Abstract
Purpose:
To examine quality indicators of end-of-life (EOL) care among privately insured people with cancer in Brazil.
Methods:
We evaluated medical records linked to health insurance databank to study consecutive patients who died of cancer. We collected information about demographics, cancer type, and quality indicators of EOL care including emergency department (ED) visits, intensive care unit (ICU) admissions, chemotherapy use, medical imaging utilization, blood transfusions, home care support, days of inpatient care, and hospital deaths.
Results:
We included 865 patients in the study. In the last 30 days of life, 62% visited the ED, 33% were admitted to the ICU, 24% received blood transfusions, and 51% underwent medical imaging. Only 1% had home care support in the last 60 days of life, and 29% used chemotherapy in the last 14 days of life. Patients had an average of 8 days of inpatient care and 52% died in the hospital. Patients with advanced cancer who used chemotherapy were more likely to visit the ED (78% vs 59%; P < .001), undergo medical imaging (67% vs 51%; P < .001), and die in the hospital (73% vs 50%; P = .03) than patients who did not use chemotherapy. In the multivariate analysis, chemotherapy use near death and advanced cancer were associated with ED visits and ICU admissions, respectively (odds ratio >1).
Conclusion:
Our study suggests that privately insured people with cancer receive poor quality EOL care in Brazil. Further research is needed to assess the impact of improvements in palliative care provision in this population.
Introduction
Quality assessment of end-of-life (EOL) care in limited-resource countries has been largely focused on the analysis of macro-indicators such as the existence of palliative care policies, morphine consumption per capita, and the proportion of medical schools that include palliative care education in undergraduate curricula. 1 Macro-indicators are important to measure the development of palliative care at national levels and facilitate comparisons between countries, but such indicators are less useful to provide practical insights into the quality of EOL care delivered to a selected group of people in the community setting.
Health administrative databases offer a valuable source of quality indicators that are relevant for people with cancer, relatively inexpensive, and highly congruent with medical records. 2 -5 Previous studies based on administrative data have suggested that people with cancer often experience negative outcomes at the EOL with high rates of emergency department (ED) visits, intensive care unit (ICU) admissions, and use of chemotherapy. 6,7 However, marked variations in the outcome rates have been demonstrated across distinct geographic regions, a fact that can be partially explained by differences in the availability of palliative care services. 8,9
According to the World Health Organization, a significant proportion of people with cancer die without receiving palliative care in Brazil, where the provision of palliative care services is isolated and insufficient to cover all demands of the population. 10 The lack of access to palliative care in fragile health-care systems poses an enormous burden to individual, family, and local community. 11 To date, little is known about the performance measures in health-care systems, where palliative care is poorly integrated into the standard oncology practice. Given the growing awareness to reduce health inequalities, it is of paramount importance that health-care systems routinely determine its measures, identify priorities, and achievable targets. The purpose of this study was to examine administrative quality indicators of EOL care among privately insured people with cancer in a community setting in Brazil and perform a multivariate analysis in this population to identify predictors of acute care utilization.
Methods
Study Design
This retrospective study used a methodology that was similar to a previous analysis we performed among patients with hematologic malignancies. 12 We evaluated consecutive patients covered by a private health insurance plan who died as a result of cancer (solid malignancies) between January 01, 2014, and June 30, 2018, in the state of Rio de Janeiro, Brazil. We included patients treated in 14 community hospitals located in the metropolitan area of Rio de Janeiro and surrounding cities (Angra dos Reis, Araruama, Barra do Piraí, Campos, Friburgo, Nova Iguaçu, Resende, Nova Iguaçu, and Volta Redonda). We excluded patients if they were less than 18 years, if they had an insurance number that does not match the database, if they died within 30 days after the diagnosis of cancer, and if they died outside the state of Rio de Janeiro. All personal identifying information was removed from the data set before analysis. The local ethical committee (94006618.3.0000.5274) approved this study.
