554325 research-article2014
PMJ0010.1177/0269216314554325Palliative MedicineMahar et al.
Original Article
Predictors of hospital stay and home care services use: A population-based, retrospective cohort study in stage IV gastric cancer
Palliative Medicine 2015, Vol. 29(2) 147–156 © The Author(s) 2014 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0269216314554325 pmj.sagepub.com
Alyson L Mahar1, Natalie G Coburn2,3,4, Raymond Viola1,5 and Ana P Johnson1,4,6
Abstract Background: Home care services use has been proposed as a means of reducing costs in palliative care by decreasing hospital stay without impacting quality of clinical care; however, little is known about utilization of these services in the time following a terminal cancer diagnosis. Aim: To examine disease, patient and healthcare system predictors of hospital stay, and home care services use in metastatic gastric cancer patients. Design: This is a population-based, retrospective cohort study. Chart review and administrative data were linked, using a 26-month time horizon to collect health services data. Participants: All patients diagnosed with metastatic gastric cancer in the province of Ontario between 2005 and 2008 were included in the study (n = 1433). Results: Age, comorbidity, tumor location, and burden of metastatic disease were identified as predictors of hospital stay and receipt of home care services. Individuals who received home care services spent fewer days in hospital than individuals who did not (relative risk: 0.44; 95% confidence interval: 0.38–0.51). Patients who interacted with a high-volume oncology specialist had shorter cumulative hospital stay (relative risk: 0.62; 95% confidence interval: 0.54–0.71) and were less likely to receive home care services (relative risk: 0.80; 95% confidence interval: 0.72–0.88) than those who did not. Conclusion: Examining how differences in hospital stay and home care services use impact clinical outcomes and how policies may reduce costs to the healthcare system is necessary.
Keywords Home care services, hospitalization, palliative care, gastric neoplasms, health services research, terminal care
What is already known on this subject? •• Hospitalizations are a major cost driver to healthcare providers. •• Receipt of home care service at end of life may be one strategy to reduce the financial burden to the system while providing optimal patient care.
1Department
of Public Health Sciences, Queen’s University, Kingston,
ON, Canada of Surgical Oncology, Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON, Canada 3Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, ON, Canada 4Institute for Clinical Evaluative Sciences, Toronto, ON, Canada 2Division
5Division
of Palliative Medicine, Department of Medicine, Queen’s University, Kingston, ON, Canada 6Centre for Health Services and Policy Research, Queen’s University, Kingston, ON, Canada Corresponding author: Natalie G Coburn, Division of Surgical Oncology, Odette Cancer Centre, Sunnybrook Health Sciences Centre, Suite T2-11, 2075 Bayview Ave, Toronto, ON M4N 3M5, Canada. Email: [email protected]
Downloaded from pmj.sagepub.com at SIMON FRASER LIBRARY on June 11, 2015
148
Palliative Medicine 29(2)
What this study adds? •• Evidence that time survived from diagnosis is implicated in the use of acute inpatient and home-based care. •• Patients who receive home care services have significantly reduced rates of hospital stay in the metastatic cancer setting. •• Characteristics of the disease and healthcare system may drive utilization. Implications for practice, theory, or policy •• Home care services should be optimized to prevent unnecessary hospitalizations at the end of life. •• Future research needs to evaluate the effectiveness of palliation during hospitalization and home care services at the end of life. •• Research needs to assess the barriers to accessing home care for palliative care patients in need.
