Improving palliative and end-of-life care with machine learning and routine data: a rapid review.
ABSTRACT: Introduction: Improving end-of-life (EOL) care is a priority worldwide as this population experiences poor outcomes and accounts disproportionately for costs. In clinical practice, physician judgement is the core method of identifying EOL care needs but has important limitations. Machine learning (ML) is a subset of artificial intelligence advancing capacity to identify patterns and make predictions using large datasets. ML approaches have the potential to improve clinical decision-making and policy design, but there has been no systematic assembly of current evidence. Methods: We conducted a rapid review, searching systematically seven databases from inception to December 31st, 2018: EMBASE, MEDLINE, Cochrane Library, PsycINFO, WOS, SCOPUS and ECONLIT. We included peer-reviewed studies that used ML approaches on routine data to improve palliative and EOL care for adults. Our specified outcomes were survival, quality of life (QoL), place of death, costs, and receipt of high-intensity treatment near end of life. We did not search grey literature and excluded material that was not a peer-reviewed article. Results: The database search identified 426 citations. We discarded 162 duplicates and screened 264 unique title/abstracts, of which 22 were forwarded for full text review. Three papers were included, 18 papers were excluded and one full text was sought but unobtainable. One paper predicted six-month mortality, one paper predicted 12-month mortality and one paper cross-referenced predicted 12-month mortality with healthcare spending. ML-informed models outperformed logistic regression in predicting mortality but poor prognosis is a weak driver of costs. Models using only routine administrative data had limited benefit from ML methods. Conclusion: While ML can in principle help to identify those at risk of adverse outcomes and inappropriate treatment near EOL, applications to policy and practice are formative. Future research must not only expand scope to other outcomes and longer timeframes, but also engage with individual preferences and ethical challenges.
Project description:For women with stage IV breast cancer (BC), the association between survival time (ST) and use of aggressive end-of-life (EOL) care is unknown.We used the SEER-Medicare database to identify women with stage IV BC diagnosed 2002-2011 who died by 12/31/2012. Aggressive EOL care was defined as receipt in the last month of life: >1 ED visit, >1 hospitalization, ICU admission, life-extending procedures, hospice admission within 3 days of death, IV chemotherapy within 14 days of death, and/or ?10 unique physician encounters in the last 6 months of life. Receipt of aggressive EOL care and hospice in the last month of life were determined using claims, and multivariable analysis was used to identify factors associated with receipt. Costs of care were also evaluated.We identified 4521 eligible patients. Of these, 2748 (60.8%) received aggressive EOL care. Factors associated with aggressive EOL care were race (OR 1.45, 95% CI 1.19-1.81 for blacks compared to whites) and more frequent oncology office visits (OR 1.56, 95% CI 1.28-1.90). Patients who lived >12 months after diagnosis were less likely to receive aggressive EOL care (OR 0.44, 95% CI 0.38-0.52), and more likely to utilize hospice (OR 1.43, 95% CI 1.21-1.69) compared to patients who lived ?6 months. Patients with a shorter ST had significantly higher costs of care per-month-alive compared to patients with longer ST.Patients with a shorter ST were more likely to receive aggressive EOL care and had higher costs of care compared to patients who lived longer.
Project description:<h4>Objective</h4>This study aimed to define the end-of-life (EOL) healthcare utilisation and its cost and determinants for cancer patients and to proactively inform related strategies in mainland China.<h4>Design</h4>A population-based retrospective study.<h4>Setting and participants</h4>Data from 894 cancer patients were collected in urban Yichang, China from 01 July 2015 to 30 June 2017.<h4>Outcome measures</h4>Emergency department (ED) visits, outpatient and inpatient hospitalisation services, intensive care unit (ICU) admission and total costs were used as the main outcomes.<h4>Results</h4>In this study, 66.8% of the 894 patients were male, and the average age was 60.4 years. Among these patients, 37.6% died at home, and patients had an average of 4.86 outpatient services, 2.23 inpatient hospitalisation services and 1.44 ED visits. Additionally, 5.9% of these patients visited the ICU at least once. During the EOL periods, the costs in the last 6 months, 3 months, 1?month and 1?week were US$18 234, US$13 043, US$6349 and US$2085, respectively. The cost increased dramatically as death approached. The estimation results of generalised linear regression models showed that aggressive care substantially affected expenditure. Patients with Urban Employee Basic Medical Insurance spent more than those with Urban Resident-based Basic Medical Insurance or the New Rural Cooperative Medical Scheme. The place of death and the survival time are also risk factors for increased EOL cost.<h4>Conclusion</h4>The findings suggested that the EOL cost for cancer patients is associated with aggressive care, insurance type and survival time. Timing palliative care is urgently needed to address ineffective and irrational healthcare utilisation and to reduce costs.<h4>Ethics and dissemination</h4>This study was approved by the Ethics Committee of the Tongji Medical College, Huazhong University of Science and Technology (IORG No.: IORG0003571). All the data used in this study were de-identified.
