1. Introduction
The rising prevalence of chronic diseases and population aging has driven a paradigm shift in global health, highlighting the need for a comprehensive re-evaluation of healthcare models. Millions of individuals live with advanced chronic conditions, both oncological and non-oncological, which substantially affect multiple dimensions of health and require a coordinated, multidimensional response [ 1 ].
Palliative care (PC) has emerged as a key component of person-centered care, addressing the complex needs of patients with life-threatening illnesses [ 2 ]. It is a comprehensive care model designed to enhance quality of life through interdisciplinary teams that manage physical symptoms, pain, and psychosocial and spiritual concerns. Furthermore, PC emphasizes the integration of patient’s support network, including caregivers and family members, who also face the emotional and clinical challenges associated with caring for these patients.
Extensive evidence supports the beneficial impact of PC, particularly in patients with advanced oncological diseases [ 3 , 4 , 5 , 6 ]. It not only improves quality of life and physical and psychological well-being but also reduces anxiety, depression, and caregiver burden. Notably, early integration of outpatient PC has been associated with improved survival rates, emphasizing the importance of incorporating PC at the initial stages of treatment for patients with advanced cancer. Although much of the evidence has focused on oncological populations, emerging studies suggest that PC can significantly improve outcomes for patients with non-oncological conditions, highlighting the need for broader application [ 7 , 8 , 9 , 10 ].
The World Health Organization (WHO) estimates that approximately 56.8 million people require PC [ 11 ]. This need encompasses patients with advanced-stage chronic diseases, including cardiovascular diseases (38.5%), cancer (34%), chronic respiratory diseases (10.3%), human immunodeficiency virus (5.7%), and diabetes (4.6%). Additionally, other conditions such as renal failure, chronic liver disease and neurodegenerative diseases like multiple sclerosis, Parkinson’s disease and dementia are increasingly recognized as requiring specialized PC services.
Timely identification of patients who may benefit from PC remains a significant challenge. Factors contributing to this difficulty include prognostic uncertainty, variability in disease trajectories, and the absence of standardized referral criteria [ 12 ]. In response to these challenges, different screening tools have been developed over the past two decades to help clinicians in recognizing patients with PC needs. These tools include traditional instruments based on clinical indicators, expert consensus, and scoring systems, as well as more recent models developed using artificial intelligence (AI) techniques. However, the characteristics, performance, and applicability of these tools vary considerably, and there is currently no consensus on which instruments are most effective or feasible across different care settings. Additionally, many tools lack robust validation or have only been tested in specific populations, limiting their generalizability. Nevertheless, global access to PC remains highly heterogeneous, with the majority of resources concentrated in high-income countries. Barriers including limited training of healthcare professionals, inadequate infrastructure, and insufficient awareness further hinder the timely integration of PC. In this context, standardized and validated screening tools could play a crucial role in bridging these gaps, ensuring early identification and equitable access to PC across diverse healthcare systems.
This systematic review aims to identify and critically appraise all screening tools developed for the early identification of patients with PC needs, encompassing both conventional instruments and models based on AI. This study examines the development methodology, clinical applicability, validation status, and key features of these tools to provide a comprehensive overview of existing approaches and guide future clinical and research initiatives.
2. Materials and Methods
This systematic review was conducted following the PRISMA 2020 (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [ 13 ] and the COSMOS-E (Conducting Systematic Reviews and Meta-analyses of Observational Studies of Etiology) recommendations [ 14 ]. Ethical approval was not required for this study as no human material or identifiable data were involved, and no human interventions were performed. For the same reason, obtaining informed consent was not necessary. The review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) under the registration number CRD420251031483 (available from https://www.crd.york.ac.uk/PROSPERO/view/CRD420251031483 , accessed on 25 April 2025).
In line with the PRISMA framework, the research question was structured using the PICO format: P (Population): patients with advanced chronic illness; I (Intervention): screening tools; C (Comparator): not applicable; O (Outcomes): early identification of patients with potential PC needs.
2.1. Literature Search
We conducted a comprehensive literature search in PubMed, Web of Science, Cochrane Library, Scopus, EMBASE, and CINAHL to identify studies describing or evaluating tools designed for the early identification of patients with PC needs, including those developed using AI techniques. The search covered studies published from January 2000 to April 2025. The search was limited to studies published from 2000 onwards, as most tools for early identification of PC needs emerged after the development of modern PC frameworks and structured screening approaches at the turn of the century. The following search terms, including MeSH and free-text terms, were used: “Advanced Illness,” “Artificial Intelligence,” “Assessment Tools,” “External Validation,” “Identification,” “Machine Learning,” “Palliative Care Needs,” “Predictive Models,” and “Screening Tools”. The search strategy applied in each database combined these terms using Boolean operators (AND, OR). The search was limited to human studies, with no language restrictions. The full search strategy for PubMed is provided in the Supplementary Data .
2.2. Inclusion and Exclusion Criteria
We included studies that described or evaluated screening instruments developed to support the early identification of adult patients with PC needs. Eligible instruments targeted patients with advanced diseases, including cancer, heart disease, chronic respiratory disease, dementia, organ failure, frailty, and other severe conditions. We focused on tools designed to identify patients who may benefit from PC across diverse care settings (hospital, primary care, nursing home, or community), encompassing traditional clinical tools and those developed using AI techniques. Instruments designed explicitly for single conditions, such as cancer or heart failure, were also considered eligible. Given the rapid changes in patient conditions in these settings, we excluded screening instruments intended solely for use in intensive care units or emergency departments, as they limit their applicability in predicting longer-term PC needs. Additionally, we excluded tools primarily designed for use in the last days or weeks of life, as these are unsuitable for early identification of patients who may benefit from PC. Methodological exclusion criteria included review articles, duplicate publications, studies without usable data, small case series, and conference abstracts.
2.3. Study Selection
After removing duplicates, three reviewers (A.B.F., A.D.C.M., and G.G.M.) independently screened the titles and abstracts of all identified records to assess their eligibility based on the predefined inclusion and exclusion criteria. The full texts of potentially relevant articles were retrieved and evaluated in detail. Reviewer disagreements during the selection process were resolved through discussion and consensus. If consensus could not be reached, a fourth reviewer (A.F.M.) was consulted to make the final decision. To ensure consistency and reproducibility in the selection process, all reviewers used a standardized screening form to guide the application of the eligibility criteria.
2.4. Data Extraction
Data extraction was performed independently by three reviewers (A.B.F., A.D.C.M., and G.G.M.) using a standardized data collection form designed explicitly for this review. For each included study, the following information was extracted: first author, year of publication, country, study design, whether the tool was developed using traditional or AI-based techniques, study population, setting (hospital, primary care, nursing home, or community), characteristics of the screening tool, target disease or conditions, outcomes, and whether the tool had undergone external validation. We also collected information on whether the instrument incorporated multidimensional domains, such as psychological, social, or spiritual aspects. Performance metrics of the external validation studies were extracted when available, including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the receiver operating characteristic curve (AUC). When multiple versions of the same tool were available, data were extracted from the original version. The extracted data were synthesized to facilitate a qualitative comparison between the different tools. Any discrepancies in data extraction were resolved through discussion and consensus among the reviewers. When consensus could not be reached, disagreements were escalated to a fourth senior reviewer (A.F.M.), who acted as the final arbiter. Inter-rater agreement statistics were not calculated; however, methodological rigor was ensured through independent duplicate assessment, standardized data extraction forms, and structured consensus procedures, in accordance with PRISMA 2020 recommendations.
