High-Cost Patients and Preventable Spending: A Population-Based Study

Background: Although high-cost (HC) patients make up a small proportion of patients, they account for most health system costs. However, little is known about HC patients with cancer or whether some of their care could potentially be prevented. This analysis sought to characterize HC patients with cancer and quantify the costs of preventable acute care (emergency department visits and inpatient hospitalizations). Methods: This analysis examined a population-based sample of all HC patients in Ontario in 2013. HC patients were defined as those above the 90th percentile of the cost distribution; all other patients were defined as non–high-cost (NHC). Patients with cancer were identified through the Ontario Cancer Registry. Sociodemographic and clinical characteristics were examined and the costs of preventable acute care for both groups by category of visit/condition were estimated using validated algorithms. Results: Compared with NHC patients with cancer (n=369,422), HC patients with cancer (n=187,770) were older (mean age 70 vs 65 years), more likely to live in low-income neighborhoods (19% vs 16%), sicker, and more likely to live in long-term care homes (8% vs 0%). Although most patients from both cohorts tended to be diagnosed with breast, prostate, or colorectal cancer, those with multiple myeloma or pancreatic or liver cancers were overrepresented among the HC group. Moreover, HC patients were more likely to have advanced cancer at diagnosis and be in the initial or terminal phase of treatment compared with NHC patients. Among HC patients with cancer, 9% of spending stemmed from potentially preventable/avoidable acute care, whereas for NHC patients, this spending was approximately 30%. Conclusions: HC patients with cancer are a unique subpopulation. Given the type of care they receive, there seems to be limited scope to prevent acute care spending among this patient group. To reduce costs, other strategies, such as making hospital care more efficient and generating less costly encounters involving chemotherapy, should be explored.

Abstract

Background: Although high-cost (HC) patients make up a small proportion of patients, they account for most health system costs. However, little is known about HC patients with cancer or whether some of their care could potentially be prevented. This analysis sought to characterize HC patients with cancer and quantify the costs of preventable acute care (emergency department visits and inpatient hospitalizations). Methods: This analysis examined a population-based sample of all HC patients in Ontario in 2013. HC patients were defined as those above the 90th percentile of the cost distribution; all other patients were defined as non–high-cost (NHC). Patients with cancer were identified through the Ontario Cancer Registry. Sociodemographic and clinical characteristics were examined and the costs of preventable acute care for both groups by category of visit/condition were estimated using validated algorithms. Results: Compared with NHC patients with cancer (n=369,422), HC patients with cancer (n=187,770) were older (mean age 70 vs 65 years), more likely to live in low-income neighborhoods (19% vs 16%), sicker, and more likely to live in long-term care homes (8% vs 0%). Although most patients from both cohorts tended to be diagnosed with breast, prostate, or colorectal cancer, those with multiple myeloma or pancreatic or liver cancers were overrepresented among the HC group. Moreover, HC patients were more likely to have advanced cancer at diagnosis and be in the initial or terminal phase of treatment compared with NHC patients. Among HC patients with cancer, 9% of spending stemmed from potentially preventable/avoidable acute care, whereas for NHC patients, this spending was approximately 30%. Conclusions: HC patients with cancer are a unique subpopulation. Given the type of care they receive, there seems to be limited scope to prevent acute care spending among this patient group. To reduce costs, other strategies, such as making hospital care more efficient and generating less costly encounters involving chemotherapy, should be explored.

Background

Recent publications have shown that patients with cancer incur high costs of care, particularly after diagnosis and in the last year of life.13 Given the increasing number of patients with cancer4 and increasing treatment costs,5 policymakers are continually seeking ways to bend the cost curve without sacrificing the quality of care. Focusing on high-cost (HC) patients is likely to have a large impact on health system costs, because these patients account for most healthcare spending. Many jurisdictions have implemented high-risk care management strategies to reduce costs and improve quality. However, it is not clear which costs can be reduced, especially among patients who require costly care, such as those with cancer. One potential solution to decrease costs without sacrificing care is to target preventable or avoidable acute care—that is, acute care for which good outpatient care could likely prevent the need for emergency department (ED) visits and/or hospitalization. Previous research has shown that ED visits and hospitalizations make up more than half of all costs among HC patients.6 Furthermore, some studies have shown that a substantial proportion of ED visits7,8 and hospitalizations912 may be prevented.

