S , Pietras C , Dowling EP. Prognostication accuracy, confidence, and decisional conflict across training levels of internal medicine providers . J Palliat Care . 2020 ; 35 ( 4 ): 226 - 231 . doi: 10.1177/0825859719867672 31405315

15 Schoenborn NL , Bowman TLII , Cayea D , Pollack CE , Feeser S , Boyd C. Primary Care Practitioners’ Views on Incorporating Long-term Prognosis in the Care of Older Adults . JAMA Intern Med . 2016 ; 176 ( 5 ): 671 - 678 . doi: 10.1001/jamainternmed.2016.0670 27064895 PMC5564293

16 Pirovano M , Maltoni M , Nanni O , et al A new palliative prognostic score: a first step for the staging of terminally Ill cancer patients . J Pain Symptom Manage . 1999 ; 17 ( 4 ): 231 - 239 . doi: 10.1016/S0885-3924(98)00145-6 10203875

17 Maltoni M , Nanni O , Pirovano M , et al Successful validation of the palliative prognostic score in terminally ill cancer patients. Italian Multicenter Study Group on Palliative Care . J Pain Symptom Manage . 1999 ; 17 ( 4 ): 240 - 247 . doi: 10.1016/s0885-3924(98)00146-8 10203876

18 Maltoni M , Caraceni A , Brunelli C , et al ; Steering Committee of the European Association for Palliative Care . Prognostic factors in advanced cancer patients: evidence-based clinical recommendations—a study by the steering committee of the European Association for Palliative Care . J Clin Oncol . 2005 ; 23 ( 25 ): 6240 - 6248 . doi: 10.1200/JClinOncol.2005.06.866 16135490

19 Stone P , Buckle P , Dolan R , et al ESMO Guidelines Committee. Electronic address: Clinicalguidelines@esmo.org . Prognostic evaluation in patients with advanced cancer in the last months of life: ESMO Clinical Practice Guideline . ⋆. ESMO Open 2023 ; 8 ( 2 ): 101195 ., doi: 10.1016/j.esmoop.2023.101195 37087198 PMC10242351

20 Stone P , Chu C , Todd C , et al The accuracy of clinician predictions of survival in the Prognosis in Palliative care Study II (PiPS2): a prospective observational study . PLoS One . 2022 ; 17 ( 4 ): e0267050 . doi: 10.1371/journal.pone.0267050 35421168 PMC9009717

21 Orlovic M , Droney J , Vickerstaff V , et al Accuracy of clinical predictions of prognosis at the end-of-life: evidence from routinely collected data in urgent care records . BMC Palliat Care . 2023 ; 22 ( 1 ): 51 . doi: 10.1186/s12904-023-01155-y 37101274 PMC10131555

22 Caraceni A , Nanni O , Maltoni M , et al ; Italian Multicenter Study Group on Palliative Care Impact of delirium on the short term prognosis of advanced cancer patients . Cancer . 2000 ; 89 ( 5 ): 1145 - 1149 . doi: 10.1002/1097-0142(20000901)89:5<1145::aid-cncr24>3.0.co;2-x 10964345

23 Scarpi E , Maltoni M , Miceli R , et al Survival prediction for terminally ill cancer patients: revision of the palliative prognostic score with incorporation of delirium . Oncologist . 2011 ; 16 ( 12 ): 1793 - 1799 . doi: 10.1634/theoncologist.2011-0130 22042788 PMC3248779

24 De Giglio A , Tassinari E , Zappi A , et al The Palliative Prognostic (PaP) score without clinical evaluation predicts early mortality among advanced NSCLC patients treated with immunotherapy . Cancers . 2022 ; 14 ( 23 ): 5845 . doi: 10.3390/cancers14235845 36497326 PMC9739118

25 Hui D , Park M , Liu D , et al Clinician prediction of survival versus the Palliative Prognostic Score: which approach is more accurate? Eur J Cancer . 2016 ; 64 : 89 - 95 . doi: 10.1016/j.ejca.2016.05.009 27372208 PMC4969216

