Abstract
Health resource utilization (HRU) and associated factors of high cost are not well understood in myeloproliferative neoplasms (MPNs). In this population-based, retrospective matched-cohort study, we used administrative health databases of Ontario, Canada to measure treatment costs and HRU for patients with MPN from 2004 to 2016 and compared them to matched controls. In 7130 patients with MPN [essential thrombocythemia (ET) = 3481; polycythemia vera (PV) = 2618; myelofibrosis (MF) = 1031], the mean annualized treatment costs were $16,646 for ET (controls, $7070); $16,360 for PV (controls, $7293); and $25,863 for MF (controls, $7386). Out of the total costs, the largest expenditure was on acute hospital care (ET: 57%, PV: 57%, MF: 66%). Older age (≥65), male gender, patients not seen by a specialist, and greater comorbidity burden were independent predictors of higher costs (p < 0.05). In addition, history of venous thrombosis in patients with ET and PV was associated with significantly higher treatment costs (p < 0.05).
Introduction
The cost of treating cancers has become unsustainable for patients, families, care providers, and society at large [Citation1,Citation2]. The American Society of Clinical Oncology (ASCO), in a guidance statement, recommended policy decisions to address the underlying factors contributing to the increased costs [Citation3]. Hematological malignancies are the second most expensive cancers to treat after breast cancers [Citation4,Citation5]. A number of economic evaluations have informed treatment availability and affordability in multiple myeloma, acute leukemia, and lymphoma [Citation6–13]. For myeloproliferative neoplasms (MPNs), the cost of treatment is two to six times higher compared to general population using US medicare databases [Citation14]. But the reasons for this high health resource use have not been investigated.
The clinical and sociodemographic characteristics of high cost users of the health systems include older age, multiple comorbid conditions, and those who are socially disadvantaged such as having low income or education [Citation15–18]. Cancer care is more complex and additional factors such as access to specialist consultation [Citation18], volume and caseload severity of the facility, institutional affiliations such as cancer care organizations, research alliances, academic status, and specialist-generalist collaborations can impact health resource utilization (HRU) [Citation19]. Patients with myeloproliferative neoplasms are older and have a high burden of arterial and venous thrombosis and comorbidities. Moreover, their clinical needs are often complex and heterogeneous because patients need both acute care services to manage exacerbation of underlying comorbidities as well as long-term cancer facilities to manage MPN-associated symptoms and transfusion requirements. The care for MPN patients has remained fragmented due to their protracted course and lack of integrated care pathways, unlike other cancers [Citation20]. Disaggregating the effects of various factors associated with costs of care will highlight clinical needs and assist policymakers in planning cost-effective measures.
In this population-based costing study our aims were to (1) measure all direct medical costs and health resource use incurred by patients with three classical MPNs [essential thrombocythemia (ET), polycythemia vera (PV), and myelofibrosis (MF)] in comparison to matched controls and (2) measure the effect of clinical and health systems factors on the cost of treatment.
Patient and methods
Study design and data sources
We conducted a population-based, retrospective matched-cohort study, using the provincial health databases of Ontario’s single payer universal health system (Supplementary Table1). These databases located at the ICES, Toronto, Canada were linked using unique encoded identifiers and analyzed at ICES. ICES is an independent, nonprofit research institute whose legal status under Ontario’s health information privacy law allows it to collect and analyze health care and demographic data, without consent, for health system evaluation and improvement. The use of data in this project was authorized under section 45 of Ontario’s Personal Health Information Protection Act, which does not require review by a Research Ethics Board.
