Measuring patient-reported outcomes to improve cancer care in Canada: an analysis of provincial survey data

Patient-reported outcomes measures (proms) are an important component of the shift from disease-centred to person-centred care. In oncology, proms describe the effects of cancer and its treatment from the patient perspective and ideally enable patients to communicate to their providers the physical symptoms and psychosocial concerns that are most relevant to them. The Edmonton Symptom Assessment System-revised (esas-r) is a commonly used and validated tool in Canada to assess symptoms related to cancer. Here, we describe the extent to which patient-reported outcome programs have been implemented in Canada and the severity of symptoms causing distress for patients with cancer. As of April 2017, 8 of 10 provinces had implemented the esas-r to assess patient-reported outcomes. Data capture methods, the proportion of cancer treatment sites that have implemented the esas-r, and the time and frequency of screening vary from province to province. From October 2016 to March 2017 in the 8 reporting provinces, 88.0% of cancer patients were screened for symptoms. Of patients who reported having symptoms, 44.3% reported depression, with 15.5% reporting moderate-to-high levels; 50.0% reported pain, with 18.6% reporting moderate-to-high levels; 56.2% reported anxiety, with 20.4% reporting moderate-to-high levels; and 75.1% reported fatigue, with 34.4% reporting moderate-to-high levels. There are some notable areas in which the implementation of proms could be improved in Canada. Findings point to a need to increase the number of cancer treatment sites that screen all patients for symptoms; to standardize when and how frequently patients are screened across the country; to screen patients for symptoms during all phases of their cancer journey, not just during treatment; and to assess whether giving cancer care providers real-time patient-reported outcomes data has led to appropriate interventions that reduce the symptom burden and improve patient outcomes. Continued measurement and reporting at the system level will allow for a better understanding of progress in proms activity over time and of the areas in which targeted quality improvement efforts could ensure that patient symptoms and concerns are being addressed.

Investigation of the international comparability of population-based routine hospital data set derived comorbidity scores for patients with lung cancer

INTRODUCTION

The International Cancer Benchmarking Partnership (ICBP) identified significant international differences in lung cancer survival. Differing levels of comorbid disease across ICBP countries has been suggested as a potential explanation of this variation but, to date, no studies have quantified its impact. This study investigated whether comparable, robust comorbidity scores can be derived from the different routine population-based cancer data sets available in the ICBP jurisdictions and, if so, use them to quantify international variation in comorbidity and determine its influence on outcome.

METHODS

Linked population-based lung cancer registry and hospital discharge data sets were acquired from nine ICBP jurisdictions in Australia, Canada, Norway and the UK providing a study population of 233 981 individuals. For each person in this cohort Charlson, Elixhauser and inpatient bed day Comorbidity Scores were derived relating to the 4-36 months prior to their lung cancer diagnosis. The scores were then compared to assess their validity and feasibility of use in international survival comparisons.

RESULTS

It was feasible to generate the three comorbidity scores for each jurisdiction, which were found to have good content, face and concurrent validity. Predictive validity was limited and there was evidence that the reliability was questionable.

CONCLUSION

The results presented here indicate that interjurisdictional comparability of recorded comorbidity was limited due to probable differences in coding and hospital admission practices in each area. Before the contribution of comorbidity on international differences in cancer survival can be investigated an internationally harmonised comorbidity index is required.

Evaluating the impact of survivorship models on health system resources and costs