26 Hui D , Maxwell JP , Paiva CE. Dealing with prognostic uncertainty: the role of prognostic models and websites for patients with advanced cancer . Curr Opin Support Palliat Care . 2019 ; 13 ( 4 ): 360 - 368 . doi: 10.1097/SPC.0000000000000459 31689273 PMC7034625

27 Stone P , White N , Oostendorp LJM , Llewellyn H , Vickerstaff V. Comparing the performance of the palliative prognostic (PaP) score with clinical predictions of survival: a systematic review . Eur J Cancer . 2021 ; 158 : 27 - 35 . doi: 10.1016/j.ejca.2021.08.049 34649086

28 Snell KIE , Levis B , Damen JAA , et al Transparent reporting of multivariable prediction models for individual prognosis or diagnosis: checklist for systematic reviews and meta-analyses (TRIPOD-SRMA) . BMJ . 2023 ; 381 : e073538 . doi: 10.1136/bmj-2022-073538 37137496 PMC10155050

29 Debray TPA , Damen JAAG , Snell KIE , et al A guide to systematic review and meta-analysis of prediction model performance . BMJ . 2017 ; 356 : i6460 . doi: 10.1136/bmj.i6460 28057641

30 Moons KGM , de Groot JAH , Bouwmeester W , et al Critical appraisal and data extraction for systematic reviews of prediction modelling studies: the CHARMS checklist . PLOS Med . 2014 ; 11 ( 10 ): e1001744 . doi: 10.1371/journal.pmed.1001744 25314315 PMC4196729

31 Salameh JP , Bossuyt PM , McGrath TA , et al Preferred reporting items for systematic review and meta-analysis of diagnostic test accuracy studies (PRISMA-DTA): explanation, elaboration, and checklist . BMJ . 2020 ; 370 : m2632 . doi: 10.1136/bmj.m2632 32816740

32 Debray TP , Damen JA , Riley RD , et al A framework for meta-analysis of prediction model studies with binary and time-to-event outcomes . Stat Methods Med Res . 2019 ; 28 ( 9 ): 2768 - 2786 . doi: 10.1177/0962280218785504 30032705 PMC6728752

33 Caetano SJ , Sonpavde G , Pond GR. C-statistic: a brief explanation of its construction, interpretation and limitations . Eur J Cancer Oxf Engl 1990 . 2018 ; 90 : 130 - 132 . doi: 10.1016/j.ejca.2017.10.027 29221899

34 Hosmer DW , Lemeshow S , Sturdivant RX. Chapter 5: Assessing the fit of the model . In: Applied Logistic Regression . 3rd ed . Wiley Series in Probability and Statistics. Hoboken, NJ : Wiley; 2013 : 177 .

35 Higgins JPT , Thomas J , Chandler J , eds. Cochrane Handbook for Systematic Reviews of Interventions . 2nd ed . John Wiley & Sons ; 2019 .

36 The Jamovi Project . Jamovi . 2023 . https://www.jamovi.org . Accessed June 2, 2023.

37 Taylor K. Data Extraction in Meta-Analysis . https://www.cebm.ox.ac.uk/resources/data-extraction-tips-meta-analysis . Accessed February 1, 2022.

38 Rutter CM , Gatsonis CA. A hierarchical regression approach to meta-analysis of diagnostic test accuracy evaluations . Stat Med . 2001 ; 20 ( 19 ): 2865 - 2884 . doi: 10.1002/sim.942 11568945

39 Takwoingi Y , Riley RD , Deeks JJ. Meta-analysis of diagnostic accuracy studies in mental health . Evid Based Ment Health . 2015 ; 18 ( 4 ): 103 - 109 . doi: 10.1136/eb-2015-102228 26446042 PMC4680179

40 Glas AS , Lijmer JG , Prins MH , Bonsel GJ , Bossuyt PMM. The diagnostic odds ratio: a single indicator of test performance . J Clin Epidemiol . 2003 ; 56 ( 11 ): 1129 - 1135 . doi: 10.1016/S0895-4356(03)00177-X 14615004