2.5. Outcomes
The objectives of this systematic review comprised identifying screening instruments designed to detect patients with early PC needs and assessing findings from external validation studies.
2.6. Risk of Bias Assessment
Four reviewers (A.B.F., A.D.C.M., A.F.M. and G.G.M.) independently assessed the methodological quality and risk of bias in the included studies. We assessed the methodological quality of the included studies using different tools tailored to each study design. For the development of instruments and questionnaires, we used the COSMIN checklist [ 15 ]. This tool evaluates content and structural validity, internal consistency, reliability, measurement error, responsiveness, and interpretability of the data. For observational studies, we applied the Newcastle–Ottawa Scale (NOS) [ 16 ]. We assessed the risk of bias in studies developing prediction models using the PROBAST tool [ 17 ]. This instrument evaluates four domains: participants, predictors, outcomes, and analysis. Each domain contains signaling questions that help identify potential sources of bias and concerns regarding applicability. This scale assesses three domains: participant selection, group comparability, and exposure or outcome ascertainment. For randomized clinical trials, we used the Cochrane Risk of Bias tool [ 18 ]. This instrument assesses bias across sequence generation, allocation concealment, blinding, incomplete outcome data, selective reporting, and other potential sources of bias. Each domain was categorized as having a low, high, or unclear risk of bias. Any discrepancies during the bias assessment process were resolved through discussion and consensus.
2.7. Data Synthesis and Analysis
The study selection process was systematically documented using a PRISMA 2020 flow diagram. Due to the heterogeneity in the development methodology, target populations, structure, and validation processes of the screening instruments, a narrative synthesis approach was employed. Data from the selected studies were systematically extracted and organized into structured tables to facilitate a comprehensive comparison. Screening tools were categorized into two main groups based on their development methodology: (1) conventional approaches and (2) AI-based techniques.
3. Results
3.1. Search Results
A total of 4820 articles were initially retrieved. After removing 2243 duplicates, 2577 records remained. Of these, 792 were screened by title and abstract, resulting in the exclusion of 562 records. Following title and abstract screening, 230 articles were selected for full-text review. Finally, 35 studies were included: 13 reported the development of screening tools [ 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 ] and 22 focused on the external validation of these instruments [ 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 ] ( Figure 1 ). Nine screening tools were developed using traditional methodologies, including the Gold Standards Framework Prognostic Indicator Guidance (GSF-PIG) [ 19 ], the Rainone [ 20 ], the Necesidades Paliativas CCOMS-ICO (NECPAL) [ 21 ], the RADboud indicators for PAlliative Care needs (RADPAC) [ 22 ], the Supportive and Palliative Care Indicators Tool (SPICT) [ 23 ], Anticipatory care in Primary care (AnticiPal) [ 24 ], Palliative Care Screening Tool (PCST) [ 25 ], the PROactive Palliative Care Identification Tool for patients with chronic obstructive pulmonary disease (COPD) (ProPal-COPD) [ 26 ], and PALLIA-10 [ 27 ], while four instruments were developed using AI techniques [ 28 , 29 , 30 , 31 ]. The Surprise Question (SQ) (“Would I be surprised if this patient died in the next 12 months?”) and the Double Surprise Question (DSQ) (when SQ is answered with ‘no’: “Would I be surprised if this patient will be still alive after 12 months?”) were identified as prognostic identification strategies and were analyzed separately but were not considered screening tools developed as standalone instruments. Among the instruments developed using traditional methods, seven were paper-based (GSF-PIG, NECPAL, RADPAC, SPICT, PCST, ProPal-COPD and PALLIA-10) and two were in electronic format (Rainone and AnticiPal).
3.2. Characteristics of the Screening Tools
The screening tools identified in this review differ in how they are intended for use in clinical practice. Some tools are designed for systematic screening, meaning that they are applied to all patients within a clearly defined population or clinical context. Other tools are not intended for universal screening, but rather to support clinician-led identification when clinical deterioration, complexity, or concern arises. Importantly, most tools do not provide explicit recommendations regarding when screening should be performed, how frequently it should be repeated, or which populations should be systematically assessed, which may hinder their consistent implementation across healthcare settings.
3.2.1. Screening Tools Developed Using Traditional Methods
Table 1 a shows the characteristics of the original studies that developed screening tools using traditional methods, while Table 1 b summarizes the main features of the screening tools for the early identification of patients with PC needs.
The GSF-PIG was developed in the United Kingdom based on expert consensus and clinical experience [ 19 ]. It is applicable across all care settings and incorporates the SQ with general indicators of decline and disease-specific criteria. A negative answer to the SQ defines a positive result combined with at least one general or disease-specific indicator, such as frequent hospital admissions in patients with heart failure or loss of swallowing function in dementia. The Rainone tool was developed in the United States [ 20 ]. It targets individuals aged 75 years or older with congestive heart failure (CHF), COPD, dementia, cancer, or human immunodeficiency virus (HIV). Development methodology was based on retrospective analysis of electronic health records (EHRs) to determine factors most frequently associated with 12-month mortality. A positive case is defined by a negative response to the SQ and/or affirmative answers to any of the five additional items. The NECPAL CCOMS-ICO tool was developed in Spain, adapting elements from the GSF-PIG and SPICT instruments [ 21 ]. NECPAL combines the SQ with indicators of severity and disease progression for conditions including cancer, CHF, dementia, and COPD. A positive case is defined by a negative response to the SQ, along with the presence of at least one indicator. The RADPAC tool was developed in the Netherlands through a Delphi consensus study with general practitioners [ 22 ]. It is specifically designed for primary care. The tool includes disease-specific indicators for advanced CHF, COPD, and cancer stages. Notably, RADPAC uniquely incorporates patient-reported indicators, such as the patient’s perception that the end of life is approaching or a diminished ‘drive to live’, reflecting a subjective dimension rarely captured by other tools. The SPICT was developed in the United Kingdom through a structured process that included a literature review, peer review, and a prospective case-finding study [ 23 ]. The tool does not include the SQ. It is a checklist that combines general indicators of deteriorating health, such as declining functional status or increasing hospital admissions, with disease-specific indicators for conditions including cancer, organ failure, and dementia. It has been translated and culturally adapted into multiple languages. The AnticiPal tool was developed in the United Kingdom based on the SPICT criteria [ 24 ]. Its creation followed an iterative process involving design, implementation, and testing phases, combining clinical expertise with data from EHRs. AnticiPal is an automated search algorithm that identifies patients with potential PC needs based on Read Codes related to malignancy and serious chronic illnesses. The PCST was developed in the United States in the medical intensive care unit (MICU) setting [ 25 ]. In its original form, the tool consisted of a brief checklist of clinical and psychosocial indicators, and any positive item triggered a palliative care consultation within 24 h. The ProPal-COPD tool was developed in the Netherlands to identify patients with advanced COPD who may benefit from proactive PC. It was designed for systematic application in patients hospitalized for acute exacerbations of COPD and uses 1-year mortality prediction as a proxy for PC needs. The tool combines clinical judgment through the SQ with disease-specific indicators, including dyspnea severity, health status, lung function, body mass index, prior hospitalizations, and the presence of relevant comorbidities [ 26 ]. Finally, the PALLIA-10 tool was developed in France using a methodology that involved expert consensus and clinical practice observation [ 27 ]. It was designed for application in hospital settings. It consists of a checklist of 10 simple clinical indicators, including advanced chronic illnesses, repeated hospitalizations, unintentional weight loss, cognitive decline, and psychosocial or ethical concerns. In PALLIA-10, a positive screen is generally defined by a total score ≥ 3; however, a higher threshold (≥5) has been suggested as more specific and appropriate for identifying true PC needs.