1

Background: The provincial cancer agency in Ontario developed and implemented a model of care (MOC) for breast cancer (BC) survivors to transition from oncology-lead care to primary care in a publically funded health care environment (2010-2013). Transition options included direct to primary care and stepped transition. The objective of our study was to examine the health system resources used by the women in the MOC group and compare them to those used by women who did not transition. Methods: A propensity score matched, quasi-experimental approach was used to compare the healthcare resource utilization and costs between BC survivors in the MOC program (case) and those receiving usual care (control). All MOC cases were linked using unique identifiers and linked into the provincial health system databases. Cases and controls were matched 1:1 on year of diagnosis and location of care and were followed from an index date to the earliest of her death date, date of last contact in the database, one day before another cancer diagnosis or the end of study available databases. The primary study outcome was overall health system utilization and mean cost during the follow-up period. Results: There were 2324 women in the MOC program. Demographic information (age, region, stage) were well balanced between cases and controls. Transitioned cases had lower hospitalizations (20.1% vs. 24.4%, p<0.05), fewer cancer clinic visits (6.0% vs. 15.1%, p<0.05), fewer medical oncologist visits (0.39 vs. 1.29, p<0.05) and fewer diagnostics (CT, MRI, ultrasound, x-rays) over an average of 25 months of follow-up. There was a trend for fewer family practice (7.35 vs. 7.91, p=0.08) and internal medical and hematology visits (0.81 vs. 1.03, p=0.08). Annual emergency visits were similar between the two groups (0.76 vs. 0.82, p=0.2). There was a $4300 (2012 $CAN) difference in the mean annual cost between cases and controls. Conclusions: Survivors in the MOC transition program used fewer health system resources and had lower health system costs when compared to controls. These findings provide real world evidence to inform transition policies for cancer survivors.

Comparative effectiveness and safety of the implementation of universal public funding of FOLFIRINOX (FFX) and gemcitabine (G) + nab-paclitaxel (GnP) in advanced pancreatic cancer (APC): A population-based study

375

Background: FFX has been universally publicly funded in Ontario, Canada, for metastatic pancreatic cancer (mPC) and unresectable locally advanced pancreatic cancer (uLAPC) since 11/2011 and 04/2015, respectively. GnP has been publicly funded for uLAPC and mPC (APC) since 04/2015. We examined the real world comparative effectiveness and safety of implementing funding of FFX and GnP for patients with APC. Methods: Patients with APC who received first-line FFX, GnP, or G from 01/2008-03/2016 were identified in CCO’s New Drug Funding Program database and divided into 3 periods: 01/2008-10/2011 (P1), 11/2011-03/2015 (P2), and 04/2015-03/2016 (P3). Data were linked with the Ontario Cancer Registry and others to ascertain demographics, comorbidities, and outcomes. Matching weights of propensity score to simultaneously compare three periods were generated using multinomial logistic regression. Crude and adjusted survival analyses were conducted to assess overall survival (OS) using Kaplan-Meier and weighted Cox regression methods.Weighted negative binomial models were used to estimate rate ratios (RR) for all-cause hospitalization (H) and ED visits. Results: We identified 3696 patients (1250 in P1, 1891 in P2, 555 in P3) (overall mean age 65, female 46%). In P2, 49% received FFX. In P3, 53% received FFX and 35% received GnP. Median OS was 5.7, 7.0, and 7.5 months for P1, P2, and P3, respectively. Median OS for FFX and GnP in mPC were 8.8 and 5.5 months, respectively. OS was improved in P2 vs. P1 (HR = 0.84, 0.78-0.90) and in P3 vs. P2 (HR = 0.82, 0.73-0.92). ED visits were similar compared P2 vs. P1 (RR=1.02, p = 0.75) and P3 vs. P2 (RR=1.04, p = 0.48), and H was reduced in P2 vs. P1 (RR = 0.86, p = 0.01), but similar in P3 vs. P2 (RR = 0.98, p = 0.78). H for febrile neutropenia (FN) was increased in P2 vs. P1 (RR = 2.18, p = 0.04) but not in P3 vs. P2 (RR = 1.32, p = 0.45). Conclusions: Implementation of universal public funding of FFX for mPC improved OS and reduced the rates H overall, but increased FN-related H. Funding of FFX for uLAPC and GnP for APC improved OS without increased in ER and H.

Chemotherapy choice in advanced pancreatic cancer: What patient and disease factors influence prescription patterns?