Cohort definition
Cases were patients aged ≥18 years, who were residents of Ontario and had a diagnosis of myeloproliferative neoplasm (Essential Thrombocythemia, ET; Polycythemia Vera, PV; and Myelofibrosis, MF) in the Ontario Cancer Registry, from January 1, 2004, to December 31, 2016 (). We defined MPNs by International Classification of Diseases for Oncology, third edition (ICD-O-3) topography code C42, and morphology codes 99623 (ET), 99503 (PV), and 99603/99613/99751 (MF) [Citation21]. The index date was the date of diagnosis of MPN. Controls were individuals from the general population of Ontario, without a diagnosis of myeloproliferative neoplasm. We matched each case to four controls on age (±3 years), sex, geographical location, and neighborhood income quintile. After matching, controls inherited the index date from their matched cases. We followed individuals from the index date to maximum study follow-up (December 31, 2017), or death.
Baseline patient and health services parameters
We determined baseline parameters during the 2 years prior to each person’s MPN diagnosis (diagnostic codes in Supplementary Tables 2–4). Rurality and neighborhood income quintile were determined using Statistics Canada’s postal code conversion file [Citation22]. We obtained prevalence of specific comorbidities, and arterial and venous thrombosis. The baseline comorbidity score was calculated as a weighted sum of the 32 John Hopkins Aggregated Diagnosis Groups (ADG, The Johns Hopkins ACG® System, version 10) [Citation23,Citation24]. The ACG System groups every ICD-9 and ICD-10 diagnosis code assigned to a health care user into one of 32 different ADGs based on five clinical and expected utilization criteria: duration of the condition, severity of the condition, diagnostic certainty, etiology of the condition, and specialty care involvement. We excluded the MPN diagnosis from the ADG score. The resource utilization bands (RUB) were calculated using the John’s Hopkins ACG System [Citation24]. They are a simplified ranking system of overall morbidity level instead of grouping by type of illness, so that individuals who are expected to use the same level of resources are grouped together, even if they have very different illnesses with different epidemiological patterns. Time to the first consultation with a specialist (hematologist or oncologist) was calculated starting 6 months prior to diagnosis using physician claims database. Survival data were obtained from the OCR and vital statistics.
Exposures and outcomes
The main exposure was a diagnosis of any of the three classical MPNs. The main outcomes were HRU and costs of treatment (data sources in Supplementary Table 5). Health resource utilization included the following parameters: average length of stay for (1) in-patient hospitalization, (2) intensive care unit, (3) long-term care, and average number of (4) outpatient visits (including specialists and general practitioners), (5) emergency room visits, and (6) home care visits. Direct medical costs of treatment was measured by accounting for all healthcare-related costs used by cases and their matched controls, and presented as mean cost of care per-person-patient-year of follow-up [Citation25]. These healthcare-related costs included costs for in-hospital care (in-patient, outpatient, emergency room), physician and nonphysician services, laboratory services, radiation therapy, chemotherapy, chronic and rehabilitation care (complex continuing care, long-term care, rehabilitation), and home care. The mean cost of hospitalization, emergency room visits, and same-day surgery for a particular year was estimated using the Resource Intensity Weights method [Citation25]. We converted all costs into 2018 Canadian dollars ($1.00CDN = $0.76USD), using the consumer price index of healthcare.
Variables
We collected information on several clinical and health service-related factors: age, sex, residence, neighborhood income quintile, ADG score, prior thrombosis, time to first consult with a specialist, and the time period of diagnosis.
Statistical analysis
We used ordinary least squares (OLS) linear regression analysis to measure the association of predictors with the annualized cost of treatment [Citation26]. Using OLS, we computed the rate ratios for the relative effect (the percent change in the covariate group) compared to the reference group within the same covariate. For descriptive statistics, we presented mean (standard deviation) and median (interquartile range) for continuous variables, and frequencies and percentages for categorical variables. We used Wilcoxon rank sum test to compare continuous variables, and Fisher exact test for categorical variables, where appropriate. For all estimates, we considered a p-value ≤0.05 statistically significant. We performed all statistical analyses using SAS 9.4 (SAS Institute, Cary, NC) and R (R Foundation, Vienna, Austria).