Introduction The costs of hospitalization make up almost one quarter of healthcare spending worldwide,1–3 and cancer care represents 6%–10% of expenditures.4,5 Hospitalizations related to cancer have the longest average length of stay,2 and the average per patient system cost of an oncologyrelated hospital stay is estimated to be CAD250,000 in Canada.4 The contribution of hospitalizations at the end of life to the overall costs of care is also substantial.4,6 On average, hospitalizations are reported to represent approximately 40% of per patient terminal care costs,4 and cancer ranks among the leading costs of inpatient deaths in the United States.6 An increased number of days spent in hospital at the end of life in cancer care is a potential indicator of uncontrolled symptoms, inadequate palliation, or inappropriately aggressive cancer management.7–13 This is also an increased financial burden to the healthcare system, compared to the use of other sites for the provision of palliative care services.14,15 In many instances, emergency room visits precede hospital admissions at end of life.10 Many of these emergency room visits are potentially avoidable, suggesting that palliative home care services could replace these hospitalizations.8 Specialized palliative care services, including the use of home care services, have been advocated as an effective and efficient means of addressing patient symptoms and needs at end of life, while decreasing the number and duration of hospitalizations.11,16,17 The use of home care services to offset hospital admissions for individuals dying of cancer may be a means of improving patient care and reducing costs together.15,16 Gastric cancer has been identified as one of the most expensive cancers to treat, and 64%–75% of estimated expenditures are related to hospitalizations.18–20 Analytic work has not been performed to investigate specific measures of healthcare resource utilization that may explain why the cost of care for these populations is so high. Metastatic gastric cancer patients survive a median of 6 months following diagnosis,21 and present an opportunity to study early junctures for intervention with home care. Therefore, the objective of this research was to
identify disease, patient and healthcare system predictors of cumulative hospital days and home care services use, and to explore the relationship between the two healthcare services within a high cost cohort of cancer patients, those with metastatic gastric cancer.
Methods The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for observational cohort studies was followed.22 Health Science Research Ethics Board approval for this project was obtained through the Sunnybrook Health Sciences Centre, Toronto, ON, Canada, and Queen’s University, Kingston, ON, Canada, and at the 116 institutions where chart reviews were performed. This was also approved by the Privacy Office at the Institute for Clinical Evaluative Sciences (ICES), which holds prescribed entity status under Ontario’s privacy law and has the authority to collect and use provincial and national administrative healthcare data without individual consent.
Study design and population This was a population-based, retrospective cohort study, using administrative data linked to in-depth chart information. This study was performed from the healthcare system perspective. Patients with a registered diagnosis of gastric cancer, aged 18–99 years, in the Ontario Cancer Registry (OCR) between 1 April 2005 and 31 March 2008 were included. In Ontario, all health services are provided publicly and free of charge from a single payer system run by the provincial government. Each individual is provided a unique Ontario Health Insurance Plan (OHIP) number. Ontario healthcare spending is allocated to 14 Local Health Integrated Networks (LHINs). Patients were included if they had a valid OHIP number, a valid ICES key number (IKN), an IKN traceable in the Registered Persons Database (RPDB, described
Downloaded from pmj.sagepub.com at SIMON FRASER LIBRARY on June 11, 2015
149
Mahar et al. below), a confirmed diagnosis of gastric adenocarcinoma, and evidence of metastatic disease (6th Edition Union for International Cancer Control/American Joint Committee on Cancer (UICC/AJCC) stage IV). Patients were excluded if they were missing metastatic stage information (MX); had a primary tumor location identified in the upper, middle, or entire esophagus; or were missing information regarding their LHIN of residence.
Data sources The chart review data were linked to the Canadian Institute of Health Information–Discharge Abstract Database (CIHI-DAD), the Ontario Home Care Database (HCD), OHIP, and the RPDB. The chart review was conducted between November 2009 and November 2011 and provided clinical disease data, including burden of metastatic disease and stage. The data were linked to administrative data holdings in January 2012. CIHI-DAD contains data on provincial acute care hospitalizations.23 The HCD was used to identify receipt of home care services.23 OHIP contains provincial physician billing claims and was used to determine visits to medical oncologists, radiation oncologists, and surgeons and surgical procedures.23 The RPDB provided patient-level demographic information, including date of death.23
Outcomes Data on cumulative hospital stay and receipt of home care services were captured from the date of diagnosis for 2 years or until death. Hospital stay was defined as the cumulative number of days spent hospitalized (for any reason, non-consecutive) over the study time horizon. The number of days, including all admission and discharge dates within that time period was summed for each unique admission. Home care services use (yes/no) was defined as at least one registered visit to a patient’s home in the records of the Community Care Access Centre, the sole provider of public home-based healthcare in Ontario. Visits could be for a variety of reasons, such as physiotherapy, speech pathology, or personal care, and by different caregivers, such as nurses, personal support workers, or physiotherapists.