327

Background: FOLFIRINOX (FFX), gemcitabine+nab-paclitaxel (GnP) and gemcitabine monotherapy (Gem)) are universally funded as first-line chemotherapy regimens for advanced pancreatic cancer (APC) in Ontario, Canada. However, there is scarce real-world data on factors that may influence choice of chemotherapy regimens in APC. Methods: Patients who received first-line chemotherapy for APC between April 2015-March 2016 in Ontario were identified from CCO’s New Drug Funding Program database and linked to the Ontario Cancer Registry and other provincial databases to ascertain baseline factors. Multinomial logistic regressions were used to examine the associations between the prescribed chemotherapy regimen and baseline factors. Results: 546 patients were identified, with a mean age of 65 and 43.6% female. 9.9% and 9.7% had received adjuvant gemcitabine and radiation treatment respectively. 17.6% had previous pancreatic resection. 68.3% had zero Charlson score and 30.6% had ECOG performance status (PS) of 0. 72.7% had metastatic disease. The majority of the patients received FFX (52.4%) compared to GnP (35.7%) and Gem (11.9%). Age and ECOG PS were strongly associated with choice of chemotherapy regimens. (See Table) Conclusions: In Ontario, increased patient age and worse ECOG PS are strongly associated with choice of Gem compared to GnP and FFX. Previous treatments and stage of disease also impact chemotherapy choice. Understanding how providers choose chemotherapy in APC aids in comprehending our practices. Odds ratio (OR) and p value from multinomial logistic regressions. [Table: see text]

Does prognostic uncertainty affect discussions of prognosis? Lessons from a survey of hematologic oncologists

45

Background: Although recent advances in cancer therapy have improved survival for patients with solid tumors, they have also increased the complexity of prognostication (Temel, JCO 2016). Prognostic uncertainty is particularly prevalent in hematologic oncology (LeBlanc, JOP 2014) and potentially a barrier to timely end-of-life (EOL) communication (Odejide, JCO 2016). Methods: In 2015, we mailed a 30-item survey to a national sample of hematologic oncologists randomly selected from the American Society of Hematology directory. The survey was developed through focus groups (n = 20) and cognitive debriefing (n = 5). We aimed to characterize respondents’ reports of prognostic discussions, as well as their timeliness and content. Results: We received 349 surveys from 48 states (response rate: 57%). Median time in practice was 25 years and 57% practiced in community settings. Overall, 60% reported discussing prognosis with “most” ( > 95%) of their patients. Those with < 15 years clinical experience (AOR = 0.54, 95% CI 0.31, 0.94) and those considering prognostic uncertainty to be a barrier to EOL care (AOR = 0.57, 95% CI 0.35, 0.92) were less likely to have prognostic discussions with “most” of their patients. When discussing prognosis, almost all (98%) reported typically having an initial discussion at diagnosis or during a period of stability; however, 18% reported either never readdressing prognosis or doing so only when death is clearly imminent. In terms of preferred terminology, 57% reported routinely having “general discussions of potentially curable disease,” while 43% preferred providing specific data such as percent chance of survival or median survival. Conclusions: The majority of hematologic oncologists in this large cohort reported discussing prognosis with their patients, but doing so qualitatively, focusing on whether cure is possible. About one-fifth reported not readdressing prognosis in a timely manner. These suggest that the prognostic uncertainty common with blood cancers fosters missed opportunities to convey what is known about prognosis. Given the growing difficulty in solid tumor prognostication, these data may foreshadow coming communication gaps for oncology as a whole.

Temporal treatments and outcomes following acute myocardial infarction among cancer survivors: A population-based study, 1995-2013