Little is known about HC patients with cancer and their healthcare utilization and spending patterns. Previous work has shown that individuals with cancer make up 20% of all HC patients.13 This is not surprising given that the population is aging, the number of patients diagnosed with cancer is increasing,4 and the introduction of newer and more expensive treatments has contributed to an increase in treatment costs.5 Wodchis et al14 examined the cost trajectories of patients with cancer over time and identified patient and system characteristics associated with high system costs after cancer treatment. The most common trajectory consisted of patients who were low-cost in the year before cancer treatment and remained low-cost after completing cancer treatment. In addition, the investigators found that increases in age and multimorbidity and low continuity of care were the strongest predictors of high costs after cancer treatment. Lam et al15 examined Medicare data for HC and non–high-cost (NHC) patients with and without cancer in 2014, and found that despite 15% of Medicare patients having a cancer diagnosis, the prevalence of cancer was higher among HC patients (33% vs 13%). Furthermore, the investigators found that HC patients had 3 times greater odds of having cancer and higher total annual spending ($66,685 vs $59,427).

However, it is not clear which costs, if any, can be reduced for this patient population. Joynt et al16 examined preventable and nonpreventable acute care spending among HC Medicare patients using validated algorithms and found that only a small percentage of costs resulted from preventable acute care. Graven et al17 replicated that analysis, using Oregon’s All Payer All Claims database and Medicaid data from the Oregon Health Authority, and found that preventable acute care spending for HC patients accounted for <6% of their total spending. Using data from Canada, Ronksley et al18 examined preventable acute care spending among hospitalized HC patients (at the Ottawa Hospital) using an ambulatory care–sensitive conditions algorithm. They found that among the HC inpatient population, most costs were due to a single nonpreventable acute care episode. Despite this work, none of these studies specifically examined HC subgroups, such as patients with cancer.

The objectives of this analysis were to provide an in-depth characterization of HC patients diagnosed with cancer, while comparing them to NHC patients with cancer, and to quantify the costs of potentially preventable/avoidable acute care (ie, ED visits and inpatient hospitalizations).

Methods

Data

We used administrative healthcare data available through ICES (Institute for Clinical Evaluative Sciences) in Toronto, Ontario, which includes individual-level linkable data on most publicly funded healthcare services for all legal residents of Ontario, Canada’s most populous province.19 Data on inpatient care are captured in the Discharge Abstract Database, the Ontario Mental Health Reporting System, the Continuing Care Reporting System, and the National Rehabilitation Reporting System, whereas data on ambulatory care are recorded in the National Ambulatory Care Reporting System. The Ontario Health Insurance Plan claims database captures data on physician visits, including fee-for-service visits and shadow-billed services, and laboratory claims. The Ontario Drug Benefit program database includes information on all outpatient prescription drugs dispensed to individuals eligible for public drug coverage (ie, individuals on social assistance and those aged ≥65 years). The Home Care Database records all unique visits by home care providers. The Ontario Cancer Registry provides data on all patients diagnosed with a malignant neoplasm in Ontario. The ICES data repository also includes other databases, which provide useful sociodemographic data on patients. The Registered Persons Database, a population-based registry, provides basic demographic data (such as age and sex) on all legal residents of Ontario. The Immigration, Refugees, and Citizenship Canada database provides information on all legal immigrants and refugees in Canada. Census data, obtained through Statistics Canada, contain information on neighborhood-level data, such as income and rurality. All databases were linked using unique encoded identifiers and analyzed at ICES, in compliance with Ontario’s privacy legislation. A full description of the databases can be found in supplemental eTable 1, available with this article at JNCCN.org.