Results
Baseline characteristics
Between 2004 and 2016, we identified 7130 patients diagnosed with MPNs (ET = 3481, PV = 2618, MF = 1031) (). For cases versus controls, the mean duration of follow-up was 3.9 versus 4.3 years for ET; 3.9 versus 4.2 years for PV; and 3.2 versus 4.9 years for MF. The proportion of individuals who developed thrombosis (arterial or venous) at any time within 2 years before the index event date was significantly higher among cases versus controls (ET, 10.4 versus 3.8%; PV, 12.6 versus 4.0%; MF, 10.1 versus 4.4%) (p < 0.001). Arterial thrombosis was more common than venous thrombosis among all three types of MPNs and their controls. Comparing cases versus controls, the median ADG score 2 years before the index event was significantly higher for cases (ET, 18 versus 6; PV, 18 versus 6; MF, 22 versus 8) (p < 0.001). A higher proportion of cases were in high or very high RUB compared to controls (ET 55.7 versus 29.1%; PV 54 versus 28.7%; MF 62.9 versus 30.8%) (p < 0.001). More than 70% cases were diagnosed after 2012, and most patients were seen by specialist within 6 months after diagnosis (ET, 76%, PV 83%, MF, 90%).
Table 1. Comparison of baseline characteristics of MPN cases and matched controls.
Health resource utilization
Patients with all three MPNs had a higher use of health resources than their matched controls, except long-term care (). The largest difference between cases and controls was noted for the duration of acute care services (hospitalization and ER services). The mean duration of hospitalizations including intensive care unit in cases compared to controls was 5.5 versus 1.6 days in ET; 5.2 versus 1.7 days in PV and 9.8 versus 1.8 days in MF. ER visits in ET, PV, and MF were higher as compared to their controls (ET, 1.1 versus 0.5; PV 1.2 versus 0.5; MF 1.8 versus 0.6). It was difficult to separate MPN-related versus non-related reasons for use of acute care services. Given the high burden of vascular events in these patients that could contribute to high use of acute care services, we specifically compared the incidence of thrombotic events in cases versus controls (Supplementary Table 6). MPN cases had significantly high incidence of thrombotic event after the index event as compared to controls. Ambulatory care was equally divided between general practitioners and specialists for both ET and PV (mean number of general practitioner versus specialist visits: 6.8 versus 6.9 for ET, 6.5 versus 7.5 for PV), but in MF, the patients were more frequently seen by the specialists (12.7 versus 7.4). Use of home care services was two times higher in cases as compared to controls (ET, 14.6 versus 7.6; PV 12.5 versus 8.2; MF 17.1 versus 8.1). ET and PV patients showed similar long-term care use as compared to their controls (ET, 10.2 versus 11.6; PV, 10.7 versus 11.3), but in MF patients, it was two times lower than their controls (MF, 4.0 versus 8.3).
Figure 1. Cohort creation. Cases with myeloproliferative neoplasms (MPN) were included from Ontario cancer registry (OCR) using ICD-0-3 diagnosis codes. Non-MPN matched controls were included from the registered persons database for the general population of Ontario, in a ratio for 1:4. Index date was the date of diagnosis of MPN from OCR; controls inherited the index date from their cases. Patients with nonclassical MPN diagnosis and CML were excluded.
Figure 2. Comparison of healthcare resource utilization between patients with myeloproliferative neoplasms and matched controls. Frequency of utilization of each type of health care resource is shown for essential thrombocythemia (A), polycythemia vera (B), and myelofibrosis (C). Data represent average number of days per-person-year for in-patient hospitalization and long-term care stay, and average number of outpatient visits per-person-year for specialist and general practitioner (GP), emergency room (ER), and home care.
