Focus on Quality
Baseline Estimates of Adherence to American Society of Clinical Oncology/American Board of Internal Medicine Choosing Wisely Initiative Among Patients With Cancer Enrolled With a Large Regional Commercial Health Insurer
See accompanying article on page
A structured approach to evaluating adherence and cost impact is needed before developing programs that are aimed at improving adherence to the American Society of Clinical Oncology Choosing Wisely measures.
The American Society of Clinical Oncology (ASCO)/American Board of Internal Medicine (ABIM) Choosing Wisely (CW) measures aim to reduce the use of interventions that lack evidence of benefit in cancer care. The study presented here characterized adherence to the 2012 ASCO/ABIM CW recommendations by linking health plan claims data with a regional cancer registry and sought to identify areas for research interventions to improve adherence.
SEER records for patients diagnosed with cancer in Western Washington State between 2007 and 2014 were linked with enrollment and claims from a large regional commercial insurance plan. Using claims and SEER records, algorithms were developed to characterize adherence to each CW measure. In addition, we calculated differences in total reimbursements and procedure-specific reimbursements for patients receiving adherent and nonadherent care.
A total of 22,359 unique individuals with cancer were linked with insurance enrollment records and met basic eligibility criteria. Overall adherence varied from 53% (breast surveillance) to 78% (breast staging). Within each measure, adherence varied substantially by stage at diagnosis and by cancer site in situations in which the CW measure affected multiple types of cancer. The difference in reimbursements between adherent and nonadherent populations across all five measures was approximately $29 million.
In April of 2012, the American Society of Clinical Oncology (ASCO) and the American Board of Internal Medicine (ABIM) Foundation, as part of the ABIM Choosing Wisely (CW) campaign, released the initial Top Five list of tests and procedures in oncology for which use should be questioned because of their failure to add clinical value (Data Supplement).1
The CW list was designed to identify practices that are costly, widely used, and for which no evidence exists to support value, and to promote conversations between physicians and patients about using the most appropriate tests and treatments as well as about avoiding care that is unnecessary or for which harm may outweigh the benefits.
Although the CW list was selected after input from more than 200 oncologists, there was no empiric validation of either the prevalence of the care processes that were included, their costs to the health care system, or the accuracy of measurement of these processes in oncology practice. Because these are important issues for health care delivery systems, we used cancer registry and health insurance claims data to test the importance of the practices that were included on the CW list, to retrospectively review oncologists' adherence to these practices, and to test the feasibility of using administrative data to measure adherence. These are issues of relevance to health care delivery systems and health insurers, given that implementation of the CW recommendations will require substantial investments on many levels.
Accordingly, the primary purpose of this study was to estimate adherence to the ASCO/ABIM recommendations in persons with cancer who are enrolled in a large regional commercial insurance plan. To further evaluate the relative level of cost savings that might be achieved through improving adherence to the measures, we also estimated total health care costs for persons whose care was adherent to CW recommendations versus costs for those with similar characteristics who had nonadherent care. Our findings may be helpful to health care organizations that are considering investment in measures and processes that are designed to improve adherence to the CW recommendations for oncology.
The study was conducted by Fred Hutchinson Cancer Research Center investigators in conjunction with leaders at Premera Blue Cross, a not-for-profit commercial Blue Cross/Blue Shield association insurer with approximately 1.2 million enrollees in Washington State, approximately 20% of all state residents younger than 65 years of age. To identify patients undergoing care to which the CW recommendations might apply, we linked Premera health plan enrollment files to cancer registry records from the Western Washington Cancer Surveillance System (CSS). As a member of the National Cancer Institute's SEER registry, CSS collects comprehensive information on staging, initial treatment, and survival for persons diagnosed with malignancies in Western Washington, excluding nonmelanoma skin cancer.2 Our linked database included persons age 18 years and older who were diagnosed with cancer between January 1, 2007, and May 31, 2014. To be included in the study, individuals needed a known date of diagnosis and could not have been diagnosed at autopsy or via death certificate. For CW measures 2 to 5, we included patients with newly diagnosed cancers and no diagnosis of other malignancy.
After linkage of CSS and insurance enrollment records, claims for inpatient and outpatient services and outpatient pharmacy were extracted for eligible individuals over the same time period. Funding for the project was provided by an unrestricted grant from Premera Blue Cross and the Fred Hutchinson Cancer Research Center. Institutional review board approval for this study was granted by the Fred Hutchinson Cancer Research Center on September 18, 2013.
CSS registry and insurance claims records were used to determine which persons were potentially eligible for evaluation with respect to each CW measure. To identify eligible patients, we matched enrollment files from approximately 1.2 million members from January 2007 through May 2014 with cancer registry records from the same time period. The Data Supplement details the criteria that were used to define the populations and determine adherence to each measure. Here we discuss specific approaches we took with respect to each measure. The CW measures are paraphrased in italics.