Patient Cohorts

We selected all adult patients (aged ≥18 years) eligible for public healthcare insurance and residing in Ontario in 2013. All individuals who did not contact the Ontario healthcare system in 2013 and those who died during 2013 (but not those in their last year of life) were excluded, in line with previous research,16 because assessing the preventability of end-of-life care costs was beyond the scope of this analysis.

HC patients were defined as those above the 90th percentile of the cost distribution (ie, the top decile) in 2013, in line with previous work.20 All other patients were defined as NHC patients. This threshold provided a larger cohort of patients compared with other thresholds. Within each group, we identified those ever diagnosed with cancer (before or during 2013) through use of the Ontario Cancer Registry. To estimate healthcare costs and thus determine patients with costs above the 90th percentile, we used a cost estimation algorithm, available at ICES.19 Healthcare costs included all costs borne by the public third-party payer (ie, the Ontario Ministry of Health and Long-Term Care)19: inpatient hospitalizations (both psychiatric and acute), other institution-based care (inpatient rehabilitation, complex continuing care, long-term care), ED visits and other ambulatory care, outpatient clinic visits, physician visits and outpatient care, outpatient prescription drugs (for individuals covered under the public provincial drug plan), and home care. Costs captured by the algorithm accounted for >90% of all government-paid healthcare costs.21

Identifying Preventable Acute Care

To identify preventable ED visits, we used an updated version of a validated algorithm created at NYU Wagner22 by John Billings and colleagues and used in prior research examining HC Medicare patients.16 The NYU algorithm uses specified diagnosis codes to determine the following type of ED visits: nonemergent, emergent but primary care treatable, emergent ED care needed but preventable, and emergent ED care needed and not preventable. The updated version of the algorithm includes additional categories for injury, mental health, alcohol use, and drug use.23 Based on these classifications, Billings et al24 compiled a set of probabilistic weights that were applied to ED discharge data using the primary discharge ICD-10 diagnosis codes to determine the percentage of ED use attributable to each category. ED visits with diagnosis codes not mapped to any of the 8 categories were assigned to the “unclassified” category. Preventable ED visits were defined as nonemergent, emergent/primary care treatable, and emergent/ED care needed but preventable, in line with prior work.16

To identify potentially preventable hospitalizations, we used a validated algorithm, Quality Prevention Quality Indicators, developed by the Agency for Healthcare Research and Quality25 and used elsewhere.2628 This algorithm defines potentially preventable hospitalizations as those related to conditions, such as heart failure, diabetes, hypertension, and asthma, for which good outpatient care can likely prevent the need for hospitalization. In addition, we used the respective prevention quality indicators for common nonpreventable clinical diagnosis groups to identify nonpreventable hospitalizations (see supplemental eAppendix 1 for codes).

Analysis

Patient Descriptives

HC and NHC patients with cancer were characterized and compared in terms of sociodemographic (sex, age, migrant status, neighborhood-level income quintile, urban/rural residence) and clinical characteristics (chronic conditions, frailty, residence in a long-term care home). We used the existing ICES-derived cohorts/registries to ascertain chronic conditions among patients2940 and the Johns Hopkins Adjusted Clinical Groups Case-Mix System41 to obtain data on patient frailty. Residence in a long-term care facility was determined through the Continuing Care Reporting System. Using information included in the Ontario Cancer Registry, we also characterized patient groups by cancer type (solid vs hematologic) and site, cancer stage at diagnosis (where available), treatment phase (initial, continuing, terminal),1,2 and days since first diagnosis (patients with multiple cancers were assigned to the first cancer diagnosed; see supplemental eTable 2 for ICD-O codes). Patient characteristics between HC and NHC individuals with cancer were compared using a chi-square test for differences.

Costs of Preventable Acute Care Among HC Patients

We estimated the total costs of preventable and nonpreventable acute care by category of visit (for ED visits) and condition (for inpatient hospitalizations) for HC and NHC patients with cancer. All costs were reported in 2016 constant Canadian dollars using the Statistics Canada Consumer Price Index for healthcare.42

All analyses were performed using SAS Enterprise Guide 7.1 (SAS Institute Inc). The study was approved by the research ethics board of Sunnybrook Health Sciences Centre.