3.2.2. Screening Tools Developed Using Artificial Intelligence Techniques
Characteristics of the original studies describing the development of screening tools using AI methods are presented in Table 2 . Avati et al. developed a mortality prediction model using a deep neural network trained on structured EHR data from the STRIDE clinical data warehouse at Stanford University [ 28 ]. The model was designed to identify hospitalized patients with a predicted risk of death within 3 to 12 months. The dataset included over 2 million patients, and training was conducted using a supervised learning approach. Wang et al. designed a deep learning model using neural networks with attention mechanisms (stacked LSTM layers) to predict 6-month, 1-year, and 2-year mortality in patients with dementia [ 29 ]. The model was trained on longitudinal clinical notes and demographic data from the Partners HealthCare System in Boston. Cary et al. employed a retrospective cohort design using Medicare administrative and clinical data from patients admitted to inpatient rehabilitation facilities following hip fracture [ 30 ]. They trained and compared logistic regression and multilayer perceptron (MLP) models to predict 30-day and 1-year mortality. The development process included extensive feature selection based on sociodemographic, clinical, and functional variables. Finally, Zhang et al. created a generalized machine learning pipeline (GMLP) to predict high-mortality-risk patients using administrative claims data from a regional health plan [ 31 ]. The pipeline included five stages: cohort creation, feature engineering, predictive modeling, scoring, and model maintenance. Development began with the application of a structured data pull strategy and inclusion/exclusion criteria based on PC eligibility. Feature engineering incorporated diagnosis codes, comorbidity indices, healthcare utilization costs and patterns, and demographic variables. None underwent external validation in independent populations or prospective clinical validation.
3.3. Risk of Bias Assessment of Screening Tool Derivation Studies
The methodological quality assessment is summarized in Tables S1–S3 (Supplementary Data files) . For the checklist-based instruments (GSF-PIG, NECPAL, RADPAC, SPICT, AnticiPal, PCST, and PALLIA-10), content validity was adequate; however, other psychometric properties, such as reliability, internal consistency, and responsiveness, were not evaluated. The Rainone study received a score of 3 out of 9 stars on the NOS, and the ProPal-COPD tool achieved a total score of 8. The studies that developed ML-based prediction models have a high risk of bias in the analysis domain due to a lack of external validation and potential overfitting.
3.4. Performance of the Screening Tools in the External Validation Studies
Twenty one external validation studies evaluating four screening tools were identified ( Table 3 ). These studies focused on predicting mortality or deterioration in various clinical settings, including acute hospitals, tertiary care centers, primary care, and specialized wards. The GSF-PIG was externally validated in three studies. Haga et al. assessed ambulatory patients with CHF NYHA III–IV, finding the highest sensitivity (83%) but the lowest specificity (22%) [ 32 ]. O’Callaghan et al. conducted a study in an acute hospital setting in New Zealand, reporting lower sensitivity (62.6%) but higher specificity (91.9%) [ 33 ]. Another study evaluated acutely hospitalized medical patients in South Africa (22% HIV and high prevalence of CHF), reporting a sensitivity of 74% and specificity of 85% [ 34 ]. The NECPAL tool has been evaluated in diverse settings. Gómez-Batiste et al. conducted a study in primary care and acute hospitals in Spain, reporting high sensitivity for mortality within 12–24 months (91.3% and 87.5%, respectively) [ 35 ]. Troncoso et al. adapted and validated the tool in Chilean primary care using a psychometric approach, reporting an AUC of 0.808 for predicting a positive response to the SQ [ 36 ]. A Calsina-Berna et al. conducted a cross-sectional prevalence study in a tertiary hospital in Spain, focusing on the identification and clinical characterization of inpatients with PC needs, without reporting metrics [ 37 ]. Another Spanish study compared the accuracy of the NECPAL to the PROFUND instrument in 146 nursing home residents (46.6% with dementia) for predicting assessed survival at 3, 6, 12, and 24 months [ 38 ]. The PROFUND instrument showed better predictive capacity in the short term (3 months), particularly among residents with dementia, whereas NECPAL performed better in the long term (24 months), especially among residents without dementia. Fisher et al. performed a cultural and linguistic adaptation of NECPAL for Israel (I-NECPAL), providing psychometric metrics (sensitivity 0.93, specificity 0.17, PPV 0.53, and NPV 0.71) related to the SQ [ 39 ]. Finally, a prospective study conducted in Italy evaluated the prognostic role of the NECPAL in 103 hospitalized older adults admitted for acute medical and surgical conditions [ 40 ]. One-year mortality was used as the primary outcome. The predictive accuracy of NECPAL was moderate (AUC 0.75). After removing the “≥2 comorbidities” item, NECPAL’s performance improved (AUC 0.78). The Taiwanese version of PCST (TW-PCST) was evaluated in four studies. A prospective study showed a good discriminative ability for 14-, 90-, and 180-day mortality in a large cohort of inpatients in an acute hospital setting in Taiwan with consistently high NPV [ 41 ]. At 90 days, a total-ABCD cut-off score of ≥2 provided the optimal balance, yielding the highest sensitivity (84.0%) together with an acceptable specificity (74.7%) and a high NPV (98.8%). Yen et al. conducted an extensive interventional cohort study to evaluate the TW-PCST for early identification of PC needs in hospitalized patients in Taipei [ 42 ]. A score ≥ 4 was independently associated with higher 90-day mortality (adjusted Odds Ratio 6.86). Yen et al. conducted another study for comparing the prognostic accuracy of the SQ and the TW-PCST for predicting 12-month mortality [ 43 ]. TW-PCST performed slightly better than SQ, as it showed higher specificity (92.0% vs. 90.6%) and AUC (0.689 vs. 0.680). In a third study conducted by Yen et al., the TW-PCST tool showed a high NPV for 6-month mortality when the score was <4 [ 44 ]. We identified seven studies that reported the clinical performance of SPICT. Among studies, one was a randomized controlled trial, and six were observational studies. In a retrospective cohort study for predicting one-year mortality, this tool showed a sensitivity of 84.1% and a specificity of 57.9% [ 45 ]. Notably, the authors analyzed the predictive performance of the two domains of the instrument separately: the general indicators of deteriorating health yielded an AUC of 0.758, and the clinical disease-specific indicators had an AUC of 0.748, with no significant difference between them ( p = 0.638). Mitchell et al. conducted a randomized controlled trial in Australian general practice to compare the use of the SQ followed by the SPICT in SQ-positive cases with unguided clinical intuition among general practitioners in predicting 12-month mortality [ 46 ]. The combined SPICT + SQ approach yielded a sensitivity of 34.0% (95% CI 25.3–42.8) and a specificity of 95.8% (95% CI 93.0–98.6), with a PPV of 20.5% and an NPV of 97.9%. Although tool-based screening increased the identification of patients at risk compared with intuition alone, its relatively low sensitivity limited its ability to capture all patients who died during follow-up. In a prospective cohort study, 3640 patients aged 75 years or older were followed for 12 months to evaluate the