Costs of treatment
The mean per-person-year direct medical costs of treatment for patients with ET was $16,646 (2.3 times ET controls, $7070), for PV $16,360 (2.2 times PV controls, $7293) and for MF $25,863 (3.5 times MF controls, $7386) (Supplementary Table 7). The comparison of drivers of costs of treatment for cases versus controls is shown in the Sankey diagram (). It represents the flow of expenditure for cases and controls with the width and darkness of the band proportional to the cost incurred for that health resource. In all MPN, the primary cost driver was acute hospital care and highest cost was incurred by MF (MF $17,186 versus $3544 in controls; ET $9508 versus $2,986 in controls; PV, $9377 versus $3,183 in controls). Because patients with MPN could incur higher costs of care when their illness is terminal, we divided the cost of treatment into three different phases to understand the trajectory of health care spending. During the initial phase of illness (3 months before the index event to 6 months after), the per 30 person-days costs of care were (cases versus controls): ET $1542 versus $122; PV $1136 versus $112; MF 1145 versus $126). The costs incurred by MPN cases during the initial phase were almost 10–12 times that incurred by controls. During continuation phase (6 months after index event to 6 months prior to death or till last date of follow-up when alive), the per 30 person-days cost of treatment were (cases versus controls): (ET $291 versus $119; PV $300 versus $128; MF 466 versus $137). During the terminal phase (6 months prior to death), the per 30 person days cost of care was (cases versus controls): ET $5842 versus $3085; PV $5029 versus $3146; MF $6177 versus $3361). Thus, the maximum difference in cost of care was seen around the period of diagnosis. Cost of physician services and chronic and rehabilitation care were the next drivers of costs of treatment. MF incurred higher cost for physician services ($4747 versus $1744 for controls) than PV ($2845 versus $1401 for controls) and ET ($2806 versus $1384 for controls). For chronic and rehabilitation care, ET ($2859 versus $1937 for controls) and PV ($2790 versus $1911 for controls) incurred higher cost than MF ($1744 versus $1551 for controls). Home care services incurred lesser costs of treatment than other cost categories (ET $1,136 versus $535 for controls; PV $965 versus $573 for controls; MF, $1566 versus $578 for controls).
Figure 3. Comparison of healthcare costs in patients with myeloproliferative neoplasms and matched controls. The Sankey diagram compares the flow of treatment costs for cases and controls toward various health care services from left to right side. The width and darkness of the band are proportional to the total costs of treatment incurred. The broader and darker the band, the higher the cost for the health care service.
Association of clinical and health services factors with costs of treatment
Among ET (, Supplementary Figure 1) and PV (, Supplementary Figure 2) patients, the highest costs of treatment were noted in individuals who were never seen by a specialist (ET, RR 2.3, 95% CI 2.0–2.6; PV, RR 3.0, 95% CI 2.5–3.6). Older patients (age >65 years) (ET, RR 2.2, 95% CI 2.0–2.4; PV, RR 2.4, 95% CI 2.1–2.7), male gender (ET, RR 1.3, 95% CI 1.2–1.4; PV, RR 1.1, 95% CI 1.0–1.2) and patients with history of venous thrombosis (ET, RR 1.5, 95% CI 1.1–2.0; PV, RR 1.5, 95% CI 1.1–1.9) also incurred higher costs of treatment. Further, comorbidity burden as represented by the ADG score was associated with increased costs of treatment (ET, RR 1.03, 95% CI 1.03–1.03; PV, RR 1.03, 95% CI 1.03–1.04). In MF (, Supplementary Figure 3), patients who were never seen by a specialist (RR 1.6, 95% CI 1.2–2.2), older patients (RR 1.3, 95% CI 1.1–1.5) and patients with higher comorbidity burden (RR 1.03, 95% CI 1.03–1.04) had increased costs of treatment. Prior history of arterial thrombosis in ET and PV or any thrombosis in MF was not associated with higher costs of treatment. No impact of neighborhood income quintile was noted on the costs of treatment. Patients seeing specialists ≥6 months after their diagnosis incurred lesser costs of treatment because of lesser comorbidity burden (Supplementary Table 8).