Data Source and Outcomes
We collected data derived from electronic medical records linked to health insurance administrative databank (Radar System, Unimed Federação, Rio de Janeiro), which contains baseline demographic information (age, gender, underlying diagnosis using the International Classification of Diseases, Tenth Revision, cancer staging), medical coding, and billing claims (outpatient and inpatient claims) for all types of medical services provided to beneficiaries. Advanced cancer was defined as locally advanced, metastatic, or recurrent disease.
To cover different aspects of EOL care, we analyzed 8 quality indicators of EOL care as follows: (1) any ED visit in the last 30 days of life, (2) any ICU admission in the last 30 days of life, (3) any packed red cell or platelet transfusion in the last 30 days of life, (4) chemotherapy use in the last 14 days of life, (5) any computed tomography (CT) scan or magnetic resonance imaging (MRI) in the last 30 days of life, (6) number of days of inpatient care in the last 30 days of life, (7) any home care visit (doctor or nurse) during the last 60 days of life, and (8) hospital death.
Statistical Analysis
We performed a descriptive analysis where the categorical variables were expressed as absolute numbers and percentages, and the continuous variables were expressed as mean and standard deviation. To better understand the role of chemotherapy near death, we also compared outcomes and health-care resource utilization among patients with advanced cancer who used chemotherapy versus those who did not use chemotherapy in the last 14 days of life. Student t test and Kruskal-Wallis test were used to compare mean values when the variables had normal or nonparametric distribution, respectively. A χ2 test was used to compare frequencies of categorical data between both the groups.
Finally, we performed a multivariate logistic regression model to evaluate whether age >60 years, gender, advanced disease, and the use of chemotherapy in the last 14 days of life were independently associated with acute care utilization. To do so, we arbitrarily used ED visits and ICU admissions as markers of acute care utilization. MedCalc version 19.0.4 package was used to perform the statistical analysis.
Results
Table 1 shows the patient characteristics and quality indicators of EOL care. The final study sample consisted of 865 patients who died from cancer. The mean age was 63 years; 39% (n = 340) were male and 61% (n = 525) were female. The 6 most common primary cancer sites were breast (n = 217, 26%), lung (n = 109, 13%), colon (n = 107, 12%), urological (n = 102, 11%), gynecological (n = 82, 9%), and pancreas (n = 43, 5%). During the last 30 days of life, 536 (62%) visited the ED, 288 (33%) were admitted to the ICU, 204 (24%) had blood transfusions, and 439 (51%) underwent a CT scan or MRI. Only 1% of patients (n = 13) had home care visits in the last 60 days of life, 29% (n = 248) used chemotherapy in the last 14 days of life. Patients received an average of 8.8 days of inpatient care, and 52% (n = 450) died in the hospital.
Characteristics of Patients and Quality Indicators of End-of-Life Care.
Abbreviations: CT, computed tomography; EOL, end of life; MRI, magnetic resonance imaging; SD, standard deviation.
Chemotherapy in Patients With Advanced Cancer
Of the 865 patients, 609 (70%) had advanced cancer. A subgroup analysis showed that 193 (31%) patients with advanced cancer used chemotherapy in the last 14 days of life. These patients were more likely to visit the ED (78%, n = 150 vs 59%, n = 244; P < .001), to undergo a CT scan or MRI (67%, n = 130 vs 51%, n = 211; P < .001), and to die in the hospital (73%, n = 141, vs 50%, n = 209; P = .03) compared to those who did not use chemotherapy. We found similar rates in terms of ICU admissions (39%, n = 75 vs 36%, n = 151; P = .60), blood transfusions (28%, n = 55 vs 25%, n = 104; P = .41), and days of inpatient care (7.3 ± 11.8 vs 9.8 ± 18.7; P = .08) between patients who used chemotherapy and those who did not use chemotherapy in the last 14 days of life (Table 2).
Chemotherapy in the Last 14 Days of Life in Patients With Advanced Cancer.
Abbreviations: CT, computed tomography; EOL, end of life; MRI, magnetic resonance imaging; Chemo, chemotherapy; SD, standard deviation.
Factors Associated With ED Visits and ICU Admissions
Table 3 summarizes the multivariate logistic regression model. Age >60 years and biological sex were not found to be independent factors associated with ED visits and ICU admissions. The use of chemotherapy in the last 14 days was independently associated with ED visit during the last 30 days of life (odds ratio: 2.64, 95% confidence interval [CI]: 1.88-3.70), while advanced disease was independently associated with ICU admission during the last 30 days of life (odds ratio: 1.44, 95% CI: 1.05-1.98).
Multivariate Logistic Regression Analysis.a
Abbreviations: CI, confidence interval; ED, emergency department; ICU, intensive care unit.
a Variables entered in the multivariate logistic regression analysis include age >60 years, gender, chemotherapy in the last 14 days of life, and advanced cancer.
Discussion
We explored an administrative database to examine quality indicators of EOL care and health-care utilization among a subgroup of patients with cancer in the community setting in Brazil. This study revealed performance measures in a health-care system with an isolated provision of palliative care services. At the time of this writing, outpatient palliative care clinics, hospital-based palliative care teams, and home care support were not easily available for people with cancer in the community setting in the state of Rio de Janeiro. Brazil has only 0.48 palliative care services per million people, which is approximately 4 times lower than the average for the Latin American region. 13 To our best knowledge, benchmark rates for quality indicators of EOL care have been proposed by studies conducted in academic centers located in high-income countries, where palliative care practice is much better organized than in Brazil. Therefore, our findings may be useful to other limited-resource countries, as it represents a more realistic scenario for performance comparisons and policymaking in the community setting.
We found that more than 60% of patients visited the ED in the last 30 days of life. Given the fact that only 1% of patients received home care visits in the last 60 days before death, these 2 outcomes were probably intrinsic linked, reflecting a combination of unmatched patients’ needs and the lack of home care support for this population. In the private health-care system, the delivery of home care services requires prior authorization from insurance payers who usually take into account the presence of artificial nutrition, intravenous hydration, prolonged course of antibiotics, and complex wound dressings, rather than typical oncology palliative care needs such as symptom management and psychosocial support. These criteria often exclude patients with cancer from receiving home care services, possibly driving unnecessary ED visits. There is evidence that some of these visits could be avoided and fall within the scope of palliative care programs. 14,15 An expert consensus has included the number of ED visits near death as a major indicator of integration between palliative care and oncology programs in acute care hospitals. 16 In the context of this study, we advocate that integrating palliative care into the primary health-care system to overcome the shortage of palliative care specialists and scarce funding would be an affordable starting point to change this grim scenario. Primary care physicians are well positioned to offer longitudinal care and identify unmatched patient’s needs early in the illness trajectory. This integration could have a positive impact on other quality indicators of EOL care, possibly reducing the number of days of inpatient care and hospital deaths. Health-care systems that rely on primary care models to deliver palliative care have more continuity of care, low rates of care setting transitions, decreased time spent in hospitals, and less acute resource utilization among dying people with cancer. 17 -19
Although there is a trend toward more aggressive cancer care, 6,7 there may be other explanations for the fact that more than 30% of patients were admitted to the ICU in the last 30 days of life. Recent advances in the management of the critically ill have resulted in better survival rates and more flexible ICU admission policies for people with cancer in Brazil. 20,21 Moreover, we analyzed information from the private health-care system, where there is an oversupply of ICU beds. In the geographic region of this study, the number of ICU beds per capita is 5 times higher than that in the United Kingdom. 22 An increase in the number of ICU beds available may interfere in triage decisions and reduce the incentive for dying patients to stay out of the ICU. 23 It is also possible that care fragmentation has influenced some ICU admissions. Discontinuity of care from the outpatient to the inpatient setting has been associated with more ICU stays among terminally ill patients with cancer. 24 In the community setting, while oncologists coordinate outpatient care, hospitalists are in charge of inpatient care. As only a minority of patients have documented advanced directives, hospitalists who are unfamiliar with patients’ wishes tend to offer life-prolonging interventions in the critical care unit rather than comfort measures in the ward. 25 A recent survey in Brazil found that in the absence of documented patient’s consent, oncologists, intensivists, and prosecutors have different views about withholding and withdrawing life-sustaining therapies in terminally ill cancer, and almost half of the physicians are afraid that their decisions regarding limitations of medical treatment could have legal implications. 26
Despite the lack of information about the patient’s performance status in our database, this study has demonstrated that 1 in 3 patients with advanced cancer received chemotherapy in the last 14 days of life. This rate was much higher than some other estimates. 27 -31 We also found that these patients were more likely to visit the ED, to undergo medical imaging, and to die in the hospital. As a rule, chemotherapy in advanced cancer should be used in patients with good performance status to improve quality of life and prolong survival. However, as death approaches, the “palliative” benefits of chemotherapy may not occur even among those with good performance status. 32 In times of financial constraints, our findings raise concerns about the sustainability of limited-resource health-care systems and the reimbursement of services in oncology. In Brazil, the current payment model is mainly fee-for-service, does not reimburse the provision of palliative care services, and rewards private oncology providers with financial incentives for prescribing expensive drugs. Perhaps, this payment model creates barriers to discontinue chemotherapy at the end of life, which is considered one of the main practices to improve the quality of cancer care and reduce costs. 33 A recent study suggests that hospitalized people with cancer who died in for-profit hospitals without palliative care units were more likely to receive chemotherapy near death than in university hospitals. 30
We found that more than half of the patients had a CT scan or MRI in the last 30 days of life. It has been found that in the ED, patients with cancer have twice the chance of undergoing CT scans than patients with noncancer. 34 Medical imaging requests in advanced cancer stages have increased more rapidly compared to early cancer stages. 35 We recognize that by ordering CT scans and MRIs, physicians can detect tumor spreading, inform prognosis, and initiate palliative care discussions. However, the overuse of medical imaging in dying patients with cancer increase costs and postpone hospice referrals. 36,37 Our findings also highlight the importance of medical data integration between outpatient and inpatient providers. In the community setting, medical records are sometimes paper-based and not well integrated resulting in poor information sharing and duplicate medical imaging. Because people with cancer frequently have multiple encounters with health-care providers in different locations, improvements in data exchange to provide clinicians a prompt and complete access to patients’ medical records may reduce unnecessary medical procedures. 38
There are some limitations to this study. First, data were originally collected for administrative purposes, not for research. Information about patient’s preferences and goals of care was not available, precluding definite conclusions about the link between health-care resource utilization and quality of care delivered. For example, blood transfusions may not indicate aggressive cancer care, as packed red cells and platelets can be used for palliative purposes. Second, we analyzed the utilization of medical services in the private health-care system, which takes care of approximately 25% of the Brazilian population. Brazil has 2 coexisting health-care systems (public and private) that do not share medical data with each other. Therefore, we cannot exclude the use of medical services in the public system. As the public system is usually overcrowded and does not cover the demand of the population, privately insured people usually prefer to use the private system. Third, our findings may not represent the country as a whole. Brazil is a continental country with drastic differences in the availability of health-care services across different regions, and social groups, limiting the generalizability of our results.
In summary, our study suggests that privately insured people with cancer experience high rates of ED visits, ICU admissions, chemotherapy use, utilization of medical imaging, and hospital deaths in Brazil. Patients with advanced cancer who used chemotherapy in the last 14 days of life were more likely to visit the ED, undergo medical imaging, and die in the hospital. In the multivariate analysis, the use of chemotherapy near death and cancer in advanced stages were associated with ED visits and ICU admissions within 30 days of life, respectively. We hope that our findings will be of use to health-care providers and policymakers who face the challenges of improving EOL care for people with cancer in fragile health-care systems. Future prospective studies should aim to identify palliative care interventions to decrease the proportion of patients experiencing these negative outcomes.
Footnotes
Authors’ Note
All authors contributed to conception and design, provision of study materials, data analysis and interpretation, manuscript writing, and final approval of manuscript. L.G.L.S. and R.V.G. contributed to Administrative Support and collection and assembly of data.
Acknowledgments
The authors indebted to Claudio Bittencourt and Luzia Santos for their administrative support.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