Disease, patient, and healthcare system characteristics Disease characteristics were identified using endoscopy, pathology, and radiology reports located in the chart. Burden of metastatic disease was defined as having either 1 or more than 1 site; primary tumor location was assigned to the gastroesophageal junction or the proximal, middle, distal, or entire stomach if known; otherwise, it was labeled unknown. Patient characteristics
were identified using administrative data, including age, sex, residence, rurality (rural or urban, according to established cut-points using the Rurality Index of Ontario24), socioeconomic status (based on data from Statistics Canada, linking postal codes to median community-level income, categorized from lowest to highest quintiles25), and Johns Hopkin’s resource utilization bands (lowest to highest quintiles of use, based on Aggregated Diagnostic Groups of comorbidity26). Health system characteristics measured included the receipt of care (consult or procedural billing codes) from a highvolume gastric cancer specialist, receipt of gastrectomy, and LHIN of residence. Gastric cancer patient volume for each physician within each specialty (surgeon, medical oncologist, radiation oncologist) was categorized into volume quartiles by specialty. High volume was defined using the cut-offs for the highest quartile: a high-volume surgeon performed an average of at least 3.5 gastrectomies/year, a high-volume medical oncologist saw on average of at least 6.7 gastric cancer patients/year, and a high-volume radiation oncologist saw on average of at least 15.8 gastric cancer patients/year. Receipt of care or a consult from at least one of the three high-volume specialists constituted a “yes” for this variable. Physician procedural billing codes in OHIP were used to identify which patients received a gastrectomy (partial or total, with or without a multivisceral resection). Patients were assigned one of 14 LHINs using their postal codes.
Statistical analysis Median survival was calculated following Kaplan–Meier methodology. Incidence rates of hospital stay per 100 days alive and their 95% confidence intervals (CIs) were calculated using the Wald method. Proportions of receipt of home care services and their 95% CIs were calculated for each disease, patient, and healthcare system characteristics. Two separate analyses were performed.27 The first was an investigation of factors influencing the length of cumulative hospital stay. For this analysis, non-parametric analysis of variance (ANOVA) tests compared mean length of hospital stay (days), and negative binomial regression was used to model independent predictors of hospital stay (days). In the second analysis, factors influencing the receipt of home care service use were evaluated. For this analysis, chi-square tests of independence were used to compare proportions of patients with home care service use, and modified Poisson regression was used to model the receipt of home care services.28 For both modeling strategies, an offset variable was incorporated (log of the number of days survived) to account for time at risk. The measure of effect for both outcomes was relative risk (RR) and 95% CIs. Two-sided hypothesis testing was performed, and an alpha of 0.05 was used to establish
Downloaded from pmj.sagepub.com at SIMON FRASER LIBRARY on June 11, 2015
150
Palliative Medicine 29(2)
Results A total of 2516 patients with gastric cancer were identified. In all, 25 patients were excluded because of missing chart review and/or staging data, 44 because they had cancers that were not adenocarcinomas, 989 because they had stage I–III disease, 23 with tumor located in the upper or mid-esophagus, and 2 because they were missing information about their LHIN of residence. This left 1433 eligible cohort members. The mean age was 67.5 years (range: 20–97 years), and almost two-thirds were male (n = 934). Just over half of the cohort had one site of metastatic disease only (n = 747). Median survival for the cohort was 6.2 months (95% CI: 5.6–6.9 months); 16% of patients survived longer than 2 years.