performance of SQ versus SPICT, as applied through an automated electronic health record search [ 47 ]. The SQ identified 50% of deaths with a specificity of 99%, while the SPICT identified 58% with a specificity of 98%. When excluding 10 deaths classified as sudden, sensitivity increased to 69% for the SQ and 81% for the SPICT. A prospective multicenter study evaluated the prognostic value of the SPICT among 458 patients admitted to acute geriatric and cardiology units [ 48 ]. In acute geriatric units, the sensitivity reached 0.82 with an AUC of 0.822, and in cardiology units, the sensitivity was 0.69 with an AUC of 0.651. SPICT was adapted and psychometrically validated in a primary care setting in Chile, focusing on identifying patients with potential palliative care needs based on clinical judgment [ 49 ]. Performance assessment was based on internal consistency (Cronbach’s alpha = 0.86) and predictive validity, as indicated by Odds Ratios for specific clinical indicators, ranging from 30.81 to 71.11. A multicenter study evaluated the predictive performance of the SPICT tool in older adults receiving home-based medical care [ 50 ]. A total of 129 patients were prospectively assessed, and 6-month mortality was used as the outcome. The SPICT demonstrated acceptable discriminative ability, with a sensitivity of 0.80, specificity of 0.56, PPV of 0.57, NPV of 0.79, and an AUC of 0.78 when using the cut-off of four general indicators plus one clinical indicator. One study culturally adapted and validated the SPICT for use in mainland China [ 51 ]. The Mandarin version (SPICT-CH) was developed and tested in 212 hospitalized cancer patients. The psychometric evaluation demonstrated an acceptable content validity (Cronbach’s α 0.76). Another cross-sectional validation study also aimed to translate and culturally adapt the SPICT tool into Chinese and assess its preliminary performance in 388 hospitalized cancer patients [ 52 ]. Its screening performance was excellent, with an accuracy of 0.905. The ProPal-COPD tool underwent external validation in a multicenter prospective cohort study. The study included 523 patients hospitalized for an acute exacerbation of COPD across 10 hospitals, with 1-year all-cause mortality as the outcome measure [ 53 ]. During follow-up, 100 patients (19.1%) died within 12 months. The tool demonstrated moderate discriminative ability, with an AUC of 0.68.
Figure 2 summarizes the distribution of reported AUC values across external validation studies of SPICT, NECPAL, and TW-PCST.
3.5. Risk of Bias Assessment of External Validation Studies
The risk of bias assessment of the external validation studies is shown in Tables S4 and S5 (Supplementary Data files) . The studies showed heterogeneous quality. A considerable proportion of the observational validations of GSF-PIG and NECPAL were rated as having a high risk of bias, primarily due to limitations in participant selection (restricted or non-representative cohorts) and the use of non-objective outcomes (e.g., reliance on the SQ or clinical judgment). By contrast, the large Taiwanese cohorts validating the TW-PCST were consistently rated as having a low risk of bias, with a robust methodology. Similarly, the majority of SPICT validations were considered low-risk. The external validation of the ProPal-COPD tool had a high risk of bias, mainly due to methodological limitations in the analysis domain.
4. Discussion
To our knowledge, this is the first systematic review focusing on screening instruments for the early identification of patients with PC needs, including those that incorporate AI-based approaches.
This systematic review identified 12 screening tools for the early identification of PC needs. Eight tools were created using traditional methodologies, and four instruments employed artificial intelligence techniques. Overall, the methodological quality of these instruments was limited. Most of the identified tools use the prediction of death as a trigger for identifying people who are likely to have unmet PC needs. Patients with advanced progressive illness experience different trajectories of decline and typically have distinct needs across different stages of the disease. Therefore, the identification process should not be based solely on predicting mortality. It should also focus on anticipating their needs whenever they occur and predicting the rate and course of functional decline.
Importantly, only NECPAL and PCST (including its Taiwanese version) assessed psychological and spiritual care domains, but their approaches differ substantially. NECPAL integrates them as a structured domain with several dedicated items, whereas the PCST acknowledges such issues only within the “other criteria” category. These represent a small proportion of the total score and do not constitute a structured domain, reflecting the tool’s primarily biomedical and functional focus. PC is based on a holistic approach that addresses the patient’s physical, psychological, social, and spiritual needs. According to the WHO definition of PC, which encompasses physical, psychological, social, and spiritual dimensions, serious illness impacts multiple interrelated domains, resulting in complex patient needs. Thus, screening instruments should comprehensively assess these domains to effectively identify unmet needs. Psychological and spiritual distress are prevalent in patients with advanced diseases and can exacerbate physical suffering, while psychological and social needs become increasingly prominent as the end of life approaches [ 54 , 55 , 56 ].
Several previous reviews have summarized tools for the early identification of palliative care needs [ 57 , 58 , 59 ]. However, the present review extends prior work by incorporating recently developed tools, including disease-specific instruments and AI-based models, and by systematically evaluating predictive performance and risk of bias using contemporary methodological frameworks. In addition, this review provides a comparative perspective between traditional screening approaches and emerging data-driven models, highlighting shared limitations and future research directions. A relevant but largely unexplored limitation identified in this review is the underutilization of rich longitudinal clinical data beyond diagnostic and mortality outcomes. In particular, pharmacy records represent a valuable yet insufficiently exploited data source for early identification of PC needs. Changes in medication patterns—such as the initiation or escalation of potent opioids, repeated adjustments in diuretic therapy in advanced heart failure, or increasing polypharmacy—may act as sensitive proxies for symptom burden, functional decline, and escalating care complexity. Future screening strategies, including AI-based models and refined traditional tools, should incorporate pharmacometrics trajectories as indirect signals to support earlier, more needs-oriented identification.
All AI-based models identified in this review were designed to predict mortality rather than multidimensional PC needs. While mortality remains a commonly used proxy, this approach risks overlooking psychological, social, and spiritual dimensions that are central to PC. Future AI-based approaches may offer opportunities to integrate multidimensional data sources, such as patient-reported outcomes, functional status, and social vulnerability, moving beyond purely prognostic models towards tools better aligned with the holistic goals of PC.
A notable gap identified in this review is the scarcity of external validation studies for the screening tools. Despite their intended clinical utility, most instruments lack robust evidence supporting their accuracy and reliability across diverse settings and populations.