Results

Patient Characteristics

In 2013, 10,031,865 individuals contacted the Ontario healthcare system; 1,003,187 were defined as HC patients, and the remainder were defined as NHC (n=9,028,678). After excluding all patients who died during 2013 (n=90,555), we were left with 929,726 HC and 9,011,584 NHC patients. Across both groups, 557,192 patients had cancer; 187,770 were HC and 369,422 were NHC.

HC patients with cancer were older (mean age, 69.9 vs 64.5 years) and slightly more represented in the lower neighborhood income quintile (19.0% vs 15.6%) compared with NHC patients, but the 2 groups were roughly the same for all other sociodemographic characteristics (Table 1). HC patients with cancer had a higher proportion of all chronic conditions examined, particularly HIV, chronic heart failure, and dementia. HC patients with cancer also had a higher degree of frailty and were more likely to be living in long-term care facilities (Table 2). Among all patients, most were diagnosed with breast, prostate, and colorectal cancers (Table 3). However, multiple myeloma and pancreatic and liver cancers were overrepresented among HC patients, whereas testicular, thyroid, and cervical cancers and melanoma were overrepresented among NHC patients. Moreover, although most patients had missing data on cancer stage, a higher proportion of patients with stages III and IV was seen in the HC group. There was also a higher proportion of patients in the initial and terminal phases of treatment among the HC group (note that the terminal phase did not include patients who died but rather those in the last year of life, for whom death likely occurred in 2014). HC patients with cancer also had a shorter length of time since first diagnosis (2,917.9 vs 3,844.6 days).

Table 1.

Sociodemographic Characteristics of Patient Cohorts

Table 1.
Table 2.

Clinical Characteristics of Patient Cohorts

Table 2.
Table 3.

Cancer-Specific Characteristics of Study Cohorts

Table 3.

Costs of Preventable and Nonpreventable Acute Care

ED Visits

Within the HC cohort, 30.8% of the ED costs were classified as preventable (Table 4). Patterns were somewhat similar for the NHC cohort, with 35.3% of costs deemed preventable. Emergent, primary care treatable ED visits made up the highest proportion of preventable care costs for both HC and NHC patients (13.2% and 15.5%, respectively). However, HC patients had a higher proportion of nonpreventable ED visit costs than NHC patients (20.1% vs 13.8%). In addition, costs of ED visits with injuries represented a relatively high proportion among NHC patients (16.4%).

Table 4.

Costs of Preventable and Nonpreventable ED Visits

Table 4.

Inpatient Care

HC patients with cancer had a higher proportion of preventable inpatient costs than NHC patients (7.3% vs 6.6%) (Table 5). The most common reasons for preventable hospitalizations among HC patients with cancer included chronic obstructive pulmonary disease (COPD), bacterial pneumonia, and urinary tract infections; this was also the case for NHC patients (but with different rankings). In contrast, the most common reasons for nonpreventable hospitalizations were cancer and chemotherapy (28.1%), orthopedic conditions (5.0%), and ischemic heart disease (4.0%) for HC patients, and cancer and chemotherapy (13.8%), gastrointestinal infections and disorders (10.7%), and syncope and dizziness (3.3%) among NHC patients (data not shown).

Table 5.

Costs of Preventable and Nonpreventable Hospitalizations

Table 5.

Combining both acute settings, 9.0% of HC patients’ costs were considered potentially preventable compared with 29.9% for HNC patients.

Discussion

We examined a population-based sample of HC patients with cancer and estimated potentially preventable acute care spending while comparing them with NHC patients with cancer. HC patients were older, slightly more represented in the lowest neighborhood income quintile, sicker, and more likely to live in long-term care homes compared with NHC patients. HC patients were also more likely to be diagnosed with high-mortality cancers, such as multiple myeloma and pancreatic and liver cancers, and to have an advanced stage at diagnosis. Moreover, HC patients were more likely to be in the initial and terminal phases of treatment and to have a shorter length of time since first diagnosis. Roughly 30.9% of ED visit costs and 7.3% of hospitalization costs among HC patients could potentially be avoided, in contrast to roughly 35.4% and 6.6%, respectively, for NHC patients. The top categories for preventable care were primary care–treatable ED visits and hospitalizations for COPD, pneumonia, and urinary tract infections. Combining both acute settings, 9.0% of HC patients’ costs were considered potentially preventable, whereas the corresponding value for NHC patients was more than 3 times higher (29.9%).