Cumulative hospital stay Overall, 95% were hospitalized at some point during the study period and the mean, cumulative hospital stay per patient was 30.2 days, with an interquartile range of 10– 37 days. A strong inverse association was documented on univariate analysis between the number of days survived from the date of diagnosis and the proportion of time alive spent in hospital (Figure 1). The mean cumulative hospital stay per patient per 100 days alive is displayed in Table 1 by disease, patient, and healthcare system characteristics. Patients with the highest average rates of days spent in hospital per 100 days alive from the time of diagnosis included those with tumors in the entire stomach (12.8 days), those living in rural areas (12.2 days), those who did not receive a gastrectomy (11.6 days), and those who resided in LHIN 12 (15.0 days) and 13 (14.3 days), the two most northern and remote regions in the province. On multivariate analysis, primary tumor location, resource utilization band, consultation or treatment from a high-volume gastric cancer specialist, receipt of a gastrectomy, and receipt of home care services were all independent predictors of cumulative hospital stay (Table 2). Compared to patients with tumors located in the distal stomach, patients with tumors at the gastroesophageal junction (RR: 0.82; 95% CI: 0.67–1.01) and proximal stomach (RR: 0.80; 95% CI: 068–0.94) had lower relative rates of cumulative hospital days, while those with tumors in the middle and entire stomach trended toward increased relative rates of hospital days (p = 0.0021). Health system characteristics played an important role in the reduction of hospital stay. While region of residence was not independently associated with the rate of hospital stay, patients who received a consult or care from a high-volume gastric cancer specialist had a 38% lower rate of hospital stay
90 % of days alive spent in hospital
statistical significance. All analyses were performed using SAS 9.2 Copyright 2008 (Cary, NC, USA).
80 70 60 50 40 30 20 10 0
Days Alive
Figure 1. Relationship between time survived from the date of diagnosis to death/censoring and accumulation of hospital days.
(RR: 0.62; 95% CI: 0.54–0.71) compared to those who did not receive this care. Patients who received a gastrectomy had a 33% reduced rate of hospital stay relative to those who did not receive this intervention (RR: 0.67; 95% CI: 0.59–0.76). Receipt of home care services was associated with the largest significant reduction in rate of hospital stay—a 56% reduction compared with those who did not receive at least one home care service visit.
Receipt of home care services Overall, 78% of patients received at least one home care service visit. A significant relationship between the number of days survived from diagnosis and the receipt of at least one home care service visit was also identified (Figure 2)—the longer the period of survival following diagnosis, the more likely a patient was to receive home care service (p
PMJ0010.1177/0269216314554325Palliative MedicineMahar et al.
Original Article
Predictors of hospital stay and home care services use: A population-based, retrospective cohort study in stage IV gastric cancer
Palliative Medicine 2015, Vol. 29(2) 147–156 © The Author(s) 2014 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0269216314554325 pmj.sagepub.com
Alyson L Mahar1, Natalie G Coburn2,3,4, Raymond Viola1,5 and Ana P Johnson1,4,6
Abstract Background: Home care services use has been proposed as a means of reducing costs in palliative care by decreasing hospital stay without impacting quality of clinical care; however, little is known about utilization of these services in the time following a terminal cancer diagnosis. Aim: To examine disease, patient and healthcare system predictors of hospital stay, and home care services use in metastatic gastric cancer patients. Design: This is a population-based, retrospective cohort study. Chart review and administrative data were linked, using a 26-month time horizon to collect health services data. Participants: All patients diagnosed with metastatic gastric cancer in the province of Ontario between 2005 and 2008 were included in the study (n = 1433). Results: Age, comorbidity, tumor location, and burden of metastatic disease were identified as predictors of hospital stay and receipt of home care services. Individuals who received home care services spent fewer days in hospital than individuals who did not (relative risk: 0.44; 95% confidence interval: 0.38–0.51). Patients who interacted with a high-volume oncology specialist had shorter cumulative hospital stay (relative risk: 0.62; 95% confidence interval: 0.54–0.71) and were less likely to receive home care services (relative risk: 0.80; 95% confidence interval: 0.72–0.88) than those who did not. Conclusion: Examining how differences in hospital stay and home care services use impact clinical outcomes and how policies may reduce costs to the healthcare system is necessary.
Keywords Home care services, hospitalization, palliative care, gastric neoplasms, health services research, terminal care
What is already known on this subject? •• Hospitalizations are a major cost driver to healthcare providers. •• Receipt of home care service at end of life may be one strategy to reduce the financial burden to the system while providing optimal patient care.