1. Avoid unnecessary anticancer therapy in patients with advanced solid tumors who are unlikely to benefit, and focus instead on symptom relief and palliative care.
Our population of interest for this measure was patients with solid tumors diagnosed at any stage of their malignancy who died during the observation period. Claims for one or more chemotherapy and radiation therapy treatments during a period of 90 days before death (as a result of any cause) were examined.
2. Do not use PET [positron emission tomography], CT [computed tomography], and radionuclide bone scans in the staging of early prostate cancer at low risk of spreading.
The eligible population for this measure included men with local-stage, low-risk prostate cancer. Using SEER records, we defined patients with local-stage, low-risk disease as those with less than T1c/T2a or T2 not-otherwise-specified prostate cancer with Gleason scores ≤ 6 or prostate-specific antigen scores ≤ 10 at diagnosis. Because documentation was not standardized, men with T2 not-otherwise-specified disease or men who did not have records for prostate-specific antigen levels or Gleason scores were categorized into an early-stage unknown risk group.
3. Do not use PET, CT, and radionuclide bone scans in the staging of early breast cancer that is at low risk of spreading.
The eligible population for this measure was women with American Joint Committee on Cancer stage 0, I, or II breast cancer. We substratified by stage to examine whether there were trends in ordering of these imaging procedures across a spectrum of stages at presentation.
4. For individuals without symptoms who have completed curative breast cancer treatment, routine blood tests for certain biomarkers (eg, CEA [carcinoembryonic antigen], CA [cancer antigen] 15-3, CA 27-29) and advanced imaging tests (PET, CT, and radionuclide bone scans) should not be used to screen for cancer recurrence.
The target population for this measure was patients with breast cancer who had a high likelihood of receiving breast cancer treatment with curative intent. Accordingly, we focused on women with American Joint Committee on Cancer stages I and II disease. Curative therapy was defined as a mastectomy within 180 days of diagnosis, or a lumpectomy within 180 days of diagnosis and start of radiation therapy within 90 days of lumpectomy. Testing during the surveillance period was measured from the beginning of the first 4-month gap in treatment (mastectomy, lumpectomy, radiation therapy, and chemotherapy) through 13 months later or at the start of new treatment (whichever was earlier). Patients diagnosed with in situ disease at diagnosis were excluded.
5. Avoid administering white-cell stimulating factors to patients undergoing chemotherapy who have less than a 20% risk for febrile neutropenia.
We focused on patients who received low– or intermediate–febrile neutropenia (FN) –risk chemotherapy for breast, non–small-cell lung, or colorectal cancer. We restricted our analysis to only the first cycle of chemotherapy for patients with no previous malignancy. When searching for chemotherapy, we classified risk on the basis of the combination of single agents received in the first 10 days of chemotherapy. The combinations of agents were assigned an FN risk on the basis of National Comprehensive Cancer Network guidelines, and assigned to the higher FN risk category if the combination of agents could represent multiple chemotherapy regimens.3 For adherence to the recommendation, we measured any use of filgrastim or pegfilgrastim within 21 days from the start of (low/intermediate risk) chemotherapy.
Cost of care was measured as total reimbursements for all claims filed during the relevant period of interest after stratifying patients with cancer into adherent and nonadherent groups. In addition to total costs, we also tallied reimbursements for the specific procedures associated with each measure (eg, reimbursement for PET scans during prostate cancer staging, CW recommendation 2).
The linkage of SEER and insurance enrollment files identified 22,444 unique individuals with cancer, 85 of whom were later excluded because of a lack of diagnosis date, diagnosis at autopsy, or diagnosis via death certificate. Table 1 shows the demographic characteristics of this commercially insured population by CW recommendation. The oldest average age of eligible persons was noted for palliative care and the youngest average age of eligible persons for the use of imaging during breast cancer staging. With the exception of the palliative care group, the majority of persons eligible for each measure had no noncancer comorbidities.
|Characteristic||Palliative Care (CW R1)||Prostate Staging (CW R2)||Breast Staging (CW R3)||Breast Surveillance (CW R4)||Colony-Stimulating Factors (CW R5)|
|No. of patients||1,565||518||1,798||629||672|
|Mean (SD) age, years||69 (14.9)||63 (9.1)||54 (12.1)||56 (12.5)||57 (12.9)|
|Non-Hispanic white, %||92||81||85||88||89|
|Stage at diagnosis, %|
|Diagnosis year, %|
|Noncancer comorbidities, %*|
|One or more||33||12||8||9||13|
|Adherence to CW measures, %|
|Stage at diagnosis|
Abbreviations: ABIM, American Board of Internal Medicine; ASCO, American Society of Clinical Oncology; CW, Choosing Wisely; R, recommendation; SD, standard deviation.