Project description:BACKGROUND:Socioeconomic inequalities in access to, and utilization of medical care have been shown in many jurisdictions. However, the extent to which they exist at end-of-life (EOL) remains unclear. METHODS:Studies in MEDLINE, EMBASE, CINAHL, ProQuest, Web of Science, Web of Knowledge, and OpenGrey databases were searched through December 2019 with hand-searching of in-text citations. No publication date or language limitations were set. Studies assessing SES (e.g. income) in adults, correlated to EOL costs in last year(s) or month(s) of life were selected. Two independent reviewers performed data abstraction and quality assessment, with inconsistencies resolved by consensus. RESULTS:A total of twenty articles met eligibility criteria. Two meta-analyses were performed on studies that examined total costs in last year of life - the first examined costs without adjustments for confounders (n?=?4), the second examined costs that adjusted for confounders, including comorbidities (n?=?2). Among studies which did not adjust for comorbidities, SES was positively correlated with EOL costs (standardized mean difference, 0.13 [95% confidence interval, 0.03 to 0.24]). However, among studies adjusting for comorbidities, SES was inversely correlated with EOL expenditures (regression coefficient, -$150.94 [95% confidence interval, -$177.69 to -$124.19], 2015 United States Dollars (USD)). Higher ambulatory care and drug expenditure were consistently found among higher SES patients irrespective of whether or not comorbidity adjustment was employed. CONCLUSION:Overall, an inequality leading to higher end-of-life expenditure for higher SES patients existed to varying extents, even within countries providing universal health care, with greatest differences seen for outpatient and prescription drug costs. The magnitude and directionality of the relationship in part depended on whether comorbidity risk-adjustment methodology was employed.
Project description:There are increasing concerns about the negative impacts of chemotherapy near the end of life (EOL). There is discrepancy among different countries about its use, and little is known about the real-world situation in China.This retrospective study was conducted at six representative hospitals across China. Adult decedents with a record of advanced solid cancer and palliative chemotherapy were consecutively screened from 2010 through 2014. The prevalence of EOL chemotherapy within the last 1 month of life was set as the primary outcome. The correlations among EOL chemotherapy, clinicopathological features, and overall survival (OS) were investigated.A total of 3,350 decedents who had had cancer were consecutively included; 2,098 (62.6%) were male and the median age was 56 years (range, 20-88). There were 177 (5.3%), 387 (11.6%), and 837 (25.0%) patients who received EOL chemotherapy within the last 2 weeks, 1 month, and 2 months of life, respectively. We identified inferior OS (median OS, 7.1 vs. 14.2 months; hazard ratio, 1.37; 95% confidence interval [CI], 1.23-1.53; p < .001), more intensive treatments (e.g., admitted to intensive care unit [ICU] in the last month of life, received cardiopulmonary resuscitation and invasive ventilation support), and hospital death (odds ratio, 1.53; 95% CI, 1.14-2.06; p?=?.005) among patients who received continued chemotherapy within the last month compared with those who did not. However, subgroup analyses indicated that receiving oral agents correlated with fewer ICU admissions and lower rates of in-hospital death.This study showed that EOL chemotherapy is commonly used in China. Intravenous chemotherapy at the EOL significantly correlated with poor outcomes and the role of oral anticancer agents warrants further investigation. The Oncologist 2017;22:53-60Implications for Practice: The role of chemotherapy toward the end of life (EOL) in patients with solid cancers is debatable. This article is believed to be the first to report the current prevalence of EOL chemotherapy in China. This study found that, compared with oral anticancer agents, intravenous chemotherapy at the EOL was significantly associated with poor outcomes. Therefore, the role of oral anticancer agents at the EOL stage deserves further investigation.