In the context of external validation, TW-PCST demonstrated the most consistent performance across multiple time points in the hospital setting. Unlike NECPAL, which exhibited high sensitivity but low specificity, TW-PCST maintained a more balanced profile, minimizing false positives. In contrast, GSF-PIG showed greater variability in its performance across different patient populations. Tools that include the SQ, such as NECPAL and GSF-PIG, may introduce subjectivity, potentially increasing the rate of false positives. This subjectivity can be particularly pronounced in non-cancer populations, where disease trajectories are less predictable and clinical judgment may be less reliable [ 60 , 61 ]. In contrast, tools that do not rely on subjective prognostication, like TW-PCST, may offer more objective results. Beyond identification accuracy, the clinical value of screening tools lies in their capacity to trigger timely and meaningful clinical action. A positive screening result should not be viewed as an endpoint, but rather as a catalyst for predefined interdisciplinary interventions. For example, the output of tools such as SPICT or an automated AI-based alert could be embedded into clinical workflows to prompt predefined interdisciplinary responses, such as a pharmacist-led medication review in the presence of escalating symptom burden, followed by a structured, team-based goals-of-care conversation when unmet needs or prognostic uncertainty are identified. Explicitly linking screening outputs to actionable care pathways may enhance the translation of early identification into tangible palliative care interventions and improve real-world clinical impact.
Integrating AI into PC is an opportunity to overcome the limitations of traditional models. Embedding AI-based screening tools within electronic health records may enhance their practical value by enabling automated, continuous identification of at-risk patients, potentially reducing clinician workload and variability associated with manual screening. In contrast, traditional screening methods require active clinician input and time for application, which may hinder scalability and consistent use in routine practice. Hybrid models combining automated AI-based screening with clinician assessment may therefore represent a promising strategy to improve efficiency while preserving clinical judgment. AI approaches can leverage extensive datasets to identify complex patterns that extend beyond clinical indicators, thereby anticipating decline trajectories. However, despite their potential, the studies included in our review reveal substantial limitations. Notably, all four AI-based studies lacked external validation. Without robust external testing, the generalizability of these models remains speculative. Furthermore, while AI models can theoretically capture multidimensional needs, including psychosocial and spiritual distress, current implementations focus predominantly on mortality prediction. Therefore, the added benefit of AI remains challenging to assess in this clinical setting.
Building on these findings, an ideal future screening tool should integrate multiple complementary dimensions. Such a framework would combine the clinical intuition and needs-oriented perspective embedded in tools like NECPAL with the predictive capacity of AI-based models, while systematically incorporating structured assessments of psychological, social, and spiritual domains. This integrated approach would allow screening tools to move beyond single-domain identification and better reflect the complexity of palliative care needs in routine clinical practice.
Why is it so complex to implement early identification processes for patients with PC needs? In light of our systematic review, one of the primary reasons is the persistence of a biomedical model that prioritizes mortality and prognosis in decision-making. We believe that incorporating the biopsychosocial model—which considers suffering as well as psychological, social, and spiritual dimensions—and advancing towards hybrid models that integrate both approaches would significantly enhance anticipation and early detection. Additional challenges further hinder this process, including insufficient professional training and limited knowledge of the biopsychosocial model, the lack of time in daily clinical practice, the absence of universal protocols, and, finally, the limitations imposed by the terminology currently used. In this regard, we propose to replace the term “palliative care” with “integrated early palliative approach” Although efforts have been made in recent years to broaden the acceptance of the term “palliative care” to promote its earlier and more widespread use, the expected impact in clinical practice has not been achieved. Therefore, a terminological shift is suggested. Nevertheless, caution is warranted. These patients should remain within the palliative perspective and approach to avoid creating parallel categories that, rather than integrating, might fragment care. These measures would help ensure greater equity in the inclusion of patients in palliative care.
Our study has some limitations. First, the systematic review included only a few high-quality studies. Therefore, the results should be interpreted with caution. Meta-analysis was not feasible due to substantial heterogeneity across studies, including differences in target populations, outcomes, and prediction horizons, and variability in cut-off thresholds used across validation studies. Second, we conducted a comprehensive literature search to capture all the relevant papers. Still, given the nature of this topic, some papers may have been missed. Third, there is currently no consensus on how to measure the accuracy of a screening tool for the early identification of patients with PC needs. In this context, specialist PC assessment may be a more appropriate reference standard for identifying PC needs than mortality outcomes, which are often used as surrogates but do not fully reflect the clinical complexity underlying PC identification.
5. Conclusions
This systematic review identified 13 studies on screening tools for detecting advanced illness and potential PC needs. Their evaluation is limited by the lack of a standardized comparator, which constrains conclusions about real-world utility. Most tools rely on mortality prediction, despite the multidimensional nature of palliative needs. Overall accuracy is moderate, reinforcing the need for clinical judgment. The findings underscore the urgency of standardized, multidimensional screening approaches linked to actionable care pathways. Future research should prioritize the development and external validation of tools that integrate clinical, functional, psychological, social, and spiritual domains into routine care. An integrated early palliative approach may enable earlier identification, timely interdisciplinary intervention, and more equitable access across disease trajectories.
Acknowledgments
Research in Palliative Care HUIL-Group: Soledad Acedo-Gutiérrez, Teresa Blanco-Moya, Natalia Pavón-Muñoz, Elena Díaz-Sánchez, Marina Chivato-Cabanal, Alicia Sánchez-Pizarro, David Escudero-Rodríguez, Beatriz Fernández-Rodríguez, Eva Moya-Mateo, Teresa Bellver-Álvarez, Marta Bustamante-Vega, Mercedes Duffort-Falcó, Virginia Pardo-Guimerá, Roberto Pedrero-Tomé.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Supplementary Materials
The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/jcm15030919/s1 : Table S1: Quality of the studies describing the development of screening tools according to the COSMIN criteria; Table S2: Quality of the studies describing the development of screening tools using the Newcastle–Ottawa Scale; Table S3: Quality of the studies describing the development of screening tools according to the PROBAST instrument; Table S4: Risk of bias in external validation studies according to PROBAST criteria; Table S5: Risk of bias in external validation studies according to the Cochrane Risk of Bias tool; Table S6: PRISMA Checklist.
Author Contributions
Conceptualization: A.F.-M., A.B.-F. and G.G.M.; methodology, A.F.-M., A.B.-F. and G.G.M.; validation, A.B.-F., A.D.C.-M., G.G.M. and A.F.-M.; formal analysis, A.B.-F. and A.F.-M.; investigation, A.B.-F., A.D.C.-M., L.H., F.G.-G., H.N.-L., R.L.-G., J.T.-M. and Research in Palliative Care HUIL-Group; data curation, A.B.-F., A.D.C.-M. and G.G.M.; writing—original draft preparation, A.B.-F. and A.F.-M.; writing—review and editing, A.B.-F., A.D.C.-M., L.H., F.G.-G., H.N.-L., R.L.-G., J.T.-M., G.G.M. and A.F.-M.; supervision, A.F.-M. and A.B.-F.; project administration, A.F.-M. All authors have read and agreed to the published version of the manuscript.
Institutional Review Board Statement
Not applicable.
Informed Consent Statement
Not applicable.
Data Availability Statement
The data supporting this study’s findings are available from the corresponding author upon reasonable request. Restrictions apply to the availability of these data, which were used under license for this systematic review.
Conflicts of Interest
The authors declare no conflicts of interest.