Previous research has found that patients with high-mortality cancers have higher treatment costs.43 This is also the case for patients with advanced cancer1 and those in the initial and terminal phases of treatment.1,2,44,45 Moreover, another study on HC patients showed that long-term care residence has a large impact on costs.46 Therefore, it is not surprising that patients with these characteristics were more often in the HC category.

Some research has examined the relationship between being an HC patient and having cancer. For example, Wodchis et al14 examined the cost trajectories of patients with cancer over time to understand the predictors of being an HC patient after a cancer diagnosis. The most common trajectory was low-cost in the years before and after cancer treatment (31.4%). The next most common trajectory, which accounted for nearly 15% of cases examined, included patients who were low-cost before diagnosis but HC in the year after cancer treatment ended, suggesting that cancer plays a role in becoming a HC patient. Lam et al15 examined Medicare data for HC and NHC patients with and without cancer in 2014 and found that the prevalence of cancer was higher among HC patients (33% vs 13%) and that these patients had 3 times greater odds of having cancer and higher total annual spending. No existing studies have specifically examined preventable acute care among specific HC subgroups, such as HC patients with cancer; most of the literature on this topic has examined all HC patients and found that only a small percentage of costs were due to preventable acute care.13,16,17 We also found this to be the case for HC patients with cancer. In particular, we found that only 9% of acute care costs were considered preventable, which is close to what other researchers have found in the United States, albeit for the entire HC population.

These results suggest that overall, among HC patients with cancer, there is limited scope to prevent acute care costs. However, a greater opportunity may exist to prevent these costs among some types of cancer and/or cancer sites. For example, despite small differences, HC patients with hematologic cancers had a slightly higher proportion of preventable acute care costs than those with solid cancers (data not reported). Future work should seek to undertake this analysis by cancer type/site. Despite the fact that more than one-third of acute care costs were for healthcare encounters that could not be classified, a large proportion of costs were for nonpreventable acute care. In particular, the largest proportion of nonpreventable hospitalization costs was for episodes of care involving chemotherapy, which is standard treatment for many patients with cancer. Chemotherapy costs tend to represent a large proportion of treatment costs, and these costs have been increasing over time.5 Thus, the ability to lower acute care costs through better outpatient care may be limited. Nonetheless, there may be instances when chemotherapy may be avoided47 or deemed unnecessary,48 despite these hospitalizations being defined as nonpreventable by the algorithm. Moreover, costs could potentially be reduced by making hospital care more efficient so that each inpatient care episode is less costly,16 and through less costly encounters involving chemotherapy.

Strengths of our study include its use of a population-based sample of all adult HC patients with cancer in Ontario, which is Canada’s most populous province, whereas many existing studies tend to be limited to older populations.15,16 Additionally, we accounted for all costs associated with ED visits and hospitalization, given the presence of a (sole) third-party public payer for acute care in Ontario. We provided an in-depth characterization of the HC patient population with cancer and quantified the costs of potentially preventable acute care.

Nonetheless, this study is not without limitations. We did not examine patients aged <18 years, because these patients typically have different cancers compared with adults and their cancer-related care is organized differently compared with adults. We also did not examine the preventability of end-of-life care, because this was outside the scope of the analysis. Additionally, data on cancer stage were missing for most patients during the analysis period. Lastly, we used algorithms developed in the United States to identify preventable acute care; however, these algorithms could not classify all visits, and considered all ED visits for mental health and alcohol and drug use as nonpreventable, although it is likely that some of these visits could potentially be preventable. Furthermore, these algorithms are not disease-specific and thus did not examine preventable cancer-specific acute care; future work should explore this area and what constitutes potentially preventable cancer-specific acute care.