Figure 4. Multivariable analysis showing association of clinical and health services factors with costs of treatment in essential thrombocythemia. Forest plot represents the rate ratios from the multivariable OLSs linear regression analysis to determine the association between the clinical and health services factors and costs of treatment (mean per-person-year in 2018 Canadian dollar) for essential thrombocythemia. The rate ratio was computed as the percent relative effect (the percent change in the covariate group) as compared to the reference group within the same covariate. The reference group is regarded as having 100% of the cost and the other covariate groups expressed relative to the reference group. For all estimates, p-value ≤0.05 was considered statistically significant.
Figure 5. Multivariable analysis showing association of clinical and health services factors with costs of treatment in polycythemia vera. Forest plot represents the rate ratios from the multivariable ordinary least squares linear regression analysis to determine the association between the clinical and health services factors and costs of treatment (mean per-person-year in 2018 Canadian dollar) for polycythemia vera. The rate ratio was computed as the percent relative effect (the percent change in the covariate group) as compared to the reference group within the same covariate. The reference group is regarded as having 100% of the cost and the other covariate groups expressed relative to the reference group. For all estimates, p-value ≤0.05 was considered statistically significant.
Figure 6. Multivariable analysis showing association of clinical and health services factors with costs of treatment in myelofibrosis. Forest plot represents the rate ratios from the multivariable ordinary least squares linear regression analysis to determine the association between the clinical and health services factors and costs of treatment (mean per-person-year in 2018 Canadian dollar) for myelofibrosis. The rate ratio was computed as the percent relative effect (the percent change in the covariate group) as compared to the reference group within the same covariate. The reference group is regarded as having 100% of the cost and the other covariate groups expressed relative to the reference group. For all estimates, p-value ≤0.05 was considered statistically significant.
Discussion
We conducted this population-based retrospective cohort study to measure the economic burden, HRU and factors associated with high cost of treatment in patients with MPNs. Patients with ET, PV, and MF incurred significantly higher cost of treatment as well as HRU compared to general population when matched for age, gender, neighborhood income quintile, and residence. These costs are largely driven by acute hospital care. Factors such as older age, male gender, venous thrombosis prior to diagnosis, increasing comorbidity burden and not seeing a specialist were associated with significantly higher treatment costs. Our findings provide an understanding of reasons for higher health resource use in patients with MPN and inform future strategies to reduce the economic burden.
The literature on economic burden of MPN is sparse. Using claims data, Mehta et al. found that patients with MPN incurred two-five times higher costs of treatment compared to matched controls [Citation26]. Their cohort was younger (median age 60 years) because of nature of eligibility for the commercial health insurance plans. The study design was cross-sectional thereby largely capturing the costs related to initial diagnostic testing and acute complications. In contrast, our population-based cohort included all MPN patients, irrespective of their age and health status. Moreover, baseline characteristics were determined in the 2 years prior to diagnosis. This allowed comprehensive capturing all health-related events and associated costs. Despite the dissimilarities in study design, both studies showed that highest proportion of resource is used for acute hospital care. Common reasons for hospital admissions in MPN patients include thrombotic or cardiovascular complications, which are potentially preventable.
In the existing literature, there is limited information on the reasons for high healthcare resource use in patients with MPN. In our multivariate model, ET and PV patients with venous thrombosis at or prior to their diagnosis incurred higher costs of treatment after adjusting for factors such as age and comorbidities. These higher future costs of care could be partly explained by recurrence of thrombosis seen in about 20–30% MPN patients [Citation27,Citation28]. Published studies have shown that patients with recurrent VTE utilize 2–3-fold higher inpatient and ER services, threefold higher outpatient medical services and more expensive to manage [Citation29,Citation30]. Further, there was markedly higher morbidity burden in MPN patients even before their diagnosis. Previous estimates of comorbidities were based on single-center studies without a control population [Citation30,Citation31]. The ADG score, a validated morbidity index, was associated with significantly higher costs of treatment in the adjusted model. Age and socioeconomic status confound comorbidities and treatment costs and therefore, were controlled in our model [Citation32].Therefore, increasing comorbidities and thrombotic complications are the key clinical factors associated excess health care expenditure in MPN. Older patients and male gender were associated with higher costs, probably due to greater complexity and severity of illness associated with them [Citation32–34].