Project description:BACKGROUND:Medical care of cancer patients at the end-of-life is costly. This study aims to describe the monthly trends of EOL medical care, drug therapy, and chemotherapy costs per patient with cancer in the last year of life in the inpatients vs. outpatient setting for the 13 most prevalent cancers in Korea. METHODS:Using the Health Insurance Review and Assessment Service (HIRA) database, we identified the patients who had been treated for the primary diagnoses of one of the 13 most prevalent cancers in Korea and died between January 1, 2013 and December 31, 2015. We calculated the mean monthly costs of medical care, drug therapy, and chemotherapy per patient in the last year of life by cancer site and patient setting (inpatient vs. outpatient). RESULTS:For most cancers, the monthly inpatient costs per patient remain stable or increased gradually from 12 months to 3 months prior to death and then increased steeply from 2 months prior to death. The mean monthly inpatient costs per patient were highest for acute myeloid leukemia (AML) throughout the last year of life; all solid tumors had similar trends of monthly inpatient costs. The mean monthly inpatient costs for AML increased from $5,465 (SD, $5,248) in 12 months prior to death to $15,033 (SD, $11,864) in the last month. The monthly outpatient costs per patient showed similar, gradually decreasing trends for most cancers. The mean outpatient costs were highest for kidney cancer; the costs sharply decreased from $954 (SD, $1,346) in 12 months prior to death to $424 (SD, $736) in the last month. The proportion of inpatients receiving chemotherapy in the last month of life was highest for AML (77%), followed by liver cancer (67%) and breast cancer (56%). CONCLUSION:The monthly inpatient medical care costs per patient with cancer increased as the patient approached death, while the monthly outpatient costs decreased. A considerable proportion of inpatient received chemotherapy in the last month of life. Efforts are needed to optimize EOL care for cancer patients.
Project description:The US has higher infant mortality than peer countries. In this paper, we combine micro-data from the US with similar data from four European countries to investigate this US infant mortality disadvantage. The US disadvantage persists after adjusting for potential di erential reporting of births near the threshold of viability. While the importance of birth weight varies across comparison countries, relative to all comparison countries the US has similar neonatal (<1 month) mortality but higher postneonatal (1-12 months) mortality. We document similar patterns across Census divisions within the US. The postneonatal mortality disadvantage is driven by poor birth outcomes among lower socioeconomic status individuals.
Project description:PURPOSE:End-of-life (EOL) chemotherapy has been described as the most widespread, wasteful, and unnecessary practice in oncology, with benchmarking aimed to reduce physician use of chemotherapy within 14 days of EOL. We evaluated the recent transformation of EOL chemotherapy and targeted therapy practices nationally. METHODS:In patients older than 65 years of age who died as a result of breast (n = 19,887), lung (n = 79,613), colorectal (n = 29,844), or prostate (n = 17,910) cancer between 2007 and 2013, we evaluated the guideline-benchmarked measure of chemotherapy use within 14 days of EOL in SEER-Medicare. Comparison outcomes were nonbenchmarked measures of chemotherapy and targeted therapy across time points within 6 months of EOL. Cochran-Armitage test was used to evaluate temporal trends. Multilevel logistic models and intraclass correlation coefficient was used to evaluate variation in EOL chemotherapy use at the physician level. RESULTS:From 2007 to 2013, chemotherapy within 14 days of EOL declined from 6.7% to 4.9% of patients (Ptrend < .001; ? = -1.8%). Similar declines occurred for chemotherapy within 1 month (Ptrend < .001; ? = -1.8%) and 2 months (Ptrend < .001; ? = -1.3%) of EOL. In contrast, chemotherapy within 4 to 6 months of EOL rose (Ptrend ? .04; ? = 0.7% to 1.7%), and 43.0% of all patients received chemotherapy within 6 months of EOL. Frequency of targeted therapy use across all time points within 6 months of EOL was stable to marginally rising from 2007 to 2013 (Ptrend = .09 to .82; ? = -0.2% to 1.8%); overall, 1.2% received targeted therapy within 14 days and 3.6% within 1 month of EOL. By 2013, 13.2% of patients received any targeted therapy within 6 months of EOL. In a multilevel model, 5.19% (intraclass correlation coefficient) of variation in 14-day EOL chemotherapy was attributed to the physician level. CONCLUSION:With national benchmarking, chemotherapy within 14 days of EOL successfully declined to less than 5%, with comprehensive benchmark uptake by physicians. Results may inform current strategies to help to achieve high-value EOL oncology practice.