Conclusions

HC patients with cancer are a unique subpopulation within the HC patient population, with specific needs and healthcare utilization patterns. Given the care they receive, there seems to be limited scope to prevent acute care spending. To reduce costs, other strategies should be explored, such as making hospital care more efficient and ensuring encounters involving chemotherapy are less costly.

References

  • 1.

    Yabroff KR, Lamont EB, Mariotto A, . Cost of care for elderly cancer patients in the United States. J Natl Cancer Inst 2008;100:630–641.

  • 2.

    de Oliveira C, Pataky R, Bremner KE, . Phase-specific and lifetime costs of cancer care in Ontario, Canada. BMC Cancer 2016;16:809.

  • 3.

    de Oliveira C, Pataky R, Bremner KE, . Estimating the cost of cancer care in British Columbia and Ontario: a Canadian inter-provincial comparison. Healthc Policy 2017;12:95–108.

    • Search Google Scholar
    • Export Citation
  • 4.

    Canadian Cancer Statistics Advisory Committee. Canadian Cancer Statistics 2017. Available at: https://www.canada.ca/en/public-health/services/chronic-diseases/cancer/canadian-cancer-statistics.html. Accessed August 1, 2019.

  • 5.

    de Oliveira C, Weir S, Rangrej J, . The economic burden of cancer care in Canada: a population-based cost study. CMAJ Open 2018;6:E1–10.

  • 6.

    Riley GF. Long-term trends in the concentration of Medicare spending. Health Aff (Millwood) 2007;26:808–816.

  • 7.

    Smulowitz PB, Lipton R, Wharam JF, . Emergency department utilization after the implementation of Massachusetts health reform. Ann Emerg Med 2011;58:225–234.e1.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 8.

    Ballard DW, Price M, Fung V, . Validation of an algorithm for categorizing the severity of hospital emergency department visits. Med Care 2010;48:58–63.

  • 9.

    Niefeld MR, Braunstein JB, Wu AW, . Preventable hospitalization among elderly Medicare beneficiaries with type 2 diabetes. Diabetes Care 2003;26:1344–1349.

  • 10.

    Billings J, Anderson GM, Newman LS. Recent findings on preventable hospitalizations. Health Aff (Millwood) 1996;15:239–249.

  • 11.

    Oster A, Bindman AB. Emergency department visits for ambulatory care sensitive conditions: insights into preventable hospitalizations. Med Care 2003;41:198–207.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12.

    Braunstein JB, Anderson GF, Gerstenblith G, . Noncardiac comorbidity increases preventable hospitalizations and mortality among Medicare beneficiaries with chronic heart failure. J Am Coll Cardiol 2003;42:1226–1233.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13.

    de Oliveira C, Cheng J, Kurdyak P. Determining preventable acute care spending among high-cost patients in a single-payer public health care system. Eur J Health Econ 2019;20:869–878.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14.

    Wodchis WP, Arthurs E, Khan AI, . Cost trajectories for cancer patients. Curr Oncol 2016;23(Suppl 1):S64–75.

  • 15.

    Lam MB, Burke LG, Orav EJ, . Proportion of patients with cancer among high-cost Medicare beneficiaries: who they are and what drives their spending. Healthc (Amst) 2018;6:46–51.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 16.

    Joynt KE, Gawande AA, Orav EJ, . Contribution of preventable acute care spending to total spending for high-cost Medicare patients. JAMA 2013;309:2572–2578.

  • 17.

    Graven PF, Meath THA, Mendelson A, . Preventable acute care spending for high-cost patients across payer types. J Health Care Finance 2016;42:1–22.

    • Search Google Scholar
    • Export Citation
  • 18.

    Ronksley PE, Kobewka DM, McKay JA, . Clinical characteristics and preventable acute care spending among a high cost inpatient population. BMC Health Serv Res 2016;16:165.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19.