*Comorbidities in the 12 months before diagnosis. Unknown comorbidity scores are the result of incomplete enrollment during that time period.
Overall adherence to CW measures varied from 53% (breast surveillance) to 78% (breast staging). However, when broken down by stage at diagnosis, variation within and across measures was much greater (Table 1). In general, adherence was poorest for persons diagnosed at more advanced stages of disease.
Adherence to the CW palliative care recommendation ranged from 60% at 90 days from the date of death to 89% at 14 days from the date of death (Figure 1A). Chemotherapy was given much more commonly than radiation therapy until 15 days before death, at which point radiation therapy was used slightly more commonly than chemotherapy.
Adherence to the early prostate cancer staging recommendation was relatively high (90%) for patients who could be identified as having low-risk disease (Figure 1B). Adherence was somewhat lower among patients with local-stage disease for whom risk status could not be estimated (70%). Adherence to the CW breast cancer staging recommendation was variable, with lower rates of adherence for persons with more advanced disease at presentation (Figure 1C). Overall adherence to staging was 78%. CT scans were the most commonly ordered nonrecommended test (282 of 1,798 patients).
Adherence to CW recommendations for breast cancer surveillance was relatively poor, with 53% of patients receiving at least one of the procedures during the follow-up period (Figure 1D). Adherence was somewhat worse for women treated with mastectomy compared with those treated with lumpectomy and radiation. Breast cancer tumor markers were the most commonly ordered test during this time period (39% of all patients).
Use of colony-stimulating factors among patients receiving low- or intermediate-risk chemotherapy occurred in 30% of patients receiving chemotherapy. Adherence in breast and non–small-cell lung cancer was poorest, with 50% and 26% of patients receiving low- or intermediate-risk chemotherapy also receiving CSF during the observation period (Figure 1E).
The differences in total care costs between adherent and nonadherent populations varied widely between measures (Table 2). Stated as percentages, the greatest relative difference in total costs between patients whose care was adherent versus nonadherent was for the palliative care group (184%), whereas the least relative difference was for the CSF group (31%). Considering reimbursements for CW-related procedures only, the entire difference between nonadherent and adherent groups was attributed to the cost of CSFs, addressed by the fifth recommendation (99%). The lowest difference was for breast cancer surveillance, for which only 7% of the cost difference was a result of the increased use of imaging and tumor markers, and the remaining 93% was from other services. The greatest absolute difference in total reimbursements and procedure-specific reimbursements was for the CW palliative care measure ($32,406 and $12,805, respectively). The difference in total reimbursements between adherent and nonadherent populations across all five measures was approximately $29 million. Considering only the cost of procedures that were directly related to the CW recommendations, the difference in total reimbursements was approximately $10 million.
|Variable||Palliative Care (CW R1)||Prostate Staging (CW R2)||Breast Staging (CW R3)||Breast Surveillance (CW R4)||Colony-Stimulating Factors (CW R5)|
|Time period||≤ 3 months from death||± 2 months from diagnosis||± 2 months from diagnosis||Surveillance period†||0-21 days from start of chemotherapy|
|Total costs during time period: adherent patients, $||17,606||5,944||20,832||14,710||15,395|
|Total costs during time period: nonadherent patients, $||50,012||8,423||33,630||23,751||20,177|
|Costs associated with nonadherent procedures, $‡||12,805||818||2,463||641||4,718|
|Percent of change directly attributed to nonadherent behavior||40||33||19||7||99|
Abbreviations: CW, Choosing Wisely; R, recommendation.
*Claims paid during the period of interest.
†Surveillance period starts at the beginning of the first 4-month gap in treatment (mastectomy, lumpectomy, radiation, chemotherapy) and ends 13 months later or at the start of a new treatment (whichever is earlier).
‡The Data Supplement contains the nonadherent procedure codes for each recommendation.
Reducing the use of unnecessary procedures is imperative in oncology, both to protect patients from harm and to reduce the financial burden on patients and the health care system. Accordingly, linking enrollment and claims records from a large commercial insurer, we sought to estimate adherence to the 2012 ASCO/ABIM CW measures for oncology. We found that adherence was highly variable across measures, with some having relatively high rates of adherence but others demonstrating quite low adherence. Moreover, because each recommendation is multidimensional with regard to clinical care and with adherence varying across those dimensions, improving adherence is likely to be complex and potentially resource intensive.
We also examined the differences in total direct medical reimbursements and procedure-specific reimbursements between patients whose care was adherent versus nonadherent to CW recommendations. We found that total incremental costs per patient for nonadherent care can differ more than 10-fold between CW measures. At the insurance plan–population level, total incremental costs for nonadherence were most pronounced for palliative care (approximately $21 million) and breast cancer staging (approximately $5 million). Procedure-related costs for nonadherence were much lower than total difference between adherent and nonadherent groups, but likely underestimate the true cost, given that CW-related procedures are likely to create additional services in a substantial fraction of patients (eg, cost of evaluating positive results for breast surveillance procedures). The true cost of nonadherence is likely between the upper bound of total cost difference and the lower bound of procedure-specific cost difference.
We believe that the evaluation of differences in overall adherence and cost is a critical first step toward making plan-level decisions regarding the ASCO/ABIM CW recommendations. Implementing any program that is aimed at improving adherence will be complex and costly, likely taking years of sustained effort. Given resource constraints, plans and practices will need to make difficult prioritization decisions among the many CW measures that have now been published. The number of patients affected, the potential clinical impact, and the relative cost of nonadherence are all factors that need to be considered when prioritizing among the CW measures. Recognizing that patient preferences and unique clinical factors mean that 100% adherence to each CW measure is not desirable, our analysis suggests ample room for improving overall adherence to the CW metrics.
There are several limitations to our analysis. First, our sample was composed of patients with cancer who are insured by a large commercial insurer, and thus was not necessarily representative of all patients with cancer in the region. Second, because we did not have access to performance status information, it is possible that some radiation therapy and chemotherapy for end-of-life care (palliative care in the CW recommendation) could have been appropriately provided to prolong life, or in other patients, for symptom relief. In addition, the diversity of this group (eg, with many cancers at various points from diagnosis) may in part explain the observed variability in adherence and cost. Third, for the CW breast cancer staging recommendation, we could not reliably identify patients with local- and regional-stage disease with genomic variants that influence risk (eg, BRCA1/2 mutation carriers), because registry and claims do not reliably record this information. Fourth, for the CW recommendation regarding use of biomarkers and imaging for surveillance after treatment, our surveillance analysis may have included some women who experienced a recurrence or progression to advanced stage disease. In addition, because it was not possible to identify the actual intent of testing, some procedures may have been undertaken to evaluate symptoms. Nevertheless, our approach provided an upper bound on use of these services among patients. In fairness to participating clinics, one would want to allow some period of time after publication of the CW recommendations to start the adherence clock. Unfortunately, we do not have sufficient study numbers for 2013 to 2014 to provide a valid assessment of a pre/post trend. Finally, the costs reflect the burden to the health plan and do not represent costs borne by patients or the total societal burden of care.
In summary, we find widely varying adherence to the 2012 ASCO/ABIM CW measures for oncology among patients covered by a large regional commercial health insurer. In addition to the potential clinical impact, we find that the potential cost implications are substantial for measures with poor adherence, large numbers of affected individuals, and large differences in procedure-specific and associated costs between patients with adherent and nonadherent care. Future research is necessary to understand the best approaches for prioritizing CW measures for subsequent intervention and for designing effective measures aimed at improving adherence.
Supported by a contract from Premera Blue Cross and the Fred Hutchinson Cancer Research Center.
Disclosures provided by the authors are available with this article at jop.ascopubs.org.
Conception and design: Scott D. Ramsey, Catherine Fedorenko, Rakesh Chauhan, Richard McGee, Gary H. Lyman
Collection and assembly of data: Scott D. Ramsey, Catherine Fedorenko, Rakesh Chauhan, Karma Kreizenbeck
Data analysis and interpretation: Scott D. Ramsey, Catherine Fedorenko, Richard McGee, Gary H. Lyman, Aasthaa Bansal
Manuscript writing: All authors
Final approval of manuscript: All authors
|1.||LE Schnipper, TJ Smith, D Raghavan , etal: American Society of Clinical Oncology identifies five key opportunities to improve care and reduce costs: The Top Five list for oncology J Clin Oncol 30: 1715– 1724,2012 Link, Google Scholar|
|2.||About the Cancer Surveillance System Fred Hutchinson Cancer Research Center http://www.fhcrc.org/en/labs/phs/projects/cancer-surveillance-system/about.html Google Scholar|
|3.||NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines): Myeloid growth factors (version 2.2014) National Comprehensive Cancer Network http://williams.medicine.wisc.edu/myeloid_growth.pdf Google Scholar|
The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or jop.ascopubs.org/site/misc/ifc.xhtml.
Consulting or Advisory Role: Genentech, Seattle Genetics
No relationship to disclose
No relationship to disclose
Leadership: First Choice Health Network
Stock or Other Ownership: Covidien, Express Scripts, Mallinckrodt, Amgen, Celgene, Myriad Genetics, Procter & Gamble, ResMed, First Choice Health Network
Consulting or Advisory Role: Palmetto, Gerson-Lehman Brothers Medical Consulting
Research Funding: Amgen (Inst)
No relationship to disclose
No relationship to disclose