Palliative prognostic tools in surgical patients at the end of life: a systematic review

Abstract

Background

Identifying surgical patients at the end of life (EOL) is the first step in integrating palliative and perioperative practices. Palliative prognostic tools (PPTs) are established frameworks from palliative care that assess patients at risk of early death. We conducted a systematic review investigating PPTs in adult surgical populations, their role in surgical decision-making, and their association with perioperative outcomes.

Methods

A prospectively registered systematic review was performed (PROSPERO registration: CRD42023411303). Ovid MEDLINE, Ovid EMBASE, and Cochrane CENTRAL (Wiley) databases were searched for studies investigating PPTs in surgical patients. The primary outcome was the decision to proceed to surgery; secondary outcomes included mortality, quality of life, palliative care consultation, and EOL documentation completion. Abstract screening, full-text review, and study quality appraisal were performed by two authors independently. Results were synthesised narratively owing to study heterogeneity.

Results

Seven studies assessing four different PPTs were included in the review. Studies identified that 12–61% of surgical patients were at the EOL. Patients identified as being at the EOL by a PPT using an illness phase, trajectory approach, or both had an increased in-hospital and 12-month mortality. The impact on decisions to proceed to surgery was uncertain because of conflicting results. Palliative care referral and EOL document completion occurred in <15% of surgical patients at the EOL. No studies described patient-reported outcomes.

Conclusions

Palliative prognostic tools have significant potential for incorporation into preoperative assessment. Future research should focus on preoperative end of life assessments and patient-reported outcomes such as quality of life, decision satisfaction, and disability-free survival.

Editor's key points.

• •The palliative care needs of surgical patients at the end of life are largely unrecognised. The role of palliative prognostic tools (PPTs) in perioperative assessment and decision-making is unknown.
• •PPTs utilising an illness phase, trajectory approach, or both can identify surgical patients at high risk of dying within 12 months. PPTs assist with identifying surgical patients at the end of life and trigger integrated perioperative palliative practices.
• •Research on prospective PPT assessments and their association with patient-reported outcomes is needed.

As global populations age alongside advances in healthcare, more people are considered for surgery at the end of life (EOL). The EOL is when a person is impaired by a life-limiting condition or serious illness including advanced age, severe frailty, organ failure, or metastatic cancer, and is at high-risk or likely to die within the next 12 months.1, 2, 3, 4, 5, 6 Combined palliative care and surgery research priorities highlight the need to identify surgical patients at the EOL and not just the imminently dying.⁷ Similarly, perioperative guidance recommends recognising the EOL and discussing prognosis as part of the decision-making process.8, 9, 10, 11

Frailty assessments have an established supportive evidence base in surgical, trauma, and critically ill patients.12, 13, 14, 15, 16, 17 In surgical patients, higher frailty scores predict increased mortality, poorer functional outcomes, and can trigger specialist geriatric input.^14,15,18, 19, 20, 21 Although overlap exists between patients with severe frailty and those at the EOL, frailty assessments may miss patients on non-frailty EOL trajectories. In palliative care, palliative prognostic tools (PPTs) are formal frameworks to identify patients at the EOL. PPTs with trigger criteria identify the illness phase or trajectory. An illness phase is a cross-sectional EOL assessment based on acute and chronic illness.22, 23, 24 An illness trajectory is a longitudinal EOL assessment based on the dominant life-limiting condition (Supplementary Fig. S1).²⁵ Examples of tools that assess illness phase, trajectory, or both are the Surprise Question (SQ), the Gold Standards Framework – Proactive Indicator Guidance (GSF-PIG), and Supportive and Palliative Care Indicator Tool (SPICT).^22,26,27 PPTs that calculate numerical scores based on function and symptom burden domains include the Palliative Performance Scale (PPS), the Palliative Prognostic Index, and Palliative Index.28, 29, 30