    Wodchis WP, Bushmeneva K, Nikitovic M, McKillop I. Guidelines on person-level costing using administrative databases in Ontario. Available at: http://www.hsprn.ca/uploads/files/Guidelines_on_PersonLevel_Costing_May_2013.pdf. Accessed August 1, 2019.

  • 20.

    de Oliveira C, Cheng J, Vigod S, . Patients with high mental health costs incur over 30 percent more costs than other high-cost patients. Health Aff (Millwood) 2016;35:36–43.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 21.

    Wodchis WP, Austin PC, Henry DA. A 3-year study of high-cost users of health care. CMAJ 2016;188:182–188.

  • 22.

    NYU Wagner. NYU ED algorithm: background. Available at: https://wagner.nyu.edu/faculty/billings/nyued-background. Accessed August 1, 2019.

  • 23.

    Johnston KJ, Allen L, Melanson TA, . A “patch” to the NYU emergency department visit algorithm. Health Serv Res 2017;52:1264–1276.

  • 24.

    Billings J, Parikh N, Mijanovich T. Emergency department use: the New York story. Issue Brief (Commonw Fund) 2000:1–12.

  • 25.

    Agency for Healthcare Research and Quality. Prevention Quality Indicators overview. Available at: http://www.qualityindicators.ahrq.gov/modules/pqi_overview.aspx. Accessed December 17, 2018.

    • Export Citation
  • 26.

    Basu J, Friedman B, Burstin H. Primary care, HMO enrollment, and hospitalization for ambulatory care sensitive conditions: a new approach. Med Care 2002;40:1260–1269.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 27.

    Bindman AB, Grumbach K, Osmond D, . Preventable hospitalizations and access to health care. JAMA 1995;274:305–311.

  • 28.

    Jiang HJ, Russo CA, Barrett ML. Nationwide frequency and costs of potentially preventable hospitalizations, 2006. HCUP statistical brief #72. April 2009. US Agency for Healthcare Research and Quality, Rockville, MD. Available at: http://www.hcup-us.ahrq.gov/reports/statbriefs/sb72.pdf. Accessed October 20, 2019.

    • PubMed
    • Export Citation
  • 29.

    Antoniou T, Zagorski B, Loutfy MR, . Validation of case-finding algorithms derived from administrative data for identifying adults living with human immunodeficiency virus infection. PLoS One 2011;6:e21748.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 30.

    Benchimol EI, Guttmann A, Mack DR, . Validation of international algorithms to identify adults with inflammatory bowel disease in health administrative data from Ontario, Canada. J Clin Epidemiol 2014;67:887–896.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 31.

    Gershon AS, Wang C, Guan J, . Identifying patients with physician-diagnosed asthma in health administrative databases. Can Respir J 2009;16:183–188.

  • 32.

    Gershon AS, Wang C, Guan J, . Identifying individuals with physician-diagnosed COPD in health administrative databases. COPD 2009;6:388–394.

  • 33.

    Hux JE, Ivis F, Flintoft V, . Diabetes in Ontario: determination of prevalence and incidence using a validated administrative data algorithm. Diabetes Care 2002;25:512–516.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 34.

    Jaakkimainen RL, Bronskill SE, Tierney MC, . Identification of physician-diagnosed Alzheimer’s disease and related dementias in population-based administrative data: a validation study using family physicians’ electronic medical records. J Alzheimers Dis 2016;54:337–349.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 35.

    Kurdyak P, Lin E, Green D, . Validation of a population-based algorithm to detect chronic psychotic illness. Can J Psychiatry 2015;60:362–368.

  • 36.

    Moist LM, Fenton S, Kim JS, . Canadian Organ Replacement Register (CORR): reflecting the past and embracing the future. Can J Kidney Health Dis 2014;1:26.

  • 37.

    Prodhan S, King MJ, De P, . Health services data: the Ontario Cancer Registry (a unique, linked, and automated population-based registry). In: Sobolev B, Levy A, Goring S, eds. Data and Measures in Health Services Research.Boston, MA: Springer; 2016:1–27.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 38.