10058

Background: There is little contemporary information regarding cardiac care and mortality differences following an acute myocardial infarction (AMI) between cancer survivors (CS) and non-cancer patients (NCP). Methods: All patients with AMI (1995-2013) in Ontario, Canada were identified through administrative databases and stratified into CS (solid or hematologic) and NCP. Those with cancer within 1 year of AMI were excluded. We used inverse probability treatment weight of propensity scores to balance confounders. Coronary intervention use and survival following index AMI were compared between CS and NCP using Modified Poisson and Cox modeling, and their temporal trends were examined. Results: Of 270,089 AMI patients (62.1% men; 87.8% >65 yrs old for CS vs. 56.2% for NCP), 22,907 were CS (prostate 26%, colorectal 17%, breast 16%) and 247,182 NCP. From 1995-2013, coronary interventions usage increased similarly for both groups (Table). The overall 30-day use did not differ between CS and NCP (angiogram unadjusted 36% vs. 50%, adjusted RR 0.96, 95% CI 0.96-1.00, p=.21; percutaneous coronary interventions unadjusted 21% vs. 31%, adjusted RR 0.98, 95% CI 0.94-1.01, p=.21; bypass surgery unadjusted 5% vs. 7%, adjusted RR 0.95, 95% CI 0.87-1.04, p=.25). Unadjusted 30-day mortality following AMI decreased similarly for CS and NCP (Table). However, adjusted 30-day mortality was worse in CS (HR 1.09, 95% CI 1.04-1.15, p<0.001). Over median follow-up of 11 yrs, CS had worse survival than NCP (HR 1.22, 95% CI 1.18-1.26, p<0.0001). CS had higher risk of heart failure than NCP (HR 1.10, 95% CI 1.05-1.15, p<0.0001), while myocardial (re)-infarction and stroke were similar (HR 0.99, 95% CI 0.96-1.02, p=.46; HR 1.09, 95% CI 1.00-1.18, p=.052). Conclusions: Following AMI, coronary intervention use increased and early mortality decreased comparably between CS and NCP over time. However, CS had worse short-term and long-term survival, suggesting that continued emphasis on cancer and cardiovascular care is needed to improve outcomes. [Table: see text]

Burden of out-of-pocket spending among high-deductible health plan members with metastatic breast cancer

1029

Background: 50% of workers have high-deductible health plans (HDHP) that require major outofpocket (OOP) spending for cancerrelated care. The OOP burden among patients with advanced cancer in HDHPs is unknown. Our objective was to estimate OOP spending for women with metastatic breast cancer (mbc) stratified by health plan type. Methods: Our data source was administrative health insurance claims and enrollment data of members insured though a large national health plan. We included 7142 women age 25-64 with mbc who had at least 6 months enrollment before the diagnosis and at least 12 months followup. We used a time series design and plotted OOP spending stratified by HDHP vs low-deductible plan. Primary outcome measures included: (1) 20042012 calendar trends in total annual OOP spending, (2) monthly total OOP spending in the 6 months before and 24 months after women were diagnosed with mbc, and (3) monthly total OOP spending in the last 6 months of life. Plots were adjusted for age, socioeconomic status, race/ethnicity, and US region of residence, and we then conducted linear regression to assess for statistical significance of trends. Results: In 2004, average annual OOP spending for women with mbc cancer in low-deductible health plans was $1196.2 and increased to $2570 in 2012, a yearly increase of $159.2 (113.2205.2). For women in HDHP average OOP spending in 2004 amounted to $2648 and increased to $3736.4 in 2012, representing an annual increase of $160.4 per year (105.4215.4) Average OOP spending per person month peaked in the month of diagnosis to $1633.8 for women in HDHPs and to $643 among low-deductible plan members. Average OOP spending in the last 6 months of life were $285.7 per person month among low-plan ($1714.2 per 6 months) and $607.3 among HDHP ($3644 per 6 months). Conclusions: To our knowledge, this is the first analysis to estimate OOP spending for women with mbc accounting for enrollment in HDHPs versus low-deductible plans. We found that OOP spending is increasing over time and is high in the last 6 months of life. HDHP members with mbc faced much higher OOP spending than women in traditional plans across all analyses. Findings raise concerns that HDHPs could worsen access to mbc treatments.

Predictors of specialized pediatric palliative care involvement and impact on patterns of end-of-life care in children with cancer: A population-based study

10573

Background: Children with cancer are at risk of receiving high-intensity (HI) care at the end-of-life (EOL) and associated high symptom burden. The impact of palliative care (PC) delivered by generalists or of specialized pediatric palliative care (SPPC) on patterns of EOL care is unknown, with previous studies limited by small sample sizes or low response rates. Methods: Using a provincial registry, we assembled a retrospective cohort of Ontario children with cancer who died between 2000-2012 and who received care through a pediatric institution with a SPPC team and a clinical PC database. Patients were linked to population-based healthcare data capturing inpatient, outpatient, and emergency visits. Clinical PC databases were used to identify patients receiving SPPC. Remaining patients were categorized as having received either general PC (GPC) or no PC depending on the presence of PC associated physician billing or inpatient codes. We determined predictors of SPPC involvement, and whether either SPPC or GPC was associated with HI-EOL outcomes: ICU admission < 30 days from death, mechanical ventilation < 14 days from death, or in hospital death. Sensitivity analyses excluded treatment-related mortality (TRM) cases. Results: 572 patients met inclusion criteria. Children less likely to receive SPPC services included those with hematologic cancers [odds ratio (OR) 0.33, 95th confidence interval (CI) 0.30-0.37; p < 0.001)], in the lowest income quintile (OR 0.44, 95CI 0.23-0.81; p = 0.009), and living at increased distance from the treatment center (OR 0.46, 95CI 0.40-0.52; p < 0.0001). In multivariate analysis, SPPC was associated with a 3-fold decrease in the odds of an EOL ICU admission (OR 0.32, 95CI 0.18-0.57), while GPC had no impact. Similar associations were seen with all other HI-EOL indicators. Excluding TRM had little impact. Conclusions: SPPC, but not GPC, is associated with lower intensity care at EOL. Access to such care however remains uneven. In the absence of randomized trials, these results provide the strongest evidence to date supporting the creation of SPPC teams. These results can be used to support PC advocacy and policy efforts.

Cetuximab plus irinotecan versus panitumumab in patients with refractory metastatic colorectal cancer in Ontario, Canada

The addition of irinotecan to an epidermal growth factor receptor (EGFR) antibody has previously been shown to improve tumor response rate and time to progression but not overall survival (OS) for refractory metastatic colorectal cancer (mCRC). We assessed the "real-world" effectiveness and toxicity of the combination versus monotherapy. In Ontario, Canada, universal public funding is available for either cetuximab plus irinotecan (Cmab + I) combination therapy or panitumumab (Pmab) monotherapy, only in patients with refractory nonmutated RAS mCRC. All patients diagnosed before December 2012 and treated with an EGFR antibody for mCRC were identified from the Ontario drug database and linked to the Ontario Cancer Registry and other administrative databases to ascertain baseline characteristics, health services utilization, and outcomes. Multivariable Cox and logistic models were constructed to compare the time to treatment discontinuation (TTD), OS, emergency department (ED) or hospital visits between Cmab + I and Pmab. Observable confounders were adjusted for using propensity score methods. One thousand and eighty-one patients were identified (Cmab + I: 278, Pmab: 803). Patients receiving Cmab + I were younger (mean age 61 vs 64 years) and had a longer duration of prior irinotecan treatment. The use of Cmab + I as compared to Pmab alone was associated with a prolonged TTD [median: 3.8 months vs 2.8 months] and an improved OS [median: 8.8 months vs. 5.9 months] with an adjusted HR of 0.62 [95% CI 0.53-0.73, p < 0.001]. Both treatment regimens afforded similar 14-day mortality and incidence of ED or hospital visits. The findings for patients over and below the age of 65 were similar.

Sensitivity and specificity of self-reported cancer history compared to cancer registry

234

Background: The Ontario Health Study (OHS) is a large prospective epidemiologic cohort study in which any Ontario resident eighteen years of older may enroll regardless of prior medical history. Baseline data are collected using web-based tools. As part of the consent process, participants are asked for consent to link study data with administrative and health care claims databases, including the Ontario Cancer Registry (OCR). There is an option to enter their Health Insurance Number (HIN) for this purpose. The purpose of this study was to link these data and evaluate the accuracy of self-reported cancer history compared to the cancer registry. Methods: Consenting participants that provided HINs were deterministically linked to the administrative data. Those that did not were probabilistically linked using name, sex and date of birth. Cancer registry records indicating a cancer diagnosed before the date of completion of the OHS baseline survey were considered the gold standard. Concordance, sensitivity, and specificity were assessed. Results: OHS records were successfully linked to administrative claims data and the Ontario Cancer Registry (OCR) with an 85.13% match rate. The final cohort consisted of 139,798 participants. A personal history of cancer was reported by 13,171 of these subjects, out of which 10,066 were found with a record in the OCR. The sensitivity of self-report was 77% and the specificity 95%. Excluding cancers diagnosed after the completion of the baseline survey increased sensitivity of self-report to 93%. The main area of discrepancy causing low sensitivity was the self-reporting of non-melanomatous skin cancers in the OHS questionnaire. Conclusions: While some self-over-reporting of cancer history may occur, cancers with lower metastatic potential tend to be under-captured in our provincial cancer registry. These findings have implications for cohort creation for research and quality improvement.

An evaluation of the breast cancer Well Follow-up Care Initiative using administrative databases: A new model of analysis

1

Background: Cancer Care Ontario (CCO) implemented the Well Follow-up Care Initiative (WFCI) to transition low-risk breast cancer (BC) survivors from oncologist to primary care providers. The objective of this work was to compare both the health system resources utilized and their associated costs, among women in the WFCI (cases) and women who were not transitioned (controls). Methods: Cases were linked to provincial administrative databases and matched to a control group based on year of diagnosis, cancer stage, age, comorbidity, income, geographic area of residence, and prior health system use. Health system resource utilization (physician, hospitalization, diagnostics, medication, and homecare) was ascertained per group. The annual mean and median costs (CAD 2014) per patient were determined. Annualized incremental costs between cases and controls were estimated using generalized estimating equations, accounting for matched pairs. Results: Results are based on 2,324 cases and 2,324 controls (mean age 64.4 and 64.9 years, respectively). During an average of 2.5 years of follow-up since the transition date, there were significant differences between the two groups for mean annual visits per patient with a medical oncologist (0.4 vs. 1.3, p<0.001) and radiation oncologist (0.2 vs. 0.4, p<0.001). There was no significant difference in mean annual family physician visits per patient (7.4 vs. 7.9, p=0.082). The intervention group had fewer inpatient hospitalizations (75.6% vs. 79.9%) and cancer clinic visits (84.9% vs. 94.0%). While there was a higher number of mammograms for cases compared to controls, other diagnostic tests (bone scan, CT, MRI, ultrasound, and x-rays) were done less frequently. The model was associated with a 39.3% reduction in mean annual costs ($6,575 among cases and $10,832 among controls) and a 22.1% reduction in median annual costs ($2,261 among cases and $2,903 among controls). Conclusions: Transitioning BC survivors to primary care was associated with fewer health system resources and had a lower annual mean cost per patient than women who were not transitioned.

Toxicity of high-dose interferon for high-risk melanoma in Ontario: A population-based study of health services use

31

Background: While High-Dose Interferon (HDIFN) is the sole approved adjuvant systemic treatment for melanoma in Ontario and many other jurisdictions, it is toxic, of modest benefit, and costly. We sought to assess the population-level impact of toxicity, particularly neuro-psychiatric toxicity. This can inform value assessment for the adjuvant treatment of melanoma. Methods: This was a retrospective population-based registry study of all patients with melanoma receiving adjuvant HDIFN in Ontario 2008-2012. HDIFN receipt was determined from provincial drug-funding data. Toxicity was investigated through health services use compatible with HDIFN toxicity (e.g. mental health physician billings). Associations between early HDIFN discontinuation and health services use were examined. Using stage data reported from cancer centers on a subset of patients, propensity matched analysis compared utilization in stage IIB-IIIC patients that did and did not receive HDIFN. Results: Of 718 patients receiving HDIFN, 12% were ≥65 years, 83% had little or no comorbidity. One third had ≥1 toxicity-associated utilization within one year of starting HDIFN. 364/420 (87%) of utilization was mental health-related: 54% were family practitioner visits, 39% psychiatrist visits. Early drug discontinuation was more likely with pre-existing mental health issues in multivariable analysis (OR 2.0 (1.1,3.4)). In propensity matched analysis, HDIFN patients were more likely than untreated matched controls to have mental health utilization (51% vs. 42%, p=0.01) between 1 year pre-melanoma diagnosis to 2 years post. Conclusions: Mental health services use is common among stage IIB-IIIC patients with melanoma, especially with HDIFN. This emphasizes an important survivorship issue for these patients, and for those receiving HDIFN, and impacts the value of care. Pre-treatment mental health services use is associated with treatment discontinuation. This is important when contemplating the value of HDIFN use for individual patients. For those receiving HDIFN, optimal support must include mental health care.

Driving quality improvement with public reporting: Use of imaging tests outside guidelines for early-stage breast cancer in Ontario

193

Background: Most patients diagnosed with breast cancer will have early stage (stage I or II) disease, with low chance of distant metastases. Thus most guidelines, including Choosing Wisely, recommend against imaging tests for distant metastases in asymptomatic early stage breast cancer. Despite this, most (86%) of these patients in Ontario received these tests from which they are not likely to benefit and may result in investigations that can be invasive and delay treatment. Publicly reported indicators, such as those in Ontario’s Cancer System Quality Index (CSQI), can bring research findings to action by identifying areas for improvement and facilitating ongoing assessment. In practice, this can be challenging due to limitations in administrative data. Moreover, relatively few quality improvement indicators focus on efficiency, the dimension of quality looking at best use of resources to achieve desired outcomes. We sought to examine trends in the use of imaging tests in early stage breast cancer and to drive quality improvement efforts via public reporting. Methods: Data from the Ontario Cancer Registry, the Discharge Abstract Database and the Ontario Health Insurance database was used to identify how many Ontario breast cancer patients diagnosed with early stage breast cancer received staging tests from 2012–2014. Imaging tests included were ultrasound, CT scan, MRI, x-ray and bone scan. The results were subsequently shared with the Regional Cancer Centres and publically released in the CSQI. Results: From 2012 to 2014, 75.1, 72.7 and 71.3% respectively of early stage breast cancer patients received at least one imaging test for staging. This is much higher than the 5-10% of patients expected to need tests due to symptoms or comorbidities. While the regional variation ranged from 47-80%, the rates were high across the province with no clear pattern. Conclusions: Public reporting may be having some effect on overtesting, but rates remain high. Following outreach by the Cancer Quality Council of Ontario and Cancer Care Ontario, targeted regional interventions are being developed and implemented, the impact of which will be assessed and reported in future releases of the CSQI.

Are we choosing wisely? Operationalizing Choosing Wisely in the Ontario cancer system

221

Background: The Choosing Wisely (CW) campaign aims to initiate conversations about unnecessary treatments contributing to the rising cost of cancer care. We aim to develop a data linkage research platform to operationalize CW recommendations within administrative health care databases and the population-based performance of these recommendations. We initiated testing with the CW recommendiation against routine surveillance imaging in patients with aggressive histology lymphoma and pancreatic/gastric (P/G) cancer treated with curative intent. Methods: We used population-based administrative databases from Ontario, Canada to examine a cohort of adult patients with diffuse large B-cell lymphoma (DLBCL) (2004-2011) and P/G cancer post-surgical resection (2003-2013). For the DLBCL cohort, we defined an index date of 2-years after the last dose of R-CHOP as the time-frame beyond which surveillance CT imaging would be inappropriate. For the P/G cohort, the index date was 6 months post-resection. The primary outcome was cumulative incidence of CT scans within 3 years of the index date. To ensure that only surveillance scans were captured, we censored 6 months prior to development of recurrent disease, a new cancer diagnosis, or death. Results: The cohort consisted of 2,838 DLBCL and 2,930 P/G patients. The cumulative incidence of receiving CT imaging in the three years post index date was 55.6% (95% CI 53.7%-57.5%) among DLBCL and 82.8% (95% CI 81.3%-84.3%) among P/G patients. DLBCL patients ≥65 were more likely to receive imaging (p<0.01) as were those with more comorbidities (p<0.01). Younger patients with P/G were more likely to receive imaging (p<0.01) as were men (p<0.01). Income and rurality did not predict for increased imaging in either cohort. Surveillance CT imaging decreased over time among DLBCL patients (p<0.01), but increased among P/G cancer patients (p<0.001). Conclusions: During a time-frame in which surveillance imaging is deemed unnecessary by the CW campaign, the practice in Ontario remains excessive. This study represents a real-world demonstration that CW statements can be operationalized within population-based administrative databases and used as quality indicators in cancer care.

Outcomes of FOLFIRINOX (FFX) and gemcitabine+nab-paclitaxel (GnP) in initially unresectable locally advanced pancreatic cancer (uLAPC): A population-based study

394

Background: Data regarding the benefits of FFX and GnP in patients (pts) with initially uLAPC is limited. FFX and GnP have been universally publicly funded for first-line uLAPC in Ontario, Canada, since April 2015. The aims of this study are to determine (1) the overall survival (OS) of pts receiving FFX and GnP, (2) the surgical conversion rate of FFX and GnP, and (3) whether resection is associated with better OS in pts with uLAPC in an unselected, real world population. Methods: All pts in Ontario who started first-line FFX, GnP or gemcitabine (G) for uLAPC between April 2015 and March 2016 were identified in Cancer Care Ontario’s New Drug Funding Program database. They were linked to the Ontario Cancer Registry and other population-based databases to ascertain baseline characteristics (age, sex, performance status (PS), locating of tumor, income quintile, and rural residence) and outcomes (pancreatic cancer resection and vital status). Crude and adjusted models of OS were generated using Kaplan-Meier the method and Cox regression. Surgical resection was modelled as a time-dependent variable to examine its association with OS. Results: We identified 147 pts with uLAPC (mean age = 65, 44% female, 31% ECOG PS 0, 61% PS 1, 60% pancreatic head). Ninety (61.2%), 40 (27.2%) and 17 (11.6%) patients were treated with FFX, GnP and G, respectively. With a median follow-up of 7.5 months, median OS was not reached. The 6-month OS rate was 87.8%, 75.1% and 76.4% for FFX, GnP and G, respectively (p = 0.33). Resection occurred in 12 (8.2%) patients, with 10 (11.1%) and 2 (5.0%) treated with FFX and GnP, respectively ( p= 0.34). Surgical resection after initial chemotherapy was not associated with better OS in multivariable analysis (HR 0.26, 95%CI 0.03-1.98, p= 0.19). Conclusions: Pts with uLAPC treated with FFX and GnP appeared to have a reasonable OS in the real world, with > 75% of pts alive at 6 months. Surgical conversion rate in this unselected population appeared to be less than other single institutional studies. The current findings do not appear to show an early surgical benefit, but longer follow-up will be required to assess the potential long-term benefit of surgery.

The Association of Dyslipidemia With Chronic Lymphocytic Leukemia: A Population-Based Study

Background

Metabolic syndrome (MetS) is a risk factor for development of cancer. Because aberrant lipid metabolism is a pathogenic feature of chronic lymphocytic leukemia (CLL), our objective was to determine if CLL patients have a higher prevalence of MetS preceding diagnosis and to determine the impact of lipid-lowering medications on survival.

Methods

We conducted a population-based case-control study in Ontario, Canada, using administrative databases of adults age 66 years and older to compare the prevalence of MetS preceding CLL with age- and sex-matched control subjects. Logistic regression was used to study the association between MetS and its components to CLL. The Kaplan-Meier method and Cox Regression were used to investigate survival. All statistical tests were two-sided.

Results

We identified 2124 persons with CLL and 7935 control subjects from January 1, 2000, to December 31, 2005, with follow-up until March 31, 2014, three years from the date of last contact with the health care system, or death. The mean age was 75.6 years, 20.2% had diabetes, 35.8% had hypertension, and 17.6% had dyslipidemia. In multivariable analysis, dyslipidemia (odds ratio [OR] = 1.26, 95% confidence interval [CI] = 1.11 to 1.44, P < .001) and hypertension (OR = 1.12, 95% CI = 1.01 to 1.25, P = .03) were associated with the development of CLL, whereas MetS and diabetes were not. Lipid-lowering medication was associated with a statistically significant improved survival in patients with CLL (HR = 0.53, 95% CI = 0.47 to 0.61, P < .001).

Conclusions

We demonstrate a higher prevalence of dyslipidemia preceding a diagnosis of CLL compared with control subjects, supporting preclinical data. Lipid-lowering medications appear to confer a survival advantage in CLL. Prospective studies are needed to confirm these results and test their potential as therapeutic applications.