Abbreviations
The following abbreviations are used in this manuscript: AI Artificial Intelligence AUC Area Under the Receiver Operating Characteristic Curve CHF Congestive Heart Failure CI Confidence Interval COPD Chronic Obstructive Pulmonary Disease COSMIN COnsensus-based Standards for the selection of health Measurement INstruments COSMOS-E Conducting Systematic Reviews and Meta-analyses of Observational Studies of Etiology DSQ Double Surprise Question EHR Electronic Health Record GSF-PIG Gold Standards Framework Prognostic Indicator Guidance HIV Human Immunodeficiency Virus ML Machine Learning NA Not Applicable NOS Newcastle–Ottawa Scale NPV Negative Predictive Value NR Not Reported NYHA New York Heart Association OR Odds Ratio PC Palliative Care PCST Palliative Care Screening Tool PPV Positive Predictive Value ProPal-COPD PROactive Palliative Care Identification Tool for patients with Chronic Obstructive Pulmonary Disease RADPAC RADboud indicators for PAlliative Care needs SPICT Supportive and Palliative Care Indicators Tool SPICT-CH Chinese version of the Supportive and Palliative Care Indicators Tool SQ Surprise Question TW-PCST Taiwanese version of the Palliative Care Screening Tool WHO World Health Organization
References
1. World Health Organization Innovative Care for Chronic Conditions: Building Blocks for Action WHO Geneva, Switzerland 2002 Available online: https://apps.who.int/iris/bitstream/handle/10665/42500/WHO_NMC_CCH_02.01.pdf (accessed on 5 September 2025)
2. World Health Organization Why Palliative Care Is an Essential Function of Primary Health Care WHO Geneva, Switzerland 2018 (WHO/HIS/SDS/2018.39) Available online: https://www.who.int/publications/i/item/WHO-HIS-SDS-2018.39 (accessed on 5 September 2025)
3. Hoerger M. Wayser G.R. Schwing G. Suzuki A. Perry L.M. Impact of Interdisciplinary Outpatient Specialty Palliative Care on Survival and Quality of Life in Adults with Advanced Cancer: A Meta-Analysis of Randomized Controlled Trials Ann. Behav. Med. 2019 53 674 685 10.1093/abm/kay077 30265282 PMC6546936
4. Haun M.W. Estel S. Rücker G. Friederich H.C. Villalobos M. Thomas M. Hartmann M. Early palliative care for adults with advanced cancer Cochrane Database Syst. Rev. 2017 6 CD011129 10.1002/14651858.CD011129.pub2 28603881 PMC6481832
5. Fulton J.J. LeBlanc T.W. Cutson T.M. Porter Starr K.N. Kamal A. Ramos K. Freiermuth C.E. McDuffie J.R. Kosinski A. Adam S. Integrated outpatient palliative care for patients with advanced cancer: A systematic review and meta-analysis Palliat. Med. 2019 33 123 134 10.1177/0269216318812633 30488781 PMC7069657
6. Hui D. Kim S.H. Roquemore J. Dev R. Chisholm G. Bruera E. Impact of timing and setting of palliative care referral on quality of end-of-life care in cancer patients Cancer 2014 120 1743 1749 10.1002/cncr.28628 24967463 PMC4073257
7. Kuebler K. Monroe T. Ricciardi R. Westlake C. Sorenson M. Pederson J. Neft M. Mitchell S.A. Daniel K. De Santo-Madeya S. Integration of palliative care in the management of multiple chronic conditions: An expert consensus paper with policy implications Nurs. Outlook 2024 72 102273 10.1016/j.outlook.2024.102273 39388799 PMC11611613
8. Quinn K.L. Stukel T. Stall N.M. Huang A. Isenberg S. Tanuseputro P. Goldman R. Cram P. Kavalieratos D. Detsky A.S. Association between palliative care and healthcare outcomes among adults with terminal non-cancer illness: Population based matched cohort study BMJ 2020 370 m2257 10.1136/bmj.m2257 32631907 PMC7336238
9. Mós J.R. Reis-Pina P. Early Integration of Palliative Care in Nononcological Patients: A Systematic Review J. Pain Symptom Manag. 2025 69 e283 e302 10.1016/j.jpainsymman.2024.12.023 39778632
10. Quinn K.L. Shurrab M. Gitau K. Kavalieratos D. Isenberg S.R. Stall N.M. Stukel T.A. Goldman R. Horn D. Cram P. Association of Receipt of Palliative Care Interventions with Health Care Use, Quality of Life, and Symptom Burden Among Adults with Chronic Noncancer Illness: A Systematic Review and Meta-analysis JAMA 2020 324 1439 1450 10.1001/jama.2020.14205 33048152 PMC8094426
11. World Health Organization Palliative Care [Internet] WHO Geneva, Switzerland 2024 Available online: https://www.who.int/news-room/fact-sheets/detail/palliative-care (accessed on 5 September 2025)
12. Murray S.A. Kendall M. Boyd K. Sheikh A. Illness trajectories and palliative care BMJ 2005 330 1007 1011 10.1136/bmj.330.7498.1007 15860828 PMC557152
13. Page M.J. McKenzie J.E. Bossuyt P.M. Boutron I. Hoffmann T.C. Mulrow C.D. Shamseer L. Tetzlaff J.M. Akl E.A. Brennan S.E. The PRISMA 2020 statement: An updated guideline for reporting systematic reviews BMJ 2021 372 n71 10.1136/bmj.n71 33782057 PMC8005924
14. Dekkers O.M. Vandenbroucke J.P. Cevallos M. Renehan A.G. Altman D.G. Egger M. COSMOS-E: Guidance on conducting systematic reviews and meta-analyses of observational studies of etiology PLoS Med. 2019 16 e1002742 10.1371/journal.pmed.1002742 30789892 PMC6383865
15. Mokkink L.B. Terwee C.B. Patrick D.L. Alonso J. Stratford P.W. Knol D.L. Bouter L.M. De Vet H.C.W. The COSMIN checklist for assessing the methodological quality of studies on measurement properties of health status measurement instruments: An international Delphi study Qual. Life Res. 2010 19 539 549 10.1007/s11136-010-9606-8 20169472 PMC2852520
16. Wells G.A. Shea B. O’Connell D. Peterson J. Welch V. Losos M. Tugwell P. The Newcastle-Ottawa Scale (NOS) for Assessing the Quality of Nonrandomized Studies in Meta-Analyses Available online: http://www.ohri.ca/programs/clinical_epidemiology/oxford.htm (accessed on 10 October 2025)
17. Wolff R.F. Moons K.G.M. Riley R.D. Whiting P.F. Westwood M. Collins G.S. Reitsma J.B. Kleijnen J. Mallett S. PROBAST: A Tool to Assess the Risk of Bias and Applicability of Prediction Model Studies Ann. Intern. Med. 2019 170 51 58 10.7326/M18-1376 30596875
18. Higgins J.P.T. Altman D.G. Gøtzsche P.C. Jüni P. Moher D. Oxman A.D. Savović J. Schulz K.F. Weeks L. Sterne J.A.C. Cochrane Bias Methods Group; Cochrane Statistical Methods Group. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials BMJ 2011 343 d5928 10.1136/bmj.d5928 22008217 PMC3196245