An important finding in our study was the two to three times higher treatment costs incurred by the patients who never saw a specialist. Possibly, the coordination of care between specialist and primary healthcare provider could result in optimal management of comorbidities and early identification for disease progression. In chronic illnesses, cohesive care has shown to reduce hospitalizations and decrease healthcare expenditures [Citation35,Citation36].
Our findings have certain limitations including biases inherent to a retrospective design, inability to include indirect, and pharmacy costs (drug reimbursement is restricted to individuals >65 years in Ontario), and inability to adjust for disease characteristics such as blood counts and molecular genetics. Besides, there was a likely reporting bias before year 2012 as majority of patients were reported to cancer registry thereafter. This is due to increased awareness of MPN following their inclusion in the WHO 2008 classification, besides the wider availability of molecular testing.
The strengths of our study are unselected population-based matched-cohort, large sample size, validated databases, extensive follow-up, and the use of baseline ADG score for risk adjustment in the OLS model, a validated tool to predict health care resource utilization, mortality and long-term hospitalization [Citation37]. Our study used an OLS regression model to predict future costs. Although cost and utilization data are usually right skewed and heteroscedastic, this was not a problem as our sample size was very large. Indeed, OLS regression has been shown to provide unbiased estimates while predicting future costs [Citation38,Citation39].
In conclusion, our study shows substantial health care burden in MPNs and generate concerning clinical factors for higher treatment costs. Reducing hospitalizations from comorbidities and thrombosis and improving coordination of care through integrated care pathways could reduce the financial burden.
Author contributions
Dr Gupta had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Bankar, Gupta.
Acquisition, analysis, or interpretation of data: Bankar, Zhao, Iqbal, Coxford, Cheung, Mozessohn, Earle, Gupta.
Drafting of the manuscript: Bankar, Gupta.
Critical revision of the manuscript for important intellectual content: all authors.
Statistical analysis: Zhao, Iqbal, Coxford.
Obtained funding: Gupta
Administrative, technical, or material support: Bankar, Iqbal, Zhao, Coxford, Earle, Gupta.
Study supervision: Gupta, Earle.
Disclaimer
This study made use of deidentified data from the Institute for Clinical Evaluative Sciences (ICES) Data Repository, which is managed by the Institute for Clinical Evaluative Sciences with support from its funders and partners: Canada’s Strategy for Patient-Oriented Research (SPOR), the Ontario SPOR Support Unit, the Canadian Institutes of Health Research and the Government of Ontario. The opinions, results and conclusions reported are those of the authors. No endorsement by ICES or any of its funders or partners is intended or should be inferred. Parts of this material are based on data and information compiled and provided by the Canadian Institute for Health Information. However, the analyses, conclusions, opinions, and statements expressed herein are those of the authors, and not necessarily those of CIHI.
Supplemental Material
Download PDF (543 KB)Acknowledgments
Parts of this study are based on data and information compiled and provided by the MOHLTC, Cancer Care Ontario, and the Canadian Institute for Health Information. The analyses, conclusions, opinions, and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred. We thank IMS Brogan Inc. for use of their Drug Information Database. The ORGD database was obtained from Service Ontario.
Disclosure statement
VG received an honorarium, clinical trial funding through the institution and served on an advisory board of Novartis, Celgene, and Sierra Oncology.
Additional information
Funding
References
- Yabroff KR, Lund J, Kepka D, et al. Economic burden of cancer in the United States: estimates, projections, and future research. Cancer Epidemiol Biomarkers Prev. 2011;20(10):2006–2014.