41 Takwoingi Y , Deeks JJ. Chapter 16: Systematic reviews of diagnostic accuracy . In: Egger M , Higgins JPT , Smith GD , eds. Systematic Reviews in Health Research. Hoboken, NJ : John Wiley & Sons, Ltd; 2022 : 296 - 323 . doi: 10.1002/9781119099369.ch16

42 Plana MN , Pérez T , Zamora J. New measures improved the reporting of heterogeneity in diagnostic test accuracy reviews: a metaepidemiological study . J Clin Epidemiol . 2021 ; 131 : 101 - 112 . doi: 10.1016/j.jclinepi.2020.11.011 33227449

43 Plana MN , Arevalo-Rodriguez I , Fernández-García S , et al Meta-DiSc 2.0: a web application for meta-analysis of diagnostic test accuracy data . BMC Med Res Methodol . 2022 ; 22 ( 1 ): 306 . doi: 10.1186/s12874-022-01788-2 36443653 PMC9707040

44 Macaskill P , Takwoingi Y , Deeks JJ , Gatsonis C. Chapter 9: Understanding meta-analysis . In: Deeks JJ , Bossuyt PM , Leeflang M , Takwoingi Y , eds. Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy . 1st ed . Hoboken, NJ : The Cochrane Collaboration and John Wiley & Sons Ltd; 2023 : 203 - 248 .

45 Royston P , Altman DG. External validation of a Cox prognostic model: principles and methods . BMC Med Res Methodol . 2013 ; 13 ( 1 ): 33 . doi: 10.1186/1471-2288-13-33 23496923 PMC3667097

46 Riley RD , Moons KGM , Snell KIE , et al A guide to systematic review and meta-analysis of prognostic factor studies . BMJ . 2019 ; 364 : k4597 . doi: 10.1136/bmj.k4597 30700442

47 McGuinness LA , Higgins JPT. Risk‐of‐bias VISualization (robvis): an R package and Shiny web app for visualizing risk‐of‐bias assessments . Res Synth Methods . 2021 ; 12 ( 1 ): 55 - 61 . doi: 10.1002/jrsm.1411 32336025

48 Moons KGM , Wolff RF , Riley RD , et al PROBAST: a tool to assess risk of bias and applicability of prediction model studies: explanation and elaboration . Ann Intern Med . 2019 ; 170 ( 1 ): W1 - W33 . doi: 10.7326/M18-1377 30596876

49 Grooten WJA , Tseli E , Äng BO , et al Elaborating on the assessment of the risk of bias in prognostic studies in pain rehabilitation using QUIPS—aspects of interrater agreement . Diagn Progn Res . 2019 ; 3 ( 1 ): 5 . doi: 10.1186/s41512-019-0050-0 31093575 PMC6460536

50 Hayden JA , van der Windt DA , Cartwright JL , Côté P , Bombardier C. Assessing bias in studies of prognostic factors . Ann Intern Med . 2013 ; 158 ( 4 ): 280 - 286 . doi: 10.7326/0003-4819-158-4-201302190-00009 23420236

51 Abraham S. Comparing the Accuracy of Survival Prediction with Two Different Prognostic Scoring Systems, in Indian Cancer Patients Receiving Palliative Care/Palliative Radiotherapy . MD Dissertation. Vellore : Christian Medical College ; 2010 . http://repository-tnmgrmu.ac.in/264/ . Accessed August 6, 2023.

52 Baba M , Maeda I , Morita T , et al Survival prediction for advanced cancer patients in the real world: a comparison of the Palliative Prognostic Score, Delirium-Palliative Prognostic Score, Palliative Prognostic Index and modified Prognosis in Palliative Care Study predictor model . Eur J Cancer Oxf Engl 1990 . 2015 ; 51 ( 12 ): 1618 - 1629 . doi: 10.1016/j.ejca.2015.04.025 26074396

53 Chung MC , Tsai PY , Chen CM , Yang CK , Chang HH. Meridian energy analysis may predict the prognosis of patients with advanced cancers receiving palliative care . J Tradit Complement Med . 2023 ; 13 ( 4 ): 389 - 396 . doi: 10.1016/j.jtcme.2023.03.003 37396153 PMC10310863