Similar to frailty assessments, PPTs may have potential for integration into preoperative evaluation to assist with decision-making for surgery. Patients assessed to be at the EOL by a PPT may experience poorer outcomes related to postoperative mortality or quality of life. The use of a PPT may be associated with indicators of high-quality EOL care delivered to patients potentially in the last 12 months of life,1, 2, 3 including palliative care consultation or advance care planning. Although multiple systematic reviews on frailty assessments in surgical patients are published,^12,31, 32, 33, 34, 35 there are none on PPTs in perioperative care. We conducted a systematic review investigating PPTs in adult surgical populations, their role in surgical decision-making, and their association with perioperative outcomes.

Methods

Search strategy

We conducted a systematic search in the electronic databases Ovid MEDLINE, Ovid EMBASE, and Cochrane CENTRAL (Wiley) using a search strategy developed by a research health sciences librarian (SL) in March 2023. The full search strategy is outlined in Supplementary Appendix A. Results were limited to English language and adults aged ≥18 yr. All publication dates were included. We conducted the systematic review in accordance with the Preferred Reporting Items for Systematic Review and Meta-analyses (PRISMA) statement. The systematic review protocol was prospectively registered with PROSPERO (registration: CRD42023411303). The search strategy was repeated in July 2024 and found no new articles for inclusion.

Palliative prognostic tool

A PPT was defined as an established tool or framework used in palliative care that identifies patients at the EOL to assess palliative care needs or predict survival. Frailty assessments were not included as they were generally considered part of geriatric medicine practice but not routine palliative care practice and thus did not fulfil the definition of a PPT.

Outcomes of interest

The primary outcome was the decision to proceed to surgery. Secondary outcomes included mortality, palliative care consultation, completion of treatment limitation documentation, advance care planning, shared decision-making, quality of life, health-related satisfaction, and function. Studies were included if primary or secondary outcomes were reported.

Inclusion and exclusion criteria

Studies were included if a PPT assessed participants who were referred for surgery, admitted to a surgical ward/unit, or underwent surgery. Included study designs were RCTs, and observational, cross-sectional, cohort, case-control, and qualitative studies. Studies investigating palliative surgical techniques (procedures to relieve symptoms from a life-limiting condition) or their outcomes were excluded. Other exclusion criteria were paediatric populations and studies that developed PPTs, developed palliative care recommendations for surgical patients, investigated prognostic tools not used in palliative care, or did not report outcomes related to PPTs. Excluded publication formats were guidelines, recommendations, commentary, editorials, conference submissions/abstracts, and review articles.

Study selection and assessment

Abstract screening and full-text review were performed independently by two authors (CWL/ABW), with disagreement resolved through consensus. Evaluation of study quality was appraised by two authors (CWL/JD) as low, moderate, or high quality using a modification of the epidemiological appraisal instrument for the evaluation of epidemiological studies³⁶ (Supplementary Appendix B), with disagreement resolved through consensus.

Data extraction and synthesis

Data were extracted from included studies regarding the study design, population, setting, number of surgical study participants, PPT investigated, proportion of surgical participants assessed as being at the EOL, primary and secondary outcomes (CWL). Because of the heterogeneity of study methods and measures, meta-analysis was not possible. Instead, a narrative synthesis of study results was performed.

Results

Our search strategy identified 4562 articles including 1566 duplicates. Abstract screening removed a further 2906 articles; 90 articles were extracted for full-text review. Of these, 83 articles were excluded, and the seven remaining articles were included for analysis (Fig. 1). Two non-English language studies^37,38 were found after screening of a further 559 English-translated abstracts retrieved when English language limitations were removed from the search strategy. These studies were not included in the analysis.

Fig 1.

PRISMA flowchart for literature search strategy.