1Department
of Public Health Sciences, Queen’s University, Kingston,
ON, Canada of Surgical Oncology, Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON, Canada 3Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, ON, Canada 4Institute for Clinical Evaluative Sciences, Toronto, ON, Canada 2Division
5Division
of Palliative Medicine, Department of Medicine, Queen’s University, Kingston, ON, Canada 6Centre for Health Services and Policy Research, Queen’s University, Kingston, ON, Canada Corresponding author: Natalie G Coburn, Division of Surgical Oncology, Odette Cancer Centre, Sunnybrook Health Sciences Centre, Suite T2-11, 2075 Bayview Ave, Toronto, ON M4N 3M5, Canada. Email: [email protected]
Downloaded from pmj.sagepub.com at SIMON FRASER LIBRARY on June 11, 2015
148
Palliative Medicine 29(2)
What this study adds? •• Evidence that time survived from diagnosis is implicated in the use of acute inpatient and home-based care. •• Patients who receive home care services have significantly reduced rates of hospital stay in the metastatic cancer setting. •• Characteristics of the disease and healthcare system may drive utilization. Implications for practice, theory, or policy •• Home care services should be optimized to prevent unnecessary hospitalizations at the end of life. •• Future research needs to evaluate the effectiveness of palliation during hospitalization and home care services at the end of life. •• Research needs to assess the barriers to accessing home care for palliative care patients in need.
Introduction The costs of hospitalization make up almost one quarter of healthcare spending worldwide,1–3 and cancer care represents 6%–10% of expenditures.4,5 Hospitalizations related to cancer have the longest average length of stay,2 and the average per patient system cost of an oncologyrelated hospital stay is estimated to be CAD250,000 in Canada.4 The contribution of hospitalizations at the end of life to the overall costs of care is also substantial.4,6 On average, hospitalizations are reported to represent approximately 40% of per patient terminal care costs,4 and cancer ranks among the leading costs of inpatient deaths in the United States.6 An increased number of days spent in hospital at the end of life in cancer care is a potential indicator of uncontrolled symptoms, inadequate palliation, or inappropriately aggressive cancer management.7–13 This is also an increased financial burden to the healthcare system, compared to the use of other sites for the provision of palliative care services.14,15 In many instances, emergency room visits precede hospital admissions at end of life.10 Many of these emergency room visits are potentially avoidable, suggesting that palliative home care services could replace these hospitalizations.8 Specialized palliative care services, including the use of home care services, have been advocated as an effective and efficient means of addressing patient symptoms and needs at end of life, while decreasing the number and duration of hospitalizations.11,16,17 The use of home care services to offset hospital admissions for individuals dying of cancer may be a means of improving patient care and reducing costs together.15,16 Gastric cancer has been identified as one of the most expensive cancers to treat, and 64%–75% of estimated expenditures are related to hospitalizations.18–20 Analytic work has not been performed to investigate specific measures of healthcare resource utilization that may explain why the cost of care for these populations is so high. Metastatic gastric cancer patients survive a median of 6 months following diagnosis,21 and present an opportunity to study early junctures for intervention with home care. Therefore, the objective of this research was to
identify disease, patient and healthcare system predictors of cumulative hospital days and home care services use, and to explore the relationship between the two healthcare services within a high cost cohort of cancer patients, those with metastatic gastric cancer.
Methods The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for observational cohort studies was followed.22 Health Science Research Ethics Board approval for this project was obtained through the Sunnybrook Health Sciences Centre, Toronto, ON, Canada, and Queen’s University, Kingston, ON, Canada, and at the 116 institutions where chart reviews were performed. This was also approved by the Privacy Office at the Institute for Clinical Evaluative Sciences (ICES), which holds prescribed entity status under Ontario’s privacy law and has the authority to collect and use provincial and national administrative healthcare data without individual consent.