    Schultz SE, Rothwell DM, Chen Z, . Identifying cases of congestive heart failure from administrative data: a validation study using primary care patient records. Chronic Dis Inj Can 2013;33:160–166.

    • Search Google Scholar
    • Export Citation
  • 39.

    Tu K, Chen Z, Lipscombe LL. Prevalence and incidence of hypertension from 1995 to 2005: a population-based study. CMAJ 2008;178:1429–1435.

  • 40.

    Widdifield J, Bernatsky S, Paterson JM, . Accuracy of Canadian health administrative databases in identifying patients with rheumatoid arthritis: a validation study using the medical records of rheumatologists. Arthritis Care Res (Hoboken) 2013;65:1582–1591.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 41.

    Weiner JP, ed. The Johns Hopkins ACG Case-Mix System. Version 7.0 Release Notes. May, 2005. Baltimore, MD: Johns Hopkins Bloomberg School of Public Health; 2005.

    • Export Citation
  • 42.

    Statistics Canada. The consumer price index. Ottawa (ON): Statistics Canada; 2016. Cat. no. 62-001-X.

  • 43.

    de Oliveira C, Bremner KE, Pataky R, . Understanding the costs of cancer care before and after diagnosis for the 21 most common cancers in Ontario: a population-based descriptive study. CMAJ Open 2013;1:E1–8.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 44.

    Laudicella M, Walsh B, Burns E, . Cost of care for cancer patients in England: evidence from population-based patient-level data. Br J Cancer 2016;114:1286–1292.

  • 45.

    Banegas MP, Yabroff KR, O’Keeffe-Rosetti MC, . Medical care costs associated with cancer in integrated delivery systems. J Natl Compr Canc Netw 2018;16:402–410.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 46.

    de Oliveira C, Cheng J, Rehm J, . The role of mental health and addiction among high-cost patients: a population-based study. J Med Econ 2018;21:348–355.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 47.

    Goyal RK, Wheeler SB, Kohler RE, . Health care utilization from chemotherapy-related adverse events among low-income breast cancer patients: effect of enrollment in a medical home program. N C Med J 2014;75:231–238.

    • Search Google Scholar
    • Export Citation
  • 48.

    Sparano JA, Gray RJ, Makower DF, . Adjuvant chemotherapy guided by a 21-gene expression assay in breast cancer. N Engl J Med 2018;379:111–121.

If the inline PDF is not rendering correctly, you can download the PDF file here.

Submitted January 8, 2019; accepted for publication July 26, 2019.Author contributions: Study concept and design: de Oliveira, Chan, Earle, Krahn, Mittmann. Data acquisition: Cheng. Data analysis and interpretation: All authors. Project management: de Oliveira. Manuscript preparation: de Oliveira. Critical revision: Cheng, Chan, Earle, Krahn, Mittmann. Disclosures: The authors have not received any financial consideration from any person or organization to support the preparation, analysis, results, or discussion of this article.Funding and disclaimers: This study was conducted with the support of Cancer Care Ontario (CCO) through funding provided by the Government of Ontario. Parts of this material are based on data and information provided by CCO. The opinions, results, view, and conclusions reported in this article are those of the authors and do not necessarily reflect those of CCO. No endorsement by CCO is intended or should be inferred. This study was also supported by the Institute for Clinical Evaluative Sciences (ICES), which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). The opinions, results, and conclusions reported in this article are those of the authors and are independent from the funding sources. No endorsement by ICES or the Ontario MOHLTC is intended or should be inferred. Parts of this material are based on data and/or information compiled and provided by the Canadian Institute for Health Information (CIHI). However, the analyses, conclusions, opinions, and statements expressed in the material are those of the author(s), and not necessarily those of CIHI.Correspondence: Claire de Oliveira, PhD, Institute for Mental Health Policy Research, Centre for Addiction and Mental Health, 33 Russell Street, Room T414, Toronto, Ontario M5S 2S1, Canada. Email: claire.deoliveira@camh.ca

Supplementary Materials