Project description:INTRODUCTION:Although evidence has been provided on the associations between psychosocial work exposures and morbidity outcomes in the literature, knowledge appears much more sparse on mortality outcomes. The objective of STRESSJEM is to explore the prospective associations between psychosocial work exposures and mortality outcomes among the national French working population. In this paper, we describe the study protocol, study population, data sources, method for exposure assessment, data analysis and future plans. METHODS AND ANALYSIS:Data sources will include: the data from the national SUMER survey from DARES on the evaluation of psychosocial work exposures and the data from the COSMOP programme from Santé publique France linking job history (DADS data from INSEE) and mortality according to causes of death (data from the national death registry, INSERM-CépiDc). A sample of 1 511 456 individuals will form the studied prospective cohort for which data are available on both job history and mortality over the period 1976-2002. Psychosocial work exposures will be imputed via a job-exposure matrix using three job title variables that are available in both the SUMER and COSMOP data sets. Our objectives will be to study the associations between various psychosocial work exposures and mortality outcomes. Psychosocial work exposures will include the job strain model factors as well as other psychosocial work factors. Various measures of exposure over time will be used. All-cause and cause-specific mortality will be studied. ETHICS AND DISSEMINATION:Both the SUMER survey and the COSMOP programme have been approved by French ethics committees. Dissemination of the study results will include a series of international peer-reviewed papers and at least one paper in French. The results will be presented in national and international conferences. This project will offer a unique opportunity to explore mortality outcomes in association with psychosocial work exposures in a large national representative sample of the working population.
Project description:Patients with hematologic malignancies are thought to receive more aggressive end-of-life (EOL) care and have suboptimal hospice use compared with patients with solid tumors, but descriptions of EOL outcomes from comprehensive cohorts have been lacking. We used the population-based Surveillance, Epidemiology, and End Results-Medicare dataset to describe hospice use and indicators of aggressive EOL care among Medicare beneficiaries who died of hematologic malignancies in 2008-2015. Overall, 56.5% of decedents used hospice services for median 9 days (interquartile range, 3-27), 33.0% died in an acute hospital setting, 36.8% had an intensive care unit (ICU) admission in the last 30 days of life, and 13.3% received chemotherapy within the last 14 days of life. Hospice use was associated with 96% lower probability of inpatient death (adjusted risk ratio [aRR], 0.038; 95% confidence interval [CI], 0.035-0.042), 44% lower probability of an ICU stay in the last 30 days of life (aRR, 0.56; 95% CI, 0.54-0.57), and 62% decrease in chemotherapy use in the last 14 days of life (aRR, 0.38; 95% CI, 0.35-0.41). Hospice enrollees spent on average 41% fewer days as inpatient during the last month of life (adjusted means ratio, 0.59; 95% CI, 0.57-0.60) and had 38% lower mean Medicare spending in the last month of life (adjusted means ratio, 0.62; 95% CI, 0.61-0.64). These associations were consistent across histologic subgroups. In conclusion, EOL care quality outcomes and hospice enrollment were suboptimal among older decedents with hematologic cancers, but hospice use was associated with a consistent decrease in aggressive care at EOL.
Project description:<h4>Background</h4>End-of-life (EOL) decisions are a serious ethical dilemma and are frequently carried out in intensive care units (ICUs). The aim of this systematic review was to investigated the different approaches used in ICUs and reported in randomized controlled trials (RCTs) to address EOL decisions and compare the impact of these different strategies regarding potential bias and mortality estimates.<h4>Methods</h4>We identified relevant RCTs published in the past 15 years via PubMed, EMBASE, and CINAHL. In addition, we searched The Cochrane Library and checked registries, including ClinicalTrials.gov to assess concordance between declared and published outcomes. Among the journals we screened were the 3 ICU specialty journals and the four general medicine journals with the highest impact factor. Only RCTs were selected in which in-ICU mortality was the primary or secondary outcome. The primary outcome was information regarding EOL decisions, and the secondary outcome was how EOL decisions were treated in the study analysis.<h4>Results</h4>A total of 178 relevant trials were identified. The details regarding the methodological aspects resulting from EOL decisions were reported in only 62 articles (35%). The manner in which EOL decisions were considered in the study analysis was very heterogeneous, often leading to a high risk of bias.<h4>Conclusion</h4>There is a heterogeneity regarding the management of data on EOL decisions in randomized control trials with mortality endpoints. Recommendations or rules are required regarding the inclusion of patients with potential EOL decisions in RCT analyses and how to manage such decisions from a methodological point of view.<h4>Trial registration</h4>PROSPERO website (CRD42013005724).