Study characteristics and quality

There were five cohort studies and two cross-sectional studies from five different countries. All were single-centre studies in tertiary, academic, or referral hospitals. The study populations were adult inpatients aged ≥18 yr old. In four studies, all participants were admitted to a surgical unit or undergoing surgery.39, 40, 41, 42 In three studies, surgical participants were a sub-cohort of total participants with selected results reported separately.43, 44, 45 Three studies were assessed as high quality, and four as low quality. The characteristics of included studies are detailed in Table 1.

Table 1.

Characteristics of included studies. CI, confidence interval; DNR, do not resuscitate; EOL, end of life; GOC, goals of care; GSF-PIG, Gold Standards Framework – Proactive Indicator Guidance; OR, odds ratio; PC, palliative care; PI, Palliative Index; RR, relative risk; SPICT, Supportive and Palliative Care Indicator Tool; SQ, Surprise Question.

Study author (year), country	Study design	Study population, setting	Study participants and age	Quality of study (modified Genaidy score)	Palliative prognostic tool	Proportion of surgical patients or admissions assessed to be at the EOL, % (n) (study criteria)	Primary outcome: decision for surgery	Secondary outcomes: mortality, advance care/GOC discussions/documentation, palliative care (PC) referral or review
Cooper and colleagues (2022), Australia45	Observational point prevalence study, single centre	Adult inpatients admitted to surgical and medical wards at a private tertiary hospital	162 surgical patients (270 total patients) Median age 70 yr (IQR not reported)	Low (28)	GSF-PIG, SQ	12% (19/162 patients) (SQ ‘No’ response and one to two specific indicators from GSF-PIG)	Not reported	EOL participants: 1.PC review: 15% (3/19)
Dogbey and colleagues (2022), South Africa39	Cross-sectional observational study (12-week period), single centre	Adult inpatients with cancer admitted to a surgical emergency centre for acute surgical emergencies at a tertiary academic hospital	100 surgical patients, 112 admissions Median age 58 yr (IQR 45–68 yr)	High (38)	SPICT, SQ	46% (52/112 admission) (SPICT one specific indicator and two general indicators) 55% (29/53 admissions) (SQ ‘No’ response)	No difference in undergoing a palliative surgery in participants identified to be at the EOL or not at the EOL: SPICT+ vs SPICT–: RR 1.0 (95% CI 0.6–1.7, P=0.90) SQ ‘No’ vs ‘Yes’: RR 1.1 (95% CI 0.5–2.4, P=0.79)	Mortality In-hospital: 1.Overall 19.6% 2.SPICT+ patients: 34.6% SPICT+ vs SPICT– 1.RR in-hospital mortality 5.2 (95% CI 1.87–13.4, P=0.002) 2.RR 6-month mortality 1.9 (95% CI 1.4–2.5, P<0.001) SQ response ‘No’ vs ‘Yes’ 1.RR in-hospital mortality 4.6 (95% CI 1.1–18.6, P=0.035) 2.RR 6-month mortality 2.1 (95% CI 1.3–3.9, P=0.004) Proportional agreement 1.SPICT+ and SQ ‘No’: 71%
Laitamäki and colleagues (2021), Finland40	Retrospective cohort study, single centre	Adult inpatients undergoing palliative emergency laparotomy surgery at a tertiary hospital	93 surgical patients Median age 69 yr (IQR not reported)	Low (29)	PI	Proportion of participants with each PI score not reported.	N/A (all patients underwent surgery)	Postoperative mortality 1.14-day: 22% 2.30-day: 41% 3.90-day: 63% 4.12-month: 97% PI not predictive of mortality or serious morbidity at above time points