- Bray F, Ferlay J, Soerjomataram I, et al. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68(6):394–424.
- Meropol NJ, Schrag D, Smith TJ, et al. American Society of Clinical Oncology guidance statement: the cost of cancer care. JCO. 2009;27(23):3868–3874.
- Gabriela Dieguez F, Ferro C, Rotter CD. The cost burden of blood cancer care. [Internet]. New York: Leukemia and Lymphoma Society; 2018 [cited November 15, 2019]. 40 p. Available from: www.lls.org/sites/default/files/Milliman%20study%20cost%20burden%20of%20 blood%20cancer%20care.pdf
- Burns R, Leal J, Sullivan R, et al. Economic burden of malignant blood disorders across Europe: a population-based cost analysis. Lancet Haematol. 2016;3(8):e362–e70.
- Zeidan AM, Mahmoud D, Kucmin-Bemelmans IT, et al. Economic burden associated with acute myeloid leukemia treatment. Expert Rev Hematol. 2016;9(1):79–89.
- Foster T, Miller JD, Boye ME, et al. Economic burden of follicular non-Hodgkin’s lymphoma. PharmacoEconomics. 2009;27(8):657–679.
- Tolley K, Morgan G, Cartwright R, et al. Economic aspects of non-hodgkin's lymphoma [Internet]. London: Office of Health Economics; 1998 [cited December 11, 2019]. 66 p. Available from: https://www.ohe.org/publications/economic-aspects-non-hodgkins-lymphoma
- Reis A, Ihle P, Paulus U, et al. Cost of illness of malignant lymphoma in Germany. Eur J Cancer Care. 2006;15(4):379–385.
- MacEwan JP, Batt K, Yin W, et al. Economic burden of multiple myeloma among patients in successive lines of therapy in the United States. Leukemia Lymphoma. 2018;59(4):941–949.
- Teitelbaum A, Ba-Mancini A, Huang H, et al. Health care costs and resource utilization, including patient burden, associated with novel-agent-based treatment versus other therapies for multiple myeloma: findings using real-world claims data. Oncologist. 2013;18(1):37–45.
- Fonseca R, Abouzaid S, Bonafede M, et al. Trends in overall survival and costs of multiple myeloma, 2000–2014. Leukemia. 2017;31(9):1915–1921.
- Kutikova L, Bowman L, Chang S, et al. Medical costs associated with non-Hodgkin’s lymphoma in the United States during the first two years of treatment. Leukemia Lymphoma. 2006;47(8):1535–1544.
- Mehta J, Wang H, Fryzek JP, et al. Health resource utilization and cost associated with myeloproliferative neoplasms in a large United States health plan. Leukemia Lymphoma. 2014;55(10):2368–2374.
- Improving the management of care for high-cost medicaid patients. Health Affairs. 2007;26(6):1643–1654.
- Lubitz J, Cai L, Kramarow E, et al. Health, life expectancy, and health care spending among the elderly. N Engl J Med. 2003;349(11):1048–1055.
- Lehnert T, Heider D, Leicht H, et al. Review: health care utilization and costs of elderly persons with multiple chronic conditions. Med Care Res Rev. 2011;68(4):387–420.
- Pitchforth E, Russell E, Van der Pol M. Access to specialist cancer care: is it equitable? Br J Cancer. 2002;87(11):1221–1226.
- Burns LR, Muller RW. Hospital‐physician collaboration: landscape of economic integration and impact on clinical integration. Milbank Quarterly. 2008;86(3):375–434.
- Schmidt I, Thor J, Davidson T, et al. The national program on standardized cancer care pathways in Sweden: observations and findings half way through. Health Policy. 2018;122(9):945–948.
- Percy FA, Jack C, Shanmugaratnam A, et al. International classification of diseases for oncology 2000.