19. Thomas K. Prognostic Indicator Guidance The Gold Standards Framework Centre London, UK 2004 Available online: https://www.futureplanning.org.uk/uploads/8/0/4/0/80407130/__gsf_prognostic_indicator_guidance.pdf?utm_source=chatgpt.com (accessed on 10 October 2025)
20. Rainone F. Blank A. Selwyn P.A. The early identification of palliative care patients: Preliminary processes and estimates from urban, family medicine practices Am. J. Hosp. Palliat. Med. 2007 24 137 140 10.1177/1049909106296973 17502439
21. Gómez-Batiste X. Martínez-Muñoz M. Blay C. Amblàs J. Vila L. Costa X. Villanueva A. Espaulella J. Espinosa J. Figuerola M. Identifying patients with chronic conditions in need of palliative care in the general population: Development of the NECPAL tool and preliminary prevalence rates in Catalonia BMJ Support. Palliat. Care 2013 3 300 308 10.1136/bmjspcare-2012-000211 24644748
22. Thoonsen B. Engels Y. van Rijswijk E. Verhagen S. van Weel C. Groot M. Vissers K. Early identification of palliative care patients in general practice: Development of RADboud indicators for PAlliative Care Needs (RADPAC) Br. J. Gen. Pract. 2012 62 e625 e631 10.3399/bjgp12X654597 22947583 PMC3426601
23. Highet G. Crawford D. Murray S.A. Boyd K. Development and evaluation of the Supportive and Palliative Care Indicators Tool (SPICT): A mixed-methods study BMJ Support. Palliat. Care 2014 4 285 290 10.1136/bmjspcare-2013-000488 24644193
24. Mason B. Boyd K. Murray S.A. Steyn J. Cormie P. Kendall M. Munday D. Weller D. Fife S. Murchie P. Developing a computerised search to help UK General Practices identify more patients for palliative care planning: A feasibility study BMC Fam. Pract. 2015 16 99 10.1186/s12875-015-0312-z 26253101 PMC4529731
25. Hurst E. Yessayan L. Mendez M. Hammad A. Jennings J. Preliminary Analysis of a Modified Screening Tool to Increase the Frequency of Palliative Care Consults Am. J. Hosp. Palliat. Med. 2018 35 417 422 10.1177/1049909117712229 28571498
26. Duenk R. Verhagen S. Bronkhorst E. Djamin R. Bosman G. Lammers E. Dekhuijzen P. Vissers K. Engels Y. Heijdra Y. Development of the ProPal-COPD tool to identify patients with COPD for proactive palliative care Int. J. Chronic Obstr. Pulm. Dis. 2017 12 2121 2128 10.2147/COPD.S140037 28790815 PMC5530053
27. Molin Y. Gallay C. Gautier J. Lardy-Cleaud A. Mayet R. Grach M. Guesdon G. Capodano G. Dubroeucq O. Bouleuc C. PALLIA-10, a screening tool to identify patients needing palliative care referral in comprehensive cancer centers: A prospective multicentric study (PREPA-10) Cancer Med. 2019 8 2950 2961 10.1002/cam4.2118 31055887 PMC6558580
28. Avati A. Jung K. Harman S. Downing L. Ng A. Shah N.H. Improving palliative care with deep learning BMC Med. Inform. Decis. Mak. 2018 18 122 10.1186/s12911-018-0677-8 30537977 PMC6290509
29. Wang L. Sha L. Lakin J.R. Bynum J. Bates D.W. Hong P. Zhou L. Development and Validation of a Deep Learning Algorithm for Mortality Prediction in Selecting Patients with Dementia for Earlier Palliative Care Interventions JAMA Netw. Open 2019 2 e196972 10.1001/jamanetworkopen.2019.6972 31298717 PMC6628612
30. Cary M.P. Jr. Zhuang F. Draelos R.L. Pan W. Amarasekara S. Douthit B.J. Kang Y. Colón-Emeric C.S. Machine Learning Algorithms to Predict Mortality and Allocate Palliative Care for Older Patients with Hip Fracture J. Am. Med. Dir. Assoc. 2021 22 291 296 10.1016/j.jamda.2020.09.025 33132014 PMC7867606
31. Zhang H. Li Y. McConnell W. Predicting potential palliative care beneficiaries for health plans: A generalized machine learning pipeline J. Biomed. Inform. 2021 123 103922 10.1016/j.jbi.2021.103922 34607012
32. Haga K. Murray S. Reid J. Ness A. O’Donnell M. Yellowlees D. Denvir M.A. Identifying community based chronic heart failure patients in the last year of life: A comparison of the Gold Standards Framework Prognostic Indicator Guide and the Seattle Heart Failure Model Heart 2012 98 579 583 10.1136/heartjnl-2011-301021 22422744
33. O’Callaghan A. Laking G. Frey R. Robinson J. Gott M. Can we predict which hospitalised patients are in their last year of life? A prospective cross-sectional study of the Gold Standards Framework Prognostic Indicator Guidance as a screening tool in the acute hospital setting Palliat. Med. 2014 28 1046 1052 10.1177/0269216314536089 24854032
34. Raubenheimer P.J. Day C. Abdullah F. Manning K. Cupido C. Peter J. The utility of a shortened palliative care screening tool to predict death within 12 months—A prospective observational study in two South African hospitals with a high HIV burden BMC Palliat. Care 2019 18 101 10.1186/s12904-019-0487-5 31722691 PMC6854790
35. Gómez-Batiste X. Martínez-Muñoz M. Blay C. Amblàs J. Vila L. Costa X. Espaulella J. Villanueva A. Oller R. Martori J.C. Utility of the NECPAL CCOMS-ICO© tool and the Surprise Question as screening tools for early palliative care and to predict mortality in patients with advanced chronic conditions: A cohort study Palliat. Med. 2017 31 754 763 10.1177/0269216316676647 27815556
36. Troncoso J. Morales-Meyer T. Villarroel L. Pérez-Cruz P. Rojas-Sepúlveda J. Arancibia-García M. Díaz-Pérez A. Moya-Castillo T. Cárdenas-Salazar J. González-González M. Adaptation and validation in Chile of the patient identification instrument needing palliative care: NECPAL-CCOMS-ICO 3.1© Aten. Primaria 2021 53 101994 (In Spanish) 10.1016/j.aprim.2021.101994 33740612 PMC7985705
37. Calsina-Berna A. Amblàs-Novellas J. González-Barboteo J. Pujol-Ribera E. Pastrana T. Gómez-Batiste X. Murray S.A. Blay C. Costa X. Vila L. Prevalence and clinical characteristics of patients with Advanced Chronic Illness and Palliative Care needs, identified with the NECPAL CCOMS-ICO© Tool at a Tertiary Care Hospital BMC Palliat. Care 2022 21 210 10.1186/s12904-022-01101-4 36443761 PMC9703744
38. Esteban-Burgos A.A. Hueso-Montoro C. Mota-Romero E. Romero-Béjar J.L. García-Caro M.P. Rodríguez-Torres M. Sánchez-Sánchez M.J. Bonilla-Vargas C. Pérez-Cruz P. Fernández-Sola C. The prognostic value of the NECPAL instrument, Palliative Prognostic Index, and PROFUND index in elderly residents of nursing homes with advanced chronic condition BMC Geriatr. 2023 23 715 10.1186/s12877-023-04409-9 37924015 PMC10623722