Study design and population This was a population-based, retrospective cohort study, using administrative data linked to in-depth chart information. This study was performed from the healthcare system perspective. Patients with a registered diagnosis of gastric cancer, aged 18–99 years, in the Ontario Cancer Registry (OCR) between 1 April 2005 and 31 March 2008 were included. In Ontario, all health services are provided publicly and free of charge from a single payer system run by the provincial government. Each individual is provided a unique Ontario Health Insurance Plan (OHIP) number. Ontario healthcare spending is allocated to 14 Local Health Integrated Networks (LHINs). Patients were included if they had a valid OHIP number, a valid ICES key number (IKN), an IKN traceable in the Registered Persons Database (RPDB, described
Downloaded from pmj.sagepub.com at SIMON FRASER LIBRARY on June 11, 2015
149
Mahar et al. below), a confirmed diagnosis of gastric adenocarcinoma, and evidence of metastatic disease (6th Edition Union for International Cancer Control/American Joint Committee on Cancer (UICC/AJCC) stage IV). Patients were excluded if they were missing metastatic stage information (MX); had a primary tumor location identified in the upper, middle, or entire esophagus; or were missing information regarding their LHIN of residence.
Data sources The chart review data were linked to the Canadian Institute of Health Information–Discharge Abstract Database (CIHI-DAD), the Ontario Home Care Database (HCD), OHIP, and the RPDB. The chart review was conducted between November 2009 and November 2011 and provided clinical disease data, including burden of metastatic disease and stage. The data were linked to administrative data holdings in January 2012. CIHI-DAD contains data on provincial acute care hospitalizations.23 The HCD was used to identify receipt of home care services.23 OHIP contains provincial physician billing claims and was used to determine visits to medical oncologists, radiation oncologists, and surgeons and surgical procedures.23 The RPDB provided patient-level demographic information, including date of death.23
Outcomes Data on cumulative hospital stay and receipt of home care services were captured from the date of diagnosis for 2 years or until death. Hospital stay was defined as the cumulative number of days spent hospitalized (for any reason, non-consecutive) over the study time horizon. The number of days, including all admission and discharge dates within that time period was summed for each unique admission. Home care services use (yes/no) was defined as at least one registered visit to a patient’s home in the records of the Community Care Access Centre, the sole provider of public home-based healthcare in Ontario. Visits could be for a variety of reasons, such as physiotherapy, speech pathology, or personal care, and by different caregivers, such as nurses, personal support workers, or physiotherapists.
Disease, patient, and healthcare system characteristics Disease characteristics were identified using endoscopy, pathology, and radiology reports located in the chart. Burden of metastatic disease was defined as having either 1 or more than 1 site; primary tumor location was assigned to the gastroesophageal junction or the proximal, middle, distal, or entire stomach if known; otherwise, it was labeled unknown. Patient characteristics
were identified using administrative data, including age, sex, residence, rurality (rural or urban, according to established cut-points using the Rurality Index of Ontario24), socioeconomic status (based on data from Statistics Canada, linking postal codes to median community-level income, categorized from lowest to highest quintiles25), and Johns Hopkin’s resource utilization bands (lowest to highest quintiles of use, based on Aggregated Diagnostic Groups of comorbidity26). Health system characteristics measured included the receipt of care (consult or procedural billing codes) from a highvolume gastric cancer specialist, receipt of gastrectomy, and LHIN of residence. Gastric cancer patient volume for each physician within each specialty (surgeon, medical oncologist, radiation oncologist) was categorized into volume quartiles by specialty. High volume was defined using the cut-offs for the highest quartile: a high-volume surgeon performed an average of at least 3.5 gastrectomies/year, a high-volume medical oncologist saw on average of at least 6.7 gastric cancer patients/year, and a high-volume radiation oncologist saw on average of at least 15.8 gastric cancer patients/year. Receipt of care or a consult from at least one of the three high-volume specialists constituted a “yes” for this variable. Physician procedural billing codes in OHIP were used to identify which patients received a gastrectomy (partial or total, with or without a multivisceral resection). Patients were assigned one of 14 LHINs using their postal codes.