Li and colleagues (2021), Kenya42	Prospective cohort study, single centre	Adult inpatients admitted to a surgical ward (neurosurgery, general surgery, or orthopaedic surgery) at a national referral hospital	411 surgical patients Median age 40 yr (IQR 29–57 yr)	Low (32)	SQ (Part of a palliative care trigger tool)	25% (102/411 patients) (SQ ‘No’ response) 25% (102/411 patients) (≥3/7 palliative care trigger tool)	Not reported	Mortality: Overall in-hospital: 7% (28/411) PC review: Overall discharge to hospice: 2% (8/411) SQ ‘No’ vs ‘Yes’ In-hospital mortality or discharge to hospice: 1.Sensitivity 64% 2.Specificity 79%
Lilley and colleagues (2017), USA41	Prospective cohort study, single centre	Adult inpatients aged 65 yr or older hospitalised with an acute surgical condition at an academic hospital	119 surgical patients, 163 physician responses Mean age: 1.EOL: Mean age 79 yr (range not reported) 2.Non-EOL: Mean age 74 yr (range not reported)	Low (30)	SQ	60% (93/119 patients) (SQ ‘No’ response)	A lower proportion of participants identified to be at the EOL underwent emergency surgery. SQ ‘No’ vs ‘Yes’: 24% vs 45% (P<0.01)	Mortality: Overall 12-month: 39% (47/119) Mortality: SQ ‘No’ vs ‘Yes’ 1.OR 5 (95% CI 2–11, P<0.001) 2.Sensitivity 81% (95% CI 71–91%) 3.Specificity 51% (95% CI 41–61%) 4.Positive predictive value 52% (95% CI 42–61%) 5.Negative predictive value 82% (95% CI 72–91%) 6.Accuracy 64% 7.Predictive performance AUROC 0.7 (0.6–0.8) SQ ‘No’: 1.Palliative care review: 6% 2.DNR document 13%
Mudge and colleagues (2018), Australia44	Prospective cohort study, single centre	Adult inpatients admitted to hospital at a metropolitan tertiary public hospital	188 surgical patients (513 total patients) Mean age 60 yr (range 19–102 yr) Mean age: 1.EOL: Mean age 69 yr (range not reported) 2.Non-EOL: Mean age 55 yr (range not reported)	High (37)	GSF-PIG, SQ, SPICT	28% (52/188 patients) (SQ ‘No’ response and two general indicators of decline from SPICT)	Not reported	Mortality: 12-month: 1.Surgical participants: not reported 2.Overall: 18% (92/513) 3.EOL participants: 38% (72/191) 4.Non-EOL participants: 6% (20/322) EOL vs non-EOL for 12-month mortality 1.Sensitivity 78% (95% CI 68–86) 2.Specificity 72% (95% CI 67–76) 3.Positive predictive value 38 (95% CI 33–42) 4.Negative predictive value 94% (95% CI 91–96) 5.Accuracy 73% (95% CI 69–77)
Orford and colleagues (2016), Australia43	Prospective cohort study, single centre	Adult inpatients referred to ICU at a regional tertiary public hospital	517 surgical patients (1024 total patients) Median age: 1.EOL: 74 yr (IQR 62–82) 2.Non-EOL: 62 yr (IQR 47–73)	High (38)	GSF-PIG, SPICT	61% (317/517 patients) (One specific or general indicator from SPICT/GSF-PIG)	Not reported	Mortality 12-month: 17.6% (91/517) GOC documentation: 14% (72/517) Comparison of 12-month mortality or GOC documentation between EOL and non-EOL surgical patients not reported

Palliative prognostic tools investigated

The PPTs investigated were the SQ (five studies^{39,41,42,44,45}), SPICT (three studies^39,43,44), GSF-PIG (three studies43, 44, 45), and Palliative Index (one study⁴⁰). Details of the PPTs investigated are in Table 2. Three studies investigated more than one PPT. Five studies identified the illness phase,^{39,41,42,44,45} and four studies identified the illness trajectory.^39,43, 44, 45 The Palliative Index was the only PPT that calculated a numerical score.⁴⁰

Table 2.

Palliative prognostic tools investigated in included studies. ASA, American Society of Anesthesiologists Physical Grading Score. ∗General indicators of decline include overall deterioration, decreasing functional activity, increasing dependence, progressive weight loss. ^†Specific indicators of decline relate to deterioration directly owing to certain conditions such as advanced cancer, end-organ failure. or dementia. (From The Gold Standards Framework – Proactive Indicator Guidance.²⁶)

Palliative prognostic tool	Criteria	Identifies	Purpose
Surprise Question (SQ)22	‘No’ response to ‘Would I be surprised if the patient dies in the next 12 months’?	Illness phase	Assess palliative care needs
Gold Standard Framework – Proactive Indicator Guidance (GSF-PIG)26	SQ ‘No’ response General indicators of decline∗ Specific indicators of decline†	Illness phase and illness trajectory	Assess palliative care needs
Supportive and Palliative Care Indicator Tool (SPICT)27	General indicators of decline∗ Specific indicators of decline†	Illness trajectory	Assess palliative care needs
Palliative Index30	Four domains 1.Active disease 2.Tumour-related indication for surgery 3.ASA physical status >3 4.Creatinine >1.3 mg dl−1	Score 0–5 Two points for active disease One point for each other criterion	Predict survival after surgery for patients with cancer (higher scores predict poorer survival)

Identification of surgical patients at the end of life

The proportion of surgical patients at the EOL ranged 12–61%.The methodology and criteria for identifying the EOL was heterogenous even when investigating the same PPT.^39,43, 44, 45 The number of EOL criteria did not vary with study quality. Higher proportions (55–61%) were identified in older patients, patients with cancer, and patients referred to ICU.^39,41,43 These studies only used one criterion to identify the EOL, and this combination of high-risk baseline population and less specific EOL criteria may account for the high proportion of EOL patients identified. In contrast, three studies investigating patients admitted to hospital surgical wards and required two to three indicators of decline demonstrated lower proportions at 12–28%.^42,44,45

Decision to proceed to surgery

The impact of identifying the EOL with a PPT on the decision to proceed to surgery was uncertain. Only two studies of contrasting quality reported on this outcome and the conflicting results and methodology made interpretation difficult.^39,41 A lower proportion of SQ ‘No’ older patients admitted with an acute surgical illness underwent surgery.⁴¹ However, the SQ was asked of treating doctors within 72 h of patient admission, by which time patients may already have undergone surgery. In patients with cancer, there was no difference in the relative risk of palliative surgery when the EOL was identified by either ‘SPICT+’ or an SQ ‘No’ answer.³⁹ In this study, the SPICT assessment was performed retrospectively and was not available for decision-making. The SQ was assessed on admission by the admitting doctor, but it is unclear if this was an operating team member.

Mortality

Five studies reported the mortality rate of surgical patients.39, 40, 41, 42, 43 The in-hospital and 12-month mortality rates ranged 7–20%^39,42 and 18–87%,^40,41,43 respectively. The EOL identified through either the SQ ‘No’ response or ‘SPICT+’ were associated with increased in-hospital and 12-month mortality in surgical patients. Although only one study was of high quality, all three studies reported statistical parameters with this finding, increasing the confidence of this result.^39,41,42 In contrast, the Palliative Index was not predictive of postoperative mortality up to 12 months after palliative emergency laparotomy.⁴⁰

Palliative care review and completion of treatment limitation documentation

Two low-quality studies reported rates of palliative care review and treatment limitation documentation. For those identified to be at the EOL, only 6–15% of patients were reviewed by palliative care.^41,45 Similarly, only 13% of EOL patients had treatment limitation documentation completed.⁴¹

Discussion

Key findings

We only identified seven studies that investigated PPTs in surgical patients. In this review, 12–61% of surgical patients were identified to be at the EOL, with the highest proportions occurring in high-risk cohorts.^41,43 The impact of the PPT assessment on the decision to proceed to surgery was uncertain.^39,41 Patients identified to be at the EOL by an illness phase, trajectory approach, or both had increased in-hospital and 12-month mortality.^39,41,42 Referral to palliative care and completion of treatment limitation documents were <15% in surgical patients at the EOL.^41,45

Relationship with previous literature

Criteria to identify the end of life and serious illness indicators

In our review, higher proportions (55–61%) of EOL surgical patients were identified in older and ICU referral cohorts, but these studies also only used one criterion to assess the EOL, thus being more sensitive but less specific in their assessment.^39,41, 42, 43, 44 Serious illness indicators also identify patients at high risk of early death. Indicators for serious illness were defined by Lee and colleagues⁴ to identify surgical patients who would benefit from palliative care. These ‘serious illness indicators’ are similar to illness trajectory ‘indicators of decline’ from the GSF-PIG and SPICT.^39,43, 44, 45 Similar to our findings, studies using serious illness indicators demonstrated more sensitivity with less indicators. In older surgical patients, 69% had unmet palliative care needs when using one indicator, but this decreased to 33–57% when using two or more indicators.^46,47 Lee and colleagues⁴ noted their definition of serious illness was highly sensitive and understanding these criteria within the perioperative context of the overall illness trajectory was essential. Whether utilising serious illness indicators or PPTs, considering how many criteria to apply must account for available perioperative palliative resources. Using only one criterion may overwhelm growing services, but too many may exclude patients in need.

Frailty assessments in surgical patients

Over the past decade, guidelines have been published incorporating perioperative frailty assessments from geriatric medicine.^11,48,49 We excluded frailty assessments from our definition of PPTs as we primarily aimed to investigate assessments from palliative care and considered frailty assessments generally more consistent with geriatric screening. However, frailty assessments can also be useful EOL identification tools as there are many commonalities in palliative care and geriatric management for patients with severe frailty. Patients with higher degrees of frailty have comparable palliative care needs to patients with advanced cancer⁵⁰ and frailty screening has been shown to increase palliative care consultation in high-risk surgical patients.⁵¹ Frailty assessments such as the Clinical Frailty Scale (CFS) identify patients with advanced frailty (CFS ≥7) as being at the EOL.⁵² Notably, a CFS 9 includes all patients who are likely to die within 6 months, even if they are not living with severe frailty.⁵³ Likewise, frailty as an EOL illness trajectory was identified in the GSF-PIG and SPICT in four included studies.^39,43, 44, 45 Although these PPTs differ from frailty assessments by also screening for non-frailty EOL illness trajectories, the co-existence of frailty with the cancer or organ failure illness trajectory is associated with worse mortality.^43,54 When assessing the EOL illness phase, a ‘No’ response to the SQ is a significant predictor of frailty (odds ratio [OR] 8.9; 95% confidence interval [CI] 2.3–27.0) identified by comprehensive geriatric assessment in older adults.⁵⁵ For PPTs that primarily assess activity performance, lower PPS scores are associated with an increasing degree of frailty on the CFS.⁵⁶ Although frailty assessments do not primarily aim to identify the EOL, in combination, both frailty assessments and PPTs can assist in identifying surgical patients at risk of early death who may benefit from a perioperative palliative approach.

Impact on decision to proceed to surgery

A previous systematic review found only two studies that explored the use of palliative care interventions on surgical decision-making.⁵⁷ These studies demonstrated that shared decision-making and multidisciplinary input to aide patient selection for surgery were associated with improved postoperative survival, functional recovery, and complications rates.^58,59 In our review, variable study quality, conflicting results, and retrospective assessment meant that the impact of PPTs on proceeding to surgery was uncertain.^39,41 International guidelines outline multiple steps between identifying a patient at the EOL and deciding whether to proceed to surgery.^{10,11,48,49,60} Common steps are: (1) establishing the patient's goals; (2) evaluating surgical intent; (3) completing advance care planning; and (4) discussing acceptability and extent of perioperative critical care and resuscitation. As identifying EOL is only one part of the process, it is perhaps expected that the impact of PPTs on decisions proceeding to surgery is unclear.

Lilley and colleagues⁴¹ reported fewer SQ ‘No’ patients underwent surgery but did not report the intent of surgery. Potentially, fewer EOL patients underwent surgery to prolong life for an acute surgical illness. Dogbey and colleagues³⁹ demonstrated no difference in the relative risk of palliative intent surgery for acutely unwell ‘SPICT+’ cancer patients compared with ‘SPICT’– patients. However, surgery to relieve symptoms performed at equal rates in cancer patients, regardless of EOL assessment, is unsurprising. These contrasting results highlight that knowing the intent of surgery and patient's goals is crucial when understanding the decision for surgery at the EOL.^10,11 Within the context of relevant guidelines and the prior review, although PPTs may identify who is at the EOL, clinicians must also know how to navigate the subsequent processes and discussions to assist with decisions around surgery.

Underutilisation of palliative care and treatment limitation documentation

Optimal standards in geriatric surgery suggest the SQ to screen patients for palliative care needs.¹¹ Once identified, patients at the EOL benefit from integrated management incorporating palliative care principles with standard treatment.⁶¹ Referral to palliative care, goals of care (GOC) discussions, and completion of treatment limitation documentation are considered indicators of high-quality EOL surgical care.^8,62 Our review demonstrated that <15% of EOL surgical patients received these indicators,^41,45 and although these were low-quality studies, this finding is consistent with the literature demonstrating unmet palliative care needs in surgical patients. In prior studies, the range of surgical EOL patients who received a palliative care consultation varied between 3% and 32%.63, 64, 65 Only one-third of surgical patients who died in hospital received a palliative care referral, with almost half of referrals occurring on the day of death.^63,66, 67, 68 Likewise, in older patients undergoing major surgery, 6% underwent advance care planning, 12% had GOC discussions, and rates of treatment limitation documentation ranged 2–5%.69, 70, 71, 72 Although the presence of advance care directives was higher at 35%, these were rarely discussed.^70,71

The findings of the landmark SUPPORT trial suggested that the engrained practices of physicians meant that indicators of high-quality EOL care did not improve despite timely prognostic patient information.⁷³ Interestingly, although palliative care trigger criteria did not improve rates of palliative consultations in surgical patients,^74,75 routine frailty screening did.⁵¹ Preoperative palliative care consultations in surgical patients were associated with lower mortality, higher rates of EOL documentation completion, better quality EOL care, and communication.^51,63,76,77 The impact of GOC discussions, advance care planning, and treatment limitation documentation on health-related satisfaction, quality of life, and communication in surgical patients was not studied in our systematic review. Although literature exploring the impact of GOC discussions in surgical patients is limited, potential benefits include confirming patient preferences for perioperative limitation on treatment and revealing unrealistic expectations of surgical outcomes.^78,79 In trauma patients at the EOL, those who had documented GOC discussions were more likely to have higher rates of palliative care consultation, pre-existing advance care planning, and treatment limitation documentation completed.⁸⁰ A qualitative study of hospital inpatients reported that patients were highly satisfied with GOC and shared decision-making discussions if they perceived that clinicians were comfortable with communication and spent enough time with them.⁸¹ This further emphasises that system-based practices around perioperative palliative care integration are essential for surgical patients at the EOL, and that further research is needed to evaluate the impact of performing preoperative GOC discussions, advance care planning, and treatment limitation documentation on patient-reported outcomes.

Implications for clinical practice

The under-recognition of surgical patients at the EOL may be attributed to narrow perceptions of EOL and palliative care in the hospital setting resulting in undue focus on active interventions and barriers to integrated management.82, 83, 84, 85, 86 Using an illness phase and trajectory approach, frequently conducted in nonsurgical populations,⁸⁷ may broaden the recognition of patients at the EOL and encourage integration of palliative care practices into surgical decision-making (Fig. 2). In this review, EOL patients identified by this approach had higher all-cause in-hospital and 12-month mortality raising questions on whether they have the time and reserve to achieve their preferred outcomes after surgery.

Fig 2.

End-of-life assessment in surgical patients and potential clinical process. GSF-PIG, Gold Standards Framework – Proactive Indicator Guidance; SPICT, Supportive and Palliative Care Indicator Tool. ^†Modified Surprise Question from Lilley EJ, Gemunden SA, Kristo G, et al. Utility of the “surprise” question in predicting survival among older patients with acute surgical conditions. J Palliat Med 2017; 20: 420–3.

Once identified, the processes to ensure meaningful and sensitive conversations are critical to navigating the decision to proceed to surgery.^88,89 Understanding patient goals, evaluating the intention of surgery, and completing advance care plans are recommended.^10,11,48,49 ‘Serious illness conversations’90, 91, 92 and ‘best case/worst case’⁹³ frameworks can be utilised in shared decision-making processes to establish whether patient's goals can be met with surgery and the acceptability of perioperative resuscitation and critical care.^{10,11,48,49,60} In most situations, this can be initiated by perioperative practitioners with referral to specialist palliative care in complex circumstances.⁸

Strengths and limitations

Our review is the first to investigate PPTs in surgical patients. Strengths include a robust search strategy, without publication time limits, devised by an experienced health sciences librarian, including studies where surgical patients were a sub-cohort if outcomes were reported separately, and abstract screening and full-text review by clinician–researchers in palliative care, geriatric medicine, and perioperative care. A major limitation is that we were unable to draw a conclusion on surgical decision-making owing to lack of studies, heterogenous methodology, and conflicting results. Other limitation included applying English language limitations, which excluded two potential non-English studies. The heterogeneity of included studies necessitated narrative synthesis of results and precluded meta-analysis. Publication bias was not assessed because of the small number of studies and observational study methodology. We also excluded frailty assessments as discussed earlier; however, we emphasise that this decision does not preclude using frailty assessments as prognostic tools in older adults.

Future direction of research

Future research could explore PPTs calculating numerical scores, prospective PPT administration, and patient-reported outcomes. PPTs utilising numerical scores such as the PPS and Palliative Prognostic Index were lacking despite their investigation in other populations.94, 95, 96, 97, 98, 99, 100, 101 Investigating the prospective administration of PPTs in the surgical population could explore their impact on decisions to proceed to surgery and quality indicators for surgical palliative care.⁶² Finally, all included studies lacked patient-reported outcomes even though older patients at the EOL may value quality of life over life prolongation.¹⁰² Avoidance of complications, days alive at home at 30 days (DAH30), and disability-free survival are important patient-reported outcomes in older surgical patients.¹⁰³ The ‘worth it’ outcome or ‘decision regret scale’ can also explore decision satisfaction for patients at the EOL.104, 105, 106 Understanding the association between EOL assessments by PPTs, these patient-reported outcomes may provide valuable information in preoperative shared decision-making, GOC, and advance care planning discussions.

Conclusions

Similar to frailty assessments, PPTs have potential for integration into perioperative care. A preoperative illness phase and trajectory assessment could potentially trigger integrated palliative and perioperative practices to assist with decisions to proceed to surgery. Future research exploring the association between EOL assessments from PPTs and patient-reported outcomes may help ensure that high-value surgery is delivered to surgical patients at the EOL.

Authors’ contributions

Conceptualisation and methodology: CWL, ABW, SL, WKL, JD

Data acquisition: CWL, SL, ABW

Data analysis: CWL, JD

Writing (original draft): CWL

Writing (review and editing): CWL, ABW, SL, WKL, JD

Funding

Australian and New Zealand College of Anaesthetists Foundation (Skantha Vallipuram Research Scholarship 2024); Australian Government Research Training Program Scholarship for the first author.

Declaration of interest

The authors disclose there are no conflicts of interest.

Handling Editor: Jonathan Hardman

Appendix A. Supplementary data

The following are the Supplementary data to this article: