Background: Home-based palliative care (HPC) is considered to moderate the problem of rising healthcare utilization of cancer patients at end-of-life. Reports however suggest a proportion of HPC patients continue to experience high care intensity. Little is known about differential trajectories of healthcare utilization in patients on HPC. Thus, we aimed to uncover the heterogeneity of healthcare utilization trajectories in HPC patients and identify predictors of each utilization pattern. Methods: This is a cohort study of adult cancer patients referred by Singapore Health Services to HCA Hospice Service who died between 1st January 2018 and 31st March 2020. We used patient-level data to capture predisposing, enabling, and need factors for healthcare utilization. Group-based multi-trajectory modelling was applied to identify trajectories for healthcare utilization based on the composite outcome of emergency department (ED) visits, hospitalization, and outpatient visits. Results: A total of 1572 cancer patients received HPC (median age, 71 years; interquartile range, 62–80 years; 51.1% female). We found three distinct trajectory groups: group 1 (31.9% of cohort) with persistently low frequencies of healthcare utilization, group 2 (44.1%) with persistently high frequencies, and group 3 (24.0%) that begin with moderate frequencies, which dropped over the next 9 months before increasing in the last 3 months. Predisposing (age, advance care plan completion, and care preferences), enabling (no medical subsidy, primary decision maker), and need factors (cancer type, comorbidity burden and performance status) were significantly associated with group membership. High symptom needs increased ED visits and hospitalizations in all three groups (ED visits, group 1–3: incidence rate ratio [IRR] 1.74–6.85; hospitalizations, group 1–3: IRR 1.69–6.60). High home visit intensity reduced outpatient visits in all three groups (group 1–3 IRR 0.54–0.84), while it contributed to reduction of ED visits (IRR 0.40; 95% CI 0.25–0.62) and hospitalizations (IRR 0.37; 95% CI 0.24–0.58) in group 2. Conclusions: This study on HPC patients highlights three healthcare utilization trajectories with implications for targeted interventions. Future efforts could include improving advance care plan completion, supporting care preferences in the community, proactive interventions among symptomatic high-risk patients, and stratification of home visit intensity.

Background: Ex-ante identification of the last year in life facilitates a proactive palliative approach. Machine learning models trained on electronic health records (EHR) demonstrate promising performance in cancer prognostication. However, gaps in literature include incomplete reporting of model performance, inadequate alignment of model formulation with implementation use-case, and insufficient explainability hindering trust and adoption in clinical settings. Hence, we aim to develop an explainable machine learning EHR-based model that prompts palliative care processes by predicting for 365-day mortality risk among patients with advanced cancer within an outpatient setting. Methods: Our cohort consisted of 5,926 adults diagnosed with Stage 3 or 4 solid organ cancer between July 1, 2017, and June 30, 2020 and receiving ambulatory cancer care within a tertiary center. The classification problem was modelled using Extreme Gradient Boosting (XGBoost) and aligned to our envisioned use-case: “Given a prediction point that corresponds to an outpatient cancer encounter, predict for mortality within 365-days from prediction point, using EHR data up to 365-days prior.” The model was trained with 75% of the dataset (n = 39,416 outpatient encounters) and validated on a 25% hold-out dataset (n = 13,122 outpatient encounters). To explain model outputs, we used Shapley Additive Explanations (SHAP) values. Clinical characteristics, laboratory tests and treatment data were used to train the model. Performance was evaluated using area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC), while model calibration was assessed using the Brier score. Results: In total, 17,149 of the 52,538 prediction points (32.6%) had a mortality event within the 365-day prediction window. The model demonstrated an AUROC of 0.861 (95% CI 0.856–0.867) and AUPRC of 0.771. The Brier score was 0.147, indicating slight overestimations of mortality risk. Explanatory diagrams utilizing SHAP values allowed visualization of feature impacts on predictions at both the global and individual levels. Conclusion: Our machine learning model demonstrated good discrimination and precision-recall in predicting 365-day mortality risk among individuals with advanced cancer. It has the potential to provide personalized mortality predictions and facilitate earlier integration of palliative care.

Factors associated with place of death inform policies with respect to allocating end-of-life care resources and tailoring supportive measures. To determine factors associated with non-hospital deaths among cancer patients. Retrospective cohort study of cancer decedents, examining factors associated with non-hospital deaths using multinomial logistic regression with hospital deaths as the reference category. Cancer patients (n = 15254) in Singapore who died during the study period from January 1, 2012 till December 31, 2105 at home, acute hospital, long-term care (LTC) or hospice were included. Increasing age (categories ≥65 years: RRR 1.25–2.61), female (RRR 1.40; 95% CI 1.28–1.52), Malays (RRR 1.67; 95% CI 1.47–1.89), Brain malignancy (RRR 1.92; 95% CI 1.15–3.23), metastatic disease (RRR 1.33–2.01) and home palliative care (RRR 2.11; 95% CI 1.95–2.29) were associated with higher risk of home deaths. Patients with low socioeconomic status were more likely to have hospice or LTC deaths: those living in smaller housing types had higher risk of dying in hospice (1–4 rooms apartment: RRR 1.13–3.17) or LTC (1–5 rooms apartment: RRR 1.36–4.11); and those with Medifund usage had higher risk of dying in LTC (RRR 1.74; 95% CI 1.36–2.21). Patients with haematological malignancies had increased risk of dying in hospital (categories of haematological subtypes: RRR 0.06–0.87). We found key sociodemographic and clinical factors associated with non-hospital deaths in cancer patients. More can be done to enable patients to die in the community and with dignity rather than in a hospital.

Introduction: End-of-life patients face difficulties in reporting respiratory distress. The Respiratory Distress Observation Scale (RDOS) is a well-known tool; however, field implementation has been challenging from ground feedback. We sought to develop a simpler scale. Setting: Patients referred for palliative consult in a tertiary hospital in Singapore were recruited. Methods: A priori , we identified 18 dyspnea physical signs and documented their presence through bedside observation. Dyspnea severity was self-reported. The cohort was randomly split into training and test sets. Partial least square regression with leave-one-out cross-validation was used to develop a four-point model from the training set. Using the test set, data fit was compared using Akaike and Bayesian Information Criterion. Discrimination was assessed using receiver operating characteristics. Results: Of 122 patients, mean age was 67.9 years (range 23–93, standard deviation 12.9), 71.3% had a primary cancer diagnosis, and 58.1% were chair/bedbound with a Palliative Performance Scale of ≤50. Median reported dyspnea scale was 5 (interquartile range 3–7). Our model (modRDOS-4) consisted of four predictors (grunting, respiratory rate, accessory muscle use, paradoxical breathing). A modRDOS-4 of ≥6 identified moderate-to-severe dyspnea with a sensitivity of 0.78 and specificity of 0.90. Using the test set, with the modRDOS-4, the Akaike Information Criterion (AIC) is 149.8, Bayesian Information Criteria (BIC) is 154.1, and the receiver operating characteristics (ROC) is 0.74. With the original RDOS, the AIC is 145.2, BIC is 149.5, and ROC is 0.76. Conclusion: For a quick assessment of dyspnea, we developed a four-item tool with a pilot web-based nomogram. External validation is needed.

Cancer is a pervasive global problem with growing healthcare utilisation and costs.1-3 This situation is similar in Singapore where cancer incidence is on the rise and accounts for nearly 30% of total population mortality.4,5 Singapore data suggests that cancer patients accounted for 13% of total healthcare costs in 2016, with a major portion driven by […]

Objectives: Patients with terminal cancer receiving home palliative care present differential healthcare utilization trajectories before death. It remains unclear which situational elements influence these trajectories among disparate patient groups. The aim of this study was to compare situational influences on “persistently high” and “low stable” trajectories of healthcare utilization in patients who received palliative care support at home.Methods: Bereaved family caregivers were recruited from our prior quantitative study investigating healthcare utilization trajectories in oncology patients on home-based palliative care. In-depth interviews were conducted with 30 family caregivers. Data were analyzed using thematic analysis.Results: Analysis of data uncovered how the 2 utilization trajectories were influenced by the interplay of 1 or more of 4 situational elements. Perceived symptom control in patients, influenced by their determination to die at home, shapes the susceptibility to situational contingencies, resulting in differential utilization trajectories. Caregivers’ mental readiness in dealing with unexpected circumstances has a significant impact on the overall manageability of care, ultimately affecting decisions related to healthcare utilization. The concordance between symptom needs and scope of homecare services in a given situation proves to be an important determinant. Lastly, perceived accessibility to informal support in times of need acts as a contextual reinforcement, either preventing or precipitating decisions regarding healthcare utilizations.Significance of results: Our findings hold important implications for the provision of homecare services, in particular, the need for comprehensive assessment of end-of-life wishes during homecare enrolment and strengthening psychological preparedness of caregivers. Expansion of home-based clinical interventions tailored to high utilizers, and funding for temporary in-home respite should be considered to optimally manage potentially preventable acute healthcare utilization.