Statistical analysis Median survival was calculated following Kaplan–Meier methodology. Incidence rates of hospital stay per 100 days alive and their 95% confidence intervals (CIs) were calculated using the Wald method. Proportions of receipt of home care services and their 95% CIs were calculated for each disease, patient, and healthcare system characteristics. Two separate analyses were performed.27 The first was an investigation of factors influencing the length of cumulative hospital stay. For this analysis, non-parametric analysis of variance (ANOVA) tests compared mean length of hospital stay (days), and negative binomial regression was used to model independent predictors of hospital stay (days). In the second analysis, factors influencing the receipt of home care service use were evaluated. For this analysis, chi-square tests of independence were used to compare proportions of patients with home care service use, and modified Poisson regression was used to model the receipt of home care services.28 For both modeling strategies, an offset variable was incorporated (log of the number of days survived) to account for time at risk. The measure of effect for both outcomes was relative risk (RR) and 95% CIs. Two-sided hypothesis testing was performed, and an alpha of 0.05 was used to establish
Downloaded from pmj.sagepub.com at SIMON FRASER LIBRARY on June 11, 2015
150
Palliative Medicine 29(2)
Results A total of 2516 patients with gastric cancer were identified. In all, 25 patients were excluded because of missing chart review and/or staging data, 44 because they had cancers that were not adenocarcinomas, 989 because they had stage I–III disease, 23 with tumor located in the upper or mid-esophagus, and 2 because they were missing information about their LHIN of residence. This left 1433 eligible cohort members. The mean age was 67.5 years (range: 20–97 years), and almost two-thirds were male (n = 934). Just over half of the cohort had one site of metastatic disease only (n = 747). Median survival for the cohort was 6.2 months (95% CI: 5.6–6.9 months); 16% of patients survived longer than 2 years.
Cumulative hospital stay Overall, 95% were hospitalized at some point during the study period and the mean, cumulative hospital stay per patient was 30.2 days, with an interquartile range of 10– 37 days. A strong inverse association was documented on univariate analysis between the number of days survived from the date of diagnosis and the proportion of time alive spent in hospital (Figure 1). The mean cumulative hospital stay per patient per 100 days alive is displayed in Table 1 by disease, patient, and healthcare system characteristics. Patients with the highest average rates of days spent in hospital per 100 days alive from the time of diagnosis included those with tumors in the entire stomach (12.8 days), those living in rural areas (12.2 days), those who did not receive a gastrectomy (11.6 days), and those who resided in LHIN 12 (15.0 days) and 13 (14.3 days), the two most northern and remote regions in the province. On multivariate analysis, primary tumor location, resource utilization band, consultation or treatment from a high-volume gastric cancer specialist, receipt of a gastrectomy, and receipt of home care services were all independent predictors of cumulative hospital stay (Table 2). Compared to patients with tumors located in the distal stomach, patients with tumors at the gastroesophageal junction (RR: 0.82; 95% CI: 0.67–1.01) and proximal stomach (RR: 0.80; 95% CI: 068–0.94) had lower relative rates of cumulative hospital days, while those with tumors in the middle and entire stomach trended toward increased relative rates of hospital days (p = 0.0021). Health system characteristics played an important role in the reduction of hospital stay. While region of residence was not independently associated with the rate of hospital stay, patients who received a consult or care from a high-volume gastric cancer specialist had a 38% lower rate of hospital stay
90 % of days alive spent in hospital
statistical significance. All analyses were performed using SAS 9.2 Copyright 2008 (Cary, NC, USA).
80 70 60 50 40 30 20 10 0
Days Alive
Figure 1. Relationship between time survived from the date of diagnosis to death/censoring and accumulation of hospital days.
(RR: 0.62; 95% CI: 0.54–0.71) compared to those who did not receive this care. Patients who received a gastrectomy had a 33% reduced rate of hospital stay relative to those who did not receive this intervention (RR: 0.67; 95% CI: 0.59–0.76). Receipt of home care services was associated with the largest significant reduction in rate of hospital stay—a 56% reduction compared with those who did not receive at least one home care service visit.
Receipt of home care services Overall, 78% of patients received at least one home care service visit. A significant relationship between the number of days survived from diagnosis and the receipt of at least one home care service visit was also identified (Figure 2)—the longer the period of survival following diagnosis, the more likely a patient was to receive home care service (p