54 Dewanti PC , Kurnianda J , Widayati K. Validation of the palliative prognostic score (PaP score) in patients with metastatic cancers in Dr Sardjito General Hospital Yogyakarta . Acta Interna J Intern Med . 2019 ; 9 ( 1 ): 15 - 23 . doi: 10.22146/actainterna.50900

55 Ermacora P , Mazzer M , Isola M , et al Prognostic evaluation in palliative care: final results from a prospective cohort study . Support Care Cancer . 2019 ; 27 ( 6 ): 2095 - 2102 . doi: 10.1007/s00520-018-4463-z 30229339

56 Glare P , Virik K. Independent prospective validation of the PaP score in terminally ill patients referred to a hospital-based palliative medicine consultation service . J Pain Symptom Manage . 2001 ; 22 ( 5 ): 891 - 898 . doi: 10.1016/S0885-3924(01)00341-4 11728792

57 Glare PA , Eychmueller S , McMahon P. Diagnostic accuracy of the palliative prognostic score in hospitalized patients with advanced cancer . J Clin Oncol . 2004 ; 22 ( 23 ): 4823 - 4828 . doi: 10.1200/JClinOncol.2004.12.056 15570085

58 Hamano J , Takeuchi A , Yamaguchi T , et al A combination of routine laboratory findings and vital signs can predict survival of advanced cancer patients without physician evaluation: a fractional polynomial model . Eur J Cancer . 2018 ; 105 : 50 - 60 . doi: 10.1016/j.ejca.2018.09.037 30391780

59 Hiratsuka Y , Kim D , Suh SY , et al ; EASED Investigators . Comparison of Objective Prognostic Score and Palliative Prognostic Score performance in inpatients with advanced cancer in Japan and Korea . Palliat Support Care . 2022 ; 20 ( 5 ): 662 - 670 . doi: 10.1017/S1478951521001589 36111731

60 Hiratsuka Y , Suh SY , Hui D , et al Are prognostic scores better than clinician judgment? A prospective study using three models . J Pain Symptom Manage . 2022 ; 64 ( 4 ): 391 - 399 . doi: 10.1016/j.jpainsymman.2022.06.008 35724924

61 Hiratsuka Y , Yoon SJ , Suh SY , et al Comparison of the accuracy of clinicians’ prediction of survival and Palliative Prognostic Score: an East Asian cross-cultural study . Support Care Cancer . 2022 ; 30 ( 3 ): 2367 - 2374 . doi: 10.1007/s00520-021-06673-0 34743238

62 Huang KS , Huang YH , Chen CT , Chou CP , Pan BL , Lee CH. Liver-specific metastases as an independent prognostic factor in cancer patients receiving hospice care in hospital . BMC Palliat Care . 2023 ; 22 ( 1 ): 62 . doi: 10.1186/s12904-023-01180-x 37221588 PMC10207633

63 Hui D , Ross J , Park M , et al Predicting survival in patients with advanced cancer in the last weeks of life: How accurate are prognostic models compared to clinicians’ estimates? Palliat Med . 2020 ; 34 ( 1 ): 126 - 133 . doi: 10.1177/0269216319873261 31564218

64 Hui D , Bansal S , Morgado M , Dev R , Chisholm G , Bruera E. Phase angle for prognostication of survival in patients with advanced cancer: preliminary findings . Cancer . 2014 ; 120 ( 14 ): 2207 - 2214 . doi: 10.1002/cncr.28624 24899148 PMC4079743

65 Hyodo I , Morita T , Adachi I , Shima Y , Yoshizawa A , Hiraga K. Development of a predicting tool for survival of terminally ill cancer patients . Jpn J Clin Oncol . 2010 ; 40 ( 5 ): 442 - 448 . doi: 10.1093/jjco/hyp182 20085908

66 Ikeguchi M , Kader A , Yoshimoto M , et al Usefulness of palliative prognostic score in the treatment of patients with non-resectable gastric cancer . Mol Clin Oncol . 2013 ; 1 ( 2 ): 253 - 256 . doi: 10.3892/mco.2013.66 24649156 PMC3915674

67 Kim AS , Youn CH , Ko HJ , Kim HM. The survival time of terminal cancer patients: prediction based on clinical parameters and simple prognostic scores . J Palliat Care . 2014 ; 30 ( 1 ): 24 - 31 . doi: 10.1177/082585971403000104 24826440

68 Lee SH , Lee JG , Choi YJ , et al Prognosis palliative care study, palliative prognostic index, palliative prognostic score and objective prognostic score in advanced cancer: a prospective comparison . BMJ Support Palliat Care . 2021 . doi: 10.1136/bmjspcare-2021-003077 34215569

69 Maltoni M , Scarpi E , Dall’Agata M , et al Prognostication in palliative radiotherapy-ProPaRT: Accuracy of prognostic scores . Front Oncol . 2022 ; 12 : 918414 . doi: 10.3389/fonc.2022.918414 36052228 PMC9425085

70 Maltoni M , Scarpi E , Pittureri C , et al Prospective comparison of prognostic scores in palliative care cancer populations . Oncologist . 2012 ; 17 ( 3 ): 446 - 454 . doi: 10.1634/theoncologist.2011-0397 22379068 PMC3316931

71 Matsunuma R , Tanbo Y , Asai N , et al Prognostic factors in patients with terminal stage lung cancer . J Palliat Med . 2014 ; 17 ( 2 ): 189 - 194 . doi: 10.1089/jpm.2013.0448 24438168

72 Mendis R , Soo WK , Zannino D , Michael N , Spruyt O. Multidisciplinary prognostication using the palliative prognostic score in an Australian Cancer Center . Palliat Care . 2015 ; 9 : 7 - 14 . doi: 10.4137/PCRT.S24411 26309410 PMC4524542

73 Miyagi T , Miyata S , Tagami K , et al Prognostic model for patients with advanced cancer using a combination of routine blood test values . Support Care Cancer . 2021 ; 29 ( 8 ): 4431 - 4437 . doi: 10.1007/s00520-020-05937-5 33443662

74 Miyashita N , Ohashi K , Fujita M , et al Prognostic factors in patients in the terminal phase of haematological malignancies who are receiving home medical care . Br J Haematol . 2023 ; 201 ( 2 ): 290 - 301 . doi: 10.1111/bjh.18623 36572123

75 Naylor C , Cerqueira L , Costa-Paiva LHS , Costa JV , Conde DM , Pinto-Neto AM. Survival of women with cancer in palliative care: Use of the palliative prognostic score in a population of Brazilian women . J Pain Symptom Manage . 2010 ; 39 ( 1 ): 69 - 75 . doi: 10.1016/j.jpainsymman.2009.05.020 20117695

76 Numico G , Occelli M , Russi EG , et al Survival prediction and frequency of anticancer treatment in cancer patients hospitalized due to acute conditions. Role of clinical parameters and PaP score . Support Care Cancer . 2011 ; 19 ( 11 ): 1823 - 1830 . doi: 10.1007/s00520-010-1024-5 21560032

77 Sonoda H , Yamaguchi T , Matsumoto M , Hisahara K. Validation of the palliative prognostic index and palliative prognostic score in a palliative care consultation team setting for patients with advanced cancers in an acute care hospital in Japan . Am J Hosp Palliat Med . 2014 ; 31 ( 7 ): 730 - 734 . doi: 10.1177/1049909113506034 24071626

78 Stiel S , Bertram L , Neuhaus S , et al Evaluation and comparison of two prognostic scores and the physicians’ estimate of survival in terminally ill patients . Support Care Cancer . 2010 ; 18 ( 1 ): 43 - 49 . doi: 10.1007/s00520-009-0628-0 19381693 PMC3085748

79 Stone P , Vickerstaff V , Kalpakidou A , et al Prognostic tools or clinical predictions: which are better in palliative care? PLoS One . 2021 ; 16 ( 4 ): e0249763 . doi: 10.1371/journal.pone.0249763 33909658 PMC8081205

80 Tarumi Y , Watanabe SM , Lau F , et al Evaluation of the Palliative Prognostic Score (PaP) and Routinely Collected Clinical Data in Prognostication of Survival for Patients Referred to a Palliative Care Consultation Service in an Acute Care Hospital . J Pain Symptom Manage . 2011 ; 42 ( 3 ): 419 - 431 . doi: 10.1016/j.jpainsymman.2010.12.013 21477974

81 Tassinari D , Montanari L , Maltoni M , et al The palliative prognostic score and survival in patients with advanced solid tumors receiving chemotherapy . Support Care Cancer off J Multinatl Assoc Support Care Cancer . 2008 ; 16 ( 4 ): 359 - 370 . doi: 10.1007/s00520-007-0302-3 17629751

82 Yoon SJ , Suh SY , Hui D , et al Accuracy of the palliative prognostic score with or without clinicians’ prediction of survival in patients with far advanced cancer . J Pain Symptom Manage . 2021 ; 61 ( 6 ): 1180 - 1187 . doi: 10.1016/j.jpainsymman.2020.10.019 33096217

83 Yoon SJ , Jung JG , Kim JS , Kim SS , Kim S. Comparison of Accuracy Among Prognostic Scores for Predicting Life Expectancy in Korean Patients With Cancer With Weeks of Survival . Am J Hosp Palliat Med . 2014 ; 31 ( 8 ): 845 - 852 . doi: 10.1177/1049909113503486 24031080

84 Gerber K , Tuer Z , Yates P. Who makes it out alive?—Predicting survival to discharge of hospital patients referred to residential aged care . Collegian . 2021 ; 28 ( 4 ): 431 - 437 . doi: 10.1016/j.colegn.2020.12.007

85 Glare P , Sinclair C , Downing M , Stone P , Maltoni M , Vigano A. Predicting survival in patients with advanced disease . Eur J Cancer . 2008 ; 44 ( 8 ): 1146 - 1156 . doi: 10.1016/j.ejca.2008.02.030 18394880

86 Flierman I , Nugteren IC , van Seben R , Buurman BM , Willems DL. How do hospital-based nurses and physicians identify the palliative phase in their patients and what difficulties exist? A qualitative interview study . BMC Palliat Care . 2019 ; 18 ( 1 ): 54 . doi: 10.1186/s12904-019-0439-0 31288816 PMC6617645

87 Pontin D , Jordan N. Issues in prognostication for hospital specialist palliative care doctors and nurses: a qualitative inquiry . Palliat Med . 2013 ; 27 ( 2 ): 165 - 171 . doi: 10.1177/0269216311432898 22190605

88 Seiler A , Blum D , Deuel JW , et al Delirium is associated with an increased morbidity and in-hospital mortality in cancer patients: results from a prospective cohort study . Palliat Support Care . 2021 ; 19 ( 3 ): 294 - 303 . doi: 10.1017/S147895152000139X 33431093

89 Bush SH , Lawlor PG , Ryan K , et al ; ESMO Guidelines Committee . Delirium in adult cancer patients: ESMO Clinical Practice Guidelines . Ann Oncol off J Eur Soc Med Oncol . 2018 ; 29 ( Suppl 4 ): iv143 - iv165 . doi: 10.1093/annonc/mdy147 29992308

90 Green JR , Smith J , Teale E , et al Use of the confusion assessment method in multicentre delirium trials: training and standardisation . BMC Geriatr . 2019 ; 19 ( 1 ): 107 . doi: 10.1186/s12877-019-1129-8 30991945 PMC6466721

91 Lewis ET , Harrison R , Nicholson M , et al Clinicians’ and public acceptability of universal risk-of-death screening for older people in routine clinical practice in Australia: Cross-sectional surveys . Aging Clin Exp Res . 2021 ; 33 ( 4 ): 1063 - 1070 . doi: 10.1007/s40520-020-01598-w 32458357

92 Malhotra C , Bundoc F , Chaudhry I , et al ; COMPASS study group . A prospective cohort study assessing aggressive interventions at the end-of-life among patients with solid metastatic cancer . BMC Palliat Care . 2022 ; 21 ( 1 ): 73 . doi: 10.1186/s12904-022-00970-z 35578270 PMC9109395

93 Cheung MC , Earle CC , Rangrej J , et al Impact of aggressive management and palliative care on cancer costs in the final month of life . Cancer . 2015 ; 121 ( 18 ): 3307 - 3315 . doi: 10.1002/cncr.29485 26031241 PMC4560956

94 Triplett DP , LeBrett WG , Bryant AK , et al Effect of palliative care on aggressiveness of end-of-life care among patients with advanced cancer . J Oncol Pract . 2017 ; 13 ( 9 ): e760 - e769 . doi: 10.1200/JOP.2017.020883 28829693 PMC5598315

95 McCormack R , Sui J , Conroy M , Stodart J. The usefulness of phlebotomy in the palliative care setting . J Palliat Med . 2011 ; 14 ( 3 ): 297 - 299 . doi: 10.1089/jpm.2010.0435 21265635

96 Zanartu C , Matti-Orozco B. Comfort measures only: agreeing on a common definition through a survey . Am J Hosp Palliat Med . 2013 ; 30 ( 1 ): 35 - 39 . doi: 10.1177/1049909112440740 22495795

97 Kawai N , Yuasa N. Laboratory prognostic score for predicting 30-day mortality in terminally ill cancer patients . Nagoya J Med Sci . 2018 ; 80 ( 4 ): 571 - 582 . doi: 10.18999/nagjms.80.4.571 30587871 PMC6295427

98 Yoong SQ , Porock D , Whitty D , Tam WWS , Zhang H. Performance of the Palliative Prognostic Index for cancer patients: a systematic review and meta-analysis . Palliat Med . 2023 ; 37 ( 8 ): 1144 - 1167 . doi: 10.1177/02692163231180657 37310019

99 Chen JJ , Gamble K , Graham-Wisener L , et al GP perceptions of the adequacy of community-based care for patients with advanced heart failure in a UK region (NI): a qualitative study . Open Heart . 2018 ; 5 ( 1 ): e000734 . doi: 10.1136/openhrt-2017-000734 29632677 PMC5888440

100 Graham C , Schonnop R , Killackey T , et al Exploring health care providers’ experiences of providing collaborative palliative care for patients with advanced heart failure at home: a qualitative study . J Am Heart Assoc . 2022 ; 11 ( 13 ): e024628 . doi: 10.1161/JAHA.121.024628 35730640 PMC9333360

101 Hallen SAM , Hootsmans NAM , Blaisdell L , Gutheil CM , Han PKJ. Physicians’ perceptions of the value of prognostic models: the benefits and risks of prognostic confidence . Health Expect Int J Public Particip Health Care Health Policy . 2015 ; 18 ( 6 ): 2266 - 2277 . doi: 10.1111/hex.12196 PMC5810722 24816136

102 Anderson F , Downing GM , Hill J , Casorso L , Lerch N. Palliative Performance Scale (PPS): a new tool . J Palliat Care . 1996 ; 12 ( 1 ): 5 - 11 . doi: 10.1177/082585979601200102 8857241

103 Cole A , Arthur A , Seymour J. Comparing the predictive ability of the Revised Minimum Dataset Mortality Risk Index (MMRI-R) with nurses’ predictions of mortality among frail older people: a cohort study . Age Ageing . 2019 ; 48 ( 3 ): 394 - 400 . doi: 10.1093/ageing/afz011 30806455

104 Wynants L , Calster BV , Collins GS , et al Prediction models for diagnosis and prognosis of covid-19: Systematic review and critical appraisal . BMJ . 2020 ; 369 : m1328 . doi: 10.1136/bmj.m1328 32265220 PMC7222643

105 Riley RD , Hayden JA , Steyerberg EW , et al ; PROGRESS Group . Prognosis Research Strategy (PROGRESS) 2: prognostic factor research . PLoS Med . 2013 ; 10 ( 2 ): e1001380 . doi: 10.1371/journal.pmed.1001380 23393429 PMC3564757