39. Fisher G. Shadmi E. Porat-Packer T. Zisberg A. Identifying patients in need of palliative care: Adaptation of the Necesidades Paliativas CCOMS-ICO© (NECPAL) screening tool for use in Israel Palliat. Support. Care 2024 22 103 109 10.1017/S1478951522001390 36285527
40. Spannella F. Falzetti S. Giulietti F. Sarnari S. Morichi V. Tamburrini P. Gattafoni P. Mannello L. Crippa M. Ferrara L. Prognostic Role of NECPAL CCOMS-ICO Tool on One-Year Mortality in a Hospitalized Older Population J. Palliat. Med. 2024 27 367 375 10.1089/jpm.2023.0038 37971772
41. Wang S.S. Huang C.M. Feng R.C. Wu Y.L. Huang S.J. Validation of a concise screening tool for the identification of palliative care needs among inpatients: A prospective study in hospital setting J. Formos. Med. Assoc. 2019 118 883 890 10.1016/j.jfma.2018.10.004 30502101
42. Yen Y.F. Lee Y.L. Hu H.Y. Lai Y.J. Sun W.J. Ko M.C. Chen C.C. Curtis J.R. Huang S.J. Chu D. An Interventional Study for the Early Identification of Patients with Palliative Care Needs and the Promotion of Advance Care Planning and Advance Directives J. Pain Symptom Manag. 2020 59 974 982.e3 10.1016/j.jpainsymman.2019.10.006 31759033
43. Yen Y.F. Lee Y.L. Hu H.Y. Sun W.J. Ko M.C. Chen C.C. Wong W.K. Morisky D.E. Huang S.J. Chu D. Early palliative care: The surprise question and the palliative care screening tool-better together BMJ Support. Palliat. Care 2022 12 211 217 10.1136/bmjspcare-2019-002116 32451326
44. Yen Y.F. Hu H.Y. Lai Y.J. Chou Y.C. Chen C.C. Ho C.Y. Comparison of intuitive assessment and palliative care screening tool in the early identification of patients needing palliative care Sci. Rep. 2022 12 4955 10.1038/s41598-022-08886-7 35322098 PMC8943025
45. De Bock R. Van Den Noortgate N. Piers R. Validation of the Supportive and Palliative Care Indicators Tool in a Geriatric Population J. Palliat. Med. 2018 21 220 224 10.1089/jpm.2017.0205 28792787
46. Mitchell G.K. Senior H.E. Rhee J.J. Ware R.S. Young S. Teo P.C. Murray S. Boyd K. Clayton J.M. Using intuition or a formal palliative care needs assessment screening process in general practice to predict death within 12 months: A randomised controlled trial Palliat. Med. 2018 32 384 394 10.1177/0269216317698621 28452570
47. van Wijmen M.P.S. Schweitzer B.P.M. Pasman H.R. Onwuteaka-Philipsen B.D. Identifying patients who could benefit from palliative care by making use of the general practice information system: The Surprise Question versus the SPICT Fam. Pract. 2020 37 641 647 10.1093/fampra/cmaa049 32424418 PMC7571774
48. Piers R. De Brauwer I. Baeyens H. Velghe A. Hens L. Deschepper E. Henrard S. De Pauw M. Van Den Noortgate N. De Saint-Hubert M. Supportive and Palliative Care Indicators Tool prognostic value in older hospitalised patients: A prospective multicentre study BMJ Support. Palliat. Care 2021 14 e003042 10.1136/bmjspcare-2021-003042 PMC11672032 34059507
49. Farfán-Zuñiga X. Zimmermann-Vildoso M. Cultural adaptation and validation of the SPICT-ES™ instrument to identify palliative care needs in Chilean older adults BMC Palliat. Care 2022 21 223 10.1186/s12904-022-01111-2 36527008 PMC9756484
50. Liao J.Y. Chang H.T. Peng J.K. Murray S. Wu C.Y. Chang H.C. Li C.M. Cheng S.Y. Tseng W.Z. Hsiung C. Adaptation of the Taiwan Version of the Supportive and Palliative Care Indicators Tool (SPICT-TW) and Its Association with Six-Month Mortality: A Multi-Center Validation Study in Older People Healthcare 2024 12 2185 10.3390/healthcare12212185 39517396 PMC11545587
51. Xie Z. Tang S. Johnson C.E. Xiao L. Ding J. Huang C. Translation, cross-cultural adaptation and validation of the Chinese version of supportive and palliative care indicators tool (SPICT-CH) to identify cancer patients with palliative care needs BMC Palliat. Care 2025 24 4 10.1186/s12904-024-01641-x 39773249 PMC11708097
52. Huang Y. Deng C. Lin H. Peng M. Hao Y. Adaptation and Validation of the Supportive and Palliative Care Indicators Tool (SPICT): A Quantitative Methodological Study J. Adv. Nurs. 2025 81 6392 6405 10.1111/jan.16782 39873129 PMC12460957
53. Broese J.M.C. van der Kleij R.M.J.J. Verschuur E.M.L. Kerstjens H.A.M. Bronkhorst E.M. Chavannes N.H. Engels Y. External Validation and User Experiences of the ProPal-COPD Tool to Identify the Palliative Phase in COPD Int. J. Chron. Obs. Pulmon Dis. 2022 17 3129 3138 10.2147/COPD.S387716 PMC9792220 36579356
54. Cerullo G. Videira-Silva A. Carrancha M. Rego F. Nunes R. Complexity of patient care needs in palliative care: A scoping review Ann. Palliat. Med. 2023 12 791 802 10.21037/apm-22-894 37038064
55. Seven A. Sert H. Anxiety, dyspnea management, and quality of life in palliative care patients: A randomized controlled trial Florence Nightingale J. Nurs. 2023 31 S13 S21 10.5152/FNJN.2023.23016 37162050 PMC10911780
56. Batstone E. Bailey C. Hallett N. Spiritual care provision to end-of-life patients: A systematic literature review J. Clin. Nurs. 2020 29 3609 3624 10.1111/jocn.15411 32645236
57. Maas E.A. Murray S.A. Engels Y. Campbell C. What tools are available to identify patients with palliative care needs in primary care: A systematic literature review and survey of European practice BMJ Support. Palliat. Care 2013 3 444 451 10.1136/bmjspcare-2013-000527 24950525
58. ElMokhallalati Y. Bradley S.H. Chapman E. Ziegler L. Murtagh F.E.M. Johnson M.J. Bennett M.I. Identification of patients with potential palliative care needs: A systematic review of screening tools in primary care Palliat. Med. 2020 34 989 1005 10.1177/0269216320929552 32507025 PMC7388141
59. Xie Z. Ding J. Jiao J. Tang S. Johnson C.E. Screening instruments for early identification of unmet palliative care needs: A systematic review and meta-analysis BMJ Support. Palliat. Care 2024 14 256 268 10.1136/spcare-2023-004465 PMC11347222 38154921
60. Coventry P.A. Grande G.E. Richards D.A. Todd C.J. Prediction of appropriate timing of palliative care for older adults with non-malignant life-threatening disease: A systematic review Age Ageing 2005 34 218 227 10.1093/ageing/afi054 15863407
61. Boland J. Martin J. Wells A.U. Ross J.R. Palliative care for people with non-malignant lung disease: Summary of current evidence and future direction Palliat. Med. 2013 27 811 816 10.1177/0269216313493467 23838376