- Statistics Canada. Postal CodeOM Conversion File (PCCF), Reference Guide [Internet]. Ottawa, ON: Ministry of Industry; 2017 [cited December 11, 2019]. 23 p. Catalogue no. 92–154-G. Available from: https://www.canadapost.ca/cpc/assets/cpc/uploads/files/marketing/2017-postal-code-conversion-file-reference-guide-en.pdf
- Austin PC, van Walraven C, Wodchis WP, et al. Using the Johns Hopkins Aggregated Diagnosis Groups (ADGs) to predict mortality in a general adult population cohort in Ontario, Canada. Med Care. 2011;49(10):932–939.
- Weiner JP, Abrams C, Bodycombe D. The Johns Hopkins ACG case-mix system version 6.0 release notes. Baltimore, MD: Johns Hopkins Bloomberg School of Public Health; 2003.
- Wodchis WP, Bushmeneva K, Nikitovic M, et al. Guidelines on Person-Level Costing Using Administrative Databases in Ontario [Internet]. The Health System Performance Research Network; 2013 [cited December 15, 2019]. 70 p. Available from: http://www.sky9games.com/hsprn/uploads/files/Guidelines_on_PersonLevel_Costing_May_2013.pdf
- Gregori D, Petrinco M, Bo S, et al. Regression models for analyzing costs and their determinants in health care: an introductory review. Int J Qual Health Care. 2011;23(3):331–341.
- De Stefano V, for the GIMEMA CMD-Working Party, Za T, Rossi E, et al. Recurrent thrombosis in patients with polycythemia vera and essential thrombocythemia: incidence, risk factors, and effect of treatments. Haematologica. 2008;93(3):372–380.
- Elliott MA, Pardanani A, Lasho TL, et al. Thrombosis in myelofibrosis: prior thrombosis is the only predictive factor and most venous events are provoked. Haematologica. 2010;95(10):1788–1791.
- Spyropoulos AC, Lin J. Direct medical costs of venous thromboembolism and subsequent hospital readmission rates: an administrative claims analysis from 30 managed care organizations. JMCP. 2007;13(6):475–486.
- Lefebvre P, Laliberté F, Nutescu EA, et al. All-cause and disease-related health care costs associated with recurrent venous thromboembolism. Thromb Haemost. 2013;110(12):1288–1297.
- Scherber RM, Senyak Z, Dueck AC, et al. The impact of medical comorbidities in MPN-related fatigue. Blood. 2014;124(21):1862–1862.
- Barnett K, Mercer SW, Norbury M, et al. Epidemiology of multimorbidity and implications for health care, research, and medical education: a cross-sectional study. Lancet. 2012;380(9836):37–43.
- Passamonti F, Cervantes F, Vannucchi AM, et al. Dynamic International Prognostic Scoring System (DIPSS) predicts progression to acute myeloid leukemia in primary myelofibrosis. Blood. 2010;116(15):2857–2858.
- Karantanos T, Chaturvedi S, Spivak JL, et al. Independent association of male sex with presentation and clinical outcomes in myeloproliferative neoplasms. Blood. 2018;132(Supplement 1):1768–1768. 2018;
- Vannucchi AM, Harrison CN. Emerging treatments for classical myeloproliferative neoplasms. Blood. 2017;129(6):693–703.
- Peikes D, Chen A, Schore J, et al. Effects of care coordination on hospitalization, quality of care, and health care expenditures among medicare beneficiaries: 15 randomized trials. JAMA. 2009;301(6):603–618.
- Austin PC, Walraven C. The mortality risk score and the ADG score: two points-based scoring systems for the Johns Hopkins aggregated diagnosis groups to predict mortality in a general adult population cohort in Ontario, Canada. Med Care. 2011;49(10):940–947.
- Barber J, Thompson S. Multiple regression of cost data: use of generalised linear models. J Health Serv Res Policy. 2004;9(4):197–204.
- Moran JL, Solomon PJ, Peisach AR, et al. New models for old questions: generalized linear models for cost prediction. J Eval Clin Pract. 2007;13